Opening this thread to discuse and replicate an NMR experiment proposed by Smudge.
It was presented 2 times already first on OUR.com, then on OU.com but quickly faded into nothingness.
First thread (2016) was here: https://www.overunityresearch.com/index.php?topic=3322.msg56207#msg56207
Second thread (2020) here: https://overunity.com/18502/nuclear-magnetic-resonance-nmr/msg546019/#msg546019
Some key PDF's presented in those threads are attached below, the important build one on top.
Also in those earlier threads good info is to be found in the form of pictures and links.
Conradelectro was so generous to 3D-print the tubular thingy to hold the water and send it to me.
I am in the process of collecting / building the rest of the setup to start experimenting.
As soon as i have something presentable i will show it here.
Be welcom to put in your comments, suggestions and replication attempts.
Regards Itsu.
First question i had was about the below pancake coil, see red circle.
It shows 11 turns of 1mm wire, which means it will be 11mm wide.
So the shown 30mm won't cut it, but Smudge confirmed it can be 30 turns, so filling up the 30mm.
Next question i have is on the 100 turns of 1mm on the toroidal coil, which will leave 1.76mm between turns.
88mm diameter = 276mm circumference / 100 = 2.76mm.
Is this interspace needed (capacitance?) or can it be closely wound (like 276 turns)?
Itsu
Hi Itsu,
The field in the water must be as uniform as possible, so the turns need to touch each other, no gaps. In my paper the conceived design was given more for showing the calculations, not the full details. Thanks for sorting this out and for trying the experiment. Thanks also to Conrad for printing the special tube.
Smudge
Thanks Smudge, your welcom.
How about the pancake coils connection, in series OK, but bucking or aiding?
This means i need to make the both pancake coils similar (bucking) or mirror like (aiding).
Thanks, Itsu
Hi Itsu,
To get the wanted radial field between the pancake coils they have to be connected in series opposing. That is the driving field. The toroidal coil is the output coil. I was wrong in my previous post where I talked about getting the field from the toroidal coil as uniform as possible, that was perhaps misleading. It is the static field from the magnet that has to be as uniform as possible throughout the water volume. That requires the geometry to be carefully made as symmetrical as possible. Unfortunately you won't have control of the magnetization in the magnets, so it may be necessary to create some form of adjustment to minimise any assymmetry of that magnetization. The static field in the water is the one that the protons precess about, and their precession frequency is determined by the field magnitude. Ideally we want uniform static field so that all the protons precess at precisely the same frequency, that would give infinite Q. It would require zero energy to get all the protons to precess in phase. But that can't happen, there will be a spread of precession frequencies hence they will never under normal circumstances be in phase, the spread of random phases results in a zero RF output. The smaller the spread of frequencies the easier it is for the RF drive to create some phase coherence. Then the precessing protons act like an array of tiny precessing bar magnets, driven by some inner quantum forces to be persistently precessing. Extracting power via the toroidal coil will try to slow those precessions down, so the field from load current in that coil does need to be uniform so as not to create varying slowing down rates that will inevitably reduce the phase coherence.
Smudge
Thanks Smudge,
so opposing (bucking) series it is for the pancake coils.
I have the both ring magnets, they are huge.
Still waiting for the special tube from Conrad, so i can concentrate on building the pancake coils for now.
Itsu
I am glad that Itsu will try the experiment. His electronics expertise is much higher than mine.
I sent the circular tube last Friday (12th June), but the Austrian postal services have a backlog of 100.000 packages because of coronavirus outbreaks in two distribution centres near Vienna. So, it might take an other week till Itsu gets the tube. Which he then has to cover with an epoxy glue, because the tube is not completely watertight.
If Itsu needs more or different tubes I am ready to print and send them.
Greetings, Conrad
Thanks Conrad.
I received the tube today, see picture 1
Up till now i have the tube, the molds for the pancake coils and the ring magnets, see picture 2
Will be needed to seal the tube first, and start winding the pancake coils etc.
Itsu
Quote from: Itsu on 2020.06.20, 09:54:46
I received the tube today, see picture 1
Up till now i have the tube, the molds for the pancake coils and the ring magnets, see picture 2
Will be needed to seal the tube first, and start winding the pancake coils etc.
Within the EU, I am willing to send a tube to experimenters. Outside the EU I will send the .stl file by E-Mail and one can have it 3D-printed locally (which is cheaper than sending by mail, there are enterprises everywhere who do the 3D-printing for a good price or you find someone who has a 3D-printer, private individuals, clubs or schools).
I also design tubes with other dimensions if needed.
Once Itsu has done first tests the requirements (specially for the electronics) will be clearer.
Amplification of the ~4 MHz drive sinus signal (from a function generator) for the input pancake coils and preamplification of the output signal (from the toroidal coil around the tube) for the scope are the issues to be solved. And Itsu has this covered.
I like the big ring magnets, are they strong enough to hurt fingers if handled carelessly?Greetings, Conrad
Quote from: Smudge on 2020.06.18, 15:57:33
It seems all NMR experiments looking for OU use ferromagnetic material, presumably on the basis that such material will yield the largest magnetic output.
This inevitably creates problems with uniformity of the static magnetic bias because of the shape dependent depolarising field.
No, these experiments target the 33 Tesla internal hyperfine fields of the ferromagnetic domains. They are very different from the water-tube experiment, which Itsu is performing in this thread.
The magnitudes of these huge internal hyperfine fields are inherently uniform because any external fields are so small compared to them (...and anti-parallel)
See the attached paper.
Quote from: Smudge on 2020.06.19, 15:05:58
It is the static field from the magnet that has to be as uniform as possible throughout the water volume. That requires the geometry to be carefully made as symmetrical as possible. Unfortunately you won't have control of the magnetization in the magnets, so it may be necessary to create some form of adjustment to minimize any asymmetry of that magnetization.
Yes, this uniformity is of paramount importance, so some kind of mechanical screw adjustment of the magnets' planes should be implemented in two perpendicular axes.
Quote from: Itsu on 2020.06.19, 15:39:54
Still waiting for the special tube from Conrad, so i can concentrate on building the pancake coils for now.
The pancake coils are the drive coils. I am very interested what inductance they will have because this inductance presents an inductive reactance to the driving signal ...and limits the AC current in the these coils.
In other words, the smaller the inductance of these pancake coils - the easier it is to push AC current into them.
The input LC tank will help a lot with maintaining the amplitude of this driving current BUT this LC circuit is frequency selective !
This selectivity complicates the entire endeavour because it takes away the freedom of an easy adjustment of the driving frequency by the means of merely turning a knob on Itsu's signal generator. (He will need to: Turn the knob & Adjust the LC tank, Turn the knob & Adjust the LC tank, Turn the knob & Adjust the LC tank...)
And this driving frequency will need to be adjusted frequently in order to match the Larmor frequency, which changes as the magnets are adjusted for the best uniformity of the magnetic field density - which is a crucial factor in any NMR.
Also, while Smudge is correct in theory that "
we can easily decouple the output from the input by the use of coils at right angles", the parallel orientation of the drive coils to the magnets, the I/O capacitive coupling and the sheer proximity of the pancake drive coils to the toroidal output coil will make it impossible in practice (they will couple even acoustically!).
Also, the nulling out of this crosstalk by "
either by careful layout taking account of the "hot" end of the drive or by adding extra trimmer" is practically impossible according to my experiences.
So in my opinion, the only way Itsu will be able to observe the NMR signal with this setup will be to
turn off the driving signal and listen for the decaying Larmor precession (FID).
In other words:
TDM (https://en.wikipedia.org/wiki/Time-division_multiplexing).
@Itsu
Notice that the flux generated by the pancake coils becomes perpendicular to the PM's flux (green vertical arrow) somewhere outside of the OD of these coils (inside gray circles), so you should make these pancake coils smaller than the mean diameter of your sample (water) holder ring.
A quick experiment with a miniature compass will show you the angle of that field when DC is flowing through the pancake coils (in absence of the magnets).
thanks verpies,
all valid points i am sure and a good time to mention them as now i still have the time and possibility to implement any hardware / electronic changes.
For the tuning a PLL system comes in mind.
Also the mentioned TDM (Time-division multiplexing) still is on the back of my mind in case needed.
Concerning the pancake coil diamater, "the mean diameter of your sample (water)" is 100mm, while the pancake coils cover 70 to 130mm diameter, so do you mean the pancakes should be like 40 to 100mm or so?
Lets see how easy(or not) it is to produce the 70 to 130mm pancakes, then i can always deviate from it.
Itsu
Quote from: Itsu on 2020.06.20, 15:50:58
For the tuning a PLL system comes in mind.
Our PLL system adjusts the driving frequency to automatically match the resonance frequency of an LC tank.
This NMR system with the LC circuit at the input, needs to change BOTH frequencies !
Quote from: Itsu on 2020.06.20, 15:50:58
Concerning the pancake coil diamater, "the mean diameter of your sample (water)" is 100mm, while the pancake coils cover 70 to 130mm diameter, so do you mean the pancakes should be like 40 to 100mm or so?
Yes. just power them up with DC and stick a small compass in the plane between them, to see their field angles.
Also, pancake coils that "
cover 70 to 130mm diameter" must have many turns...and consequently a large inductance (even when connected in series).
Consider how much RF current you will be able to push into coils of such large inductive reactance ?
...and
how small the capacitor needs to be to resonate with the pancake coils at the Larmor frequency (and will the intrawinding capacitance be larger than that ) ?
QuoteOur PLL system adjusts the driving frequency to automatically match the resonance frequency of an LC tank.
This NMR system with the LC circuit at the input, needs to change BOTH frequencies !
Yes, during my diner just now that dawned on me.
QuoteYes. just power them up with DC and stick a small compass in the plane between them, to see their field angles.
Also, pancake coils that "cover 70 to 130mm diameter" must have many turns...and consequently a large inductance (even when connected in series).
Consider how much RF current you will be able to push into coils of such large inductive reactance ?
...and how small the capacitor needs to be to resonate with the pancake coils at the Larmor frequency (and will the intrawinding capacitance be larger than that ) ?
In my 2th post above you see the initial pancake coil turns were 11, but it showed a 30mm area, so we now changed that to 30 turns.
I will make one and do the compass test, measure the inductance and we go from what we find.
Itsu
Quote from: verpies on 2020.06.20, 15:15:40
@Itsu
Notice that the flux generated by the pancake coils becomes perpendicular to the PM's flux (green vertical arrow) somewhere outside of the OD of these coils (inside gray circles), so you should make these pancake coils smaller than the mean diameter of your sample (water) holder ring.
A quick experiment with a miniature compass will show you the angle of that field when DC is flowing through the pancake coils (in absence of the magnets).
Verpies,
Sorry, but that image is misleading as it doesn't show pancake coils. For the NMR it is the flux flowing over the surface.of the pancake which is radial to the axis of the water ring. It is that flux which is then perpendicular to the magnet flux in the water.. What you show are two non-pancake current rings which of course require a special geometry to get their flux to meet this requirement of perpendicular fluxes in the water. I agree with your concerns for getting the needed RF
current in the pancake coils.
Smudge
Here is an FEMM result to show the flux between the pancake coils. The outer blue circle is simply the boundary for the FEMM simulation. The pancake coils are modeled as current strips with the current going into or out of the screen, which is a reasonable representation for a single layer of closely spaced wires. Of course FEMM is only a 2D simulation so it is showing the results for current strips that are infinitely long: it will not give an accurate figure for the actual flux from coils, but it does give a picture of what the flux looks like.
With regard to the problem of searching for the NMR effect, having to adjust both the tuning of the pancake coils and the drive frequency together, I would hope that we can make a good stab at the field from the magnets. If we know that accurately then we also accurately know the NMR frequency, so the search window is not very wide.
Smudge
Quote from: Smudge on 2020.06.21, 07:04:16
Sorry, but that image is misleading as it doesn't show pancake coils.
...
Here is an FEMM result to show the flux between the pancake coils.
The micro compass measurement will demonstrate that empirically
First pancake coil is drying, it measures 115uH @ 100Khz.
Inner diameter is 7cm, outer diameter 13cm, 30 turns of 1mm wire
Micro compass on its way.
Itsu
So if you apply a 100V RMS, 4MHz sine tone to this coil, then 35mA RMS current will flow through it, because at this frequency its inductive reactance is 2890Ω
The capacitance needed to make a 115µH inductance resonate at 4MHz is 14pF. An impractical value because this coil also has some intrawinding capacitance, which is hard to measure. It can be calculated from its self-resonance frequency which can be determined experimentally with an FM sweep (which you've done many times before).
Also, when two pancake coils are connected and sandwiched, their inductance will change due to their connection and mutual inductance.
OK, 14pF is indeed somewhat impractical as probably the intrawinding capacitance will be similar.
When the coil is dry, i will try to establish its intrawinding capacitance.
It could be we have to go back in number of turns (30) to the original 11 or so.
Itsu
Here I show the calculation of self resonance and self capacitance of a pancake coil (note the description of the video), but Itsu knows that anyway.
https://www.youtube.com/watch?v=spQ9yLdb7v4
Small capacitors (like 14 pF) can be easily built and they are even variable:
Variant 1: Cut two pieces from a metal sheet (aluminium, copper, steel) about 5 cm x 10 cm. Put a plastic foil in between and let the two metal sheets overlap more or less to adjust capacitance. Use bigger or smaller metal sheets to adjust maximum capacitance.
Variant 2: Cut two lengths of insulated wire (10 cm or 20 cm) and twist the two wires for a shorter or longer length to adjust capacitance, see screenshot of a 1 pF twisted wire cap.
https://www.youtube.com/watch?v=v1wWL6TGWOE This person makes a 6 pF cap at the end of the video.
Greetings, Conrad
Don't forget that the two pancake coils are connected in series opposing, and they are relatively close to each other. In FEMM I get 100uH for one coil and 67uH for the two coils in series opposing. So that 100:67 is the ratio you can expect.
Smudge
Thanks Conrad, thanks Smudge.
I did some measurements on the pancake coil, like the FM sweep (1Khz - 10Mhz), see screenshot 1 and a square wave ping and capture the damped oscillations, see screenshot 2.
Both show a resonance point of the pancake coil (30t) to be around 4.58Mhz or so.
Probe used was 3.9pF / 10MOhm.
Itsu
The 1st screenshot was taken with the other scope / probe, meaning 8pF / 10MOhm.
White is when the probe tip is not touching the coil (only the ground lead), blue is when the probe tip is connected too.
The resonance changes from about 4.5Mhz (white) to 3.3Mhz (blue).
The 2th screenshot is from my Spectrum Analyzer / Tracking Generator sweeping the same pancake coil.
Here the resonance found is around 4.45Mhz.
Itsu
So the self-resonance of this one coil is approximately the NMR of hydrogen in the magnetic field density of typical ceramic magnets. All without any external capacitances.
If we assume the 115uH to be accurate then the interwinding capacitance can be estimated as ~12pF.
Of course that will change once you connect and sandwich two such coils.
Right, and that change will be downwards for both inductance and interwinding capacitance, thus i can continue with a second 30 turns pancake coil.
Itsu
Quote from: conradelektro on 2020.06.22, 10:04:39
Here I show the calculation of self resonance and self capacitance of a pancake coil (note the description of the video), but Itsu knows that anyway.
The 1 pF decoupling capacitor was a good idea.
I am somewhat puzzled by the excitor coil's axis being perpendicular to the pancake coil's axis. Can you elucidate that geometrical arrangement?
Also, in my opinion the most natural and energy preserving state for a coil is a shorted coil (conversely, for a capacitor it is an open circuit) so if you short the pancake coil and do your measurements across a loosely coupled Tx/Rx exciter/sense coil (which has a low inductance and low capacitance), then these measurements will be even more accurate.
Alternatively, a "ping" by a powerful nanopulser fed
* into the exciter coil would work well, too, because the self-oscillations of a shorted pancake coil would decay
for a long time after the "ping". To avoid the need for RF Tx/Rx switching you could use a separate loosely coupled "Rx/sense coil".
*
In order to prevent exciter coil's self-oscillation, the exciter coil should be fed through a good series rectifying diode ( not a diode with DSR properties ! ).
A good rectifying diode will block the reverse current in the exciter coil IMMEDIATELY and will allow it to self-oscillate only for 1/2 cycle. OTOH, DSR diodes and other crappy diodes do NOT block the reverse current immediately
Quote from: verpies on 2020.06.23, 09:17:54
The 1 pF decoupling capacitor was a good idea.
I am somewhat puzzled by the excitor coil's axis being perpendicular to the pancake coil's axis. Can you elucidate that geometrical arrangement?
Thank you for the measurement education. Doing truthful and exact measurements of analog circuits is more difficult than the layman realises. I have no electronics education only mathematics, software, hardware near software and law.
The perpendicular position of the exciter coil in my video was dictated by laziness. I could just wind it over the plexiglas coil support without any additional fixtures. And because it somehow worked I left it like this.
The measurements in my video were inspired by TinselKoala und Gyulasun helping a tinkerer. I soon lost interest. I like to build things with my hands, just a hobby. And I like the fringe of all sciences out of curiosity. Normal things are so boring. But I dislike people who pretend to have built or done something extraordinary. Trying the impossible is courageous, but lying, cheating and having illusions is bad.
Greetings, Conrad
Quote from: conradelektro on 2020.06.23, 09:51:00
I have no electronics education only mathematics, software, hardware near software and law.
What programming languages are you good at ?
Quote from: conradelektro on 2020.06.23, 09:51:00
The perpendicular position of the exciter coil in my video was dictated by laziness.
That is an honest answer and it explains a lot. Such geometric arrangement of the coil makes it loosely coupled to the pancake coil. That is a good thing in this case even if it is inadvertent.
Quote from: conradelektro on 2020.06.23, 09:51:00
I could just wind it over the plexiglas coil support without any additional fixtures. And because it somehow worked I left it like this.
Loose magnetic coupling minimizes the influence of the exciter coil on the DUT. To be optimal, it should be placed further away to minimize the capacitive coupling, too.
Quote from: conradelektro on 2020.06.23, 09:51:00
Normal things are so boring.
How true!
Quote from: conradelektro on 2020.06.23, 09:51:00
But I dislike people who pretend to have built or done something extraordinary. Trying the impossible is courageous, but lying, cheating and having illusions is bad.
There is an immense pull for the human mind across this demarcation line.
Quote from: verpies on 2020.06.23, 10:00:19
What programming languages are you good at ?
The programming language used is a secondary consideration. If you ask a butcher "what knife do you use", he will tell you "the one required for the task". One is good in a certain software area, like database systems, networks, operating systems, event driven applications, hardware drivers, and so on. Depending on the available software tools (compilers, assemblers, integrated development systems) for a certain processor platform, you chose the programming language. For internet based applications it is often Java and JavaScript. Drivers for various hardware are nowadays written in C or C++, at my time all this was written in Assembler, because it is time critical. Often the use of a certain programming language is driven by hype, like the Ada hype or Cobol in the 70ies. I would also call Java a hype. There are many very good tools for C++ on many processor platforms.
Let's not deviate from the topic, which is NMR.
Greetings, Conrad
Quote from: conradelektro on 2020.06.23, 11:44:47
Let's not deviate from the topic, which is NMR.
This is not a deviation from the topic because I wrote a piece of software for NMR that interfaces with Itsu's equipment. I would like to finish it and when I read that you can program, I thought you might be able to help me finish it.
The software would make his life easier ...and IF HE CHOOSES TO, it would allow us to do NMR experiments remotely on his equipment, such as interesting pulse sequences for spin echo experiments. I could also give Itsu remote access to my equipment and to some of you, too.
Quote from: conradelektro on 2020.06.23, 11:44:47
The programming language used is a secondary consideration. If you ask a butcher "what knife do you use", he will tell you "the one required for the task".
Not in this case because, the part I already have is written in concrete language.
Quote from: conradelektro on 2020.06.23, 11:44:47
For internet based applications it is often Java and JavaScript.
A necessary evil
Quote from: conradelektro on 2020.06.23, 11:44:47
Drivers for various hardware are nowadays written in C or C++, at my time all this was written in Assembler, because it is time critical.
I think that drivers with Interrupt Service Routines should still be written in Assembler. The interrupt mechanism is always platform specific so there is no advantage in the portability of higher-level languages. A similar case can be made for code serving other hardware peripherals such as DMA, timers, UARTs, etc...
Less demanding drivers can be written in C because nowadays compilers generate well optimized code. C++ is OK for non-realtime drivers, too, but its exception handling scheme is generally incompatible with SEH in kernel mode. However if one knows what's going on behind the scenes and implements all the speedups offered by it, such as R-value references then the results can be acceptable.
Quote from: conradelektro on 2020.06.23, 11:44:47
Often the use of a certain programming language is driven by hype, like the Ada hype or Cobol in the 70ies.
I didn't know that about Cobol! I thought that in the 70s, it was all they had...
Quote from: conradelektro on 2020.06.23, 11:44:47
I would also call Java a hype.
Oh definitely !
In my opinion any language that doesn't compile to machine code is not worth my time. All this hand-holding just takes away too much control from me as a programmer >:(
Quote from: verpies on 2020.06.23, 12:57:47
This is not a deviation from the topic because I wrote a piece of software for NMR that interfaces with Itsu's equipment. I would like to finish it and when I read that you can program, I thought you might be able to help me finish it.
The software would make his life easier ...and IF HE CHOOSES TO, it would allow us to do NMR experiments remotely on his equipment, such as interesting pulse sequences for spin echo experiments. I could also give Itsu remote access to my equipment and to some of you, too.
I agree, we were doing some NMR on iron toroid experiments last year when interrupted by domestic problems overhere.
Does this interfacing concerns the mentioned airspy receiver (which i now have one) mentioned here:
https://www.overunityresearch.com/index.php?topic=3691.msg77379#msg77379
or is it something different?
Itsu
Quote from: Itsu on 2020.06.23, 15:47:43
Does this interfacing concerns the mentioned airspy receiver (which i now have one) mentioned here:
https://www.overunityresearch.com/index.php?topic=3691.msg77379#msg77379
or is it something different?
It is different because it involves interfacing and synchronizing your FG, SA and scope at the same time, as they all have Ethernet ports. I could incorporate the AirSpy as the receiver, too but because it has
only an USB port I'd have to find a linkable API for it. Not some damned C# library ! @&*$#
How to work with a scope from anywhere in the world:
1) Person A has a scope and a PC and Person B only has a PC.
2) Both persons download AnyDesk to their PCs https://anydesk.com/en/features/remote-support.
3) Person A connects his scope to his PC with an USB cable (the software comes with the scope) and allows Person B remote access to his PC via AnyDesk.
There are other programs for remote assistance, I just mentioned AnyDesk because it is free for personal use.
Greetings, Conrad
Quote from: conradelektro on 2020.06.23, 19:10:43
How to work with a scope from anywhere in the world:
Yes, remote access tools for single instruments exist even for free from the OEM but they do not allow for synchronization of these instruments. Especially through the LXI Ethernet ports they require a PC in the proximity of the instruments because of the USB latency, distance and routing limitation.
This is not much different than accessing a scope via intermediary PC running windows over RDP (which is built into the OS since WinXP for free).
Quote from: verpies on 2020.06.23, 19:14:49
Yes, remote access tools for single instruments exist even for free from the OEM but they do not allow for synchronization of these instruments. Especially through the LXI Ethernet ports they require a PC in the proximity of the instruments because of the USB distance and routing limitation.
Are the instruments to be synchronised connected to the same remote PC? If yes, the other person can see via AnyDesk everything that is happening on the remote PC (any instrument connected to the remote PC).
If the instruments to be synchronised are remote from each other (not connected to the same PC) it will be rather difficult.
Greetings, Conrad
Quote from: conradelektro on 2020.06.23, 19:22:17
Are the instruments to be synchronised connected to the same remote PC?
No, an intermediate PC is not even necessary when accessing these instruments through the LXI Ethernet ports and TCP/UDP/IP.
USB is not a serious control solution.
Quote from: verpies on 2020.06.23, 19:25:38
No, an intermediate PC is not even necessary when accessing these instruments through the LXI Ethernet ports and TCP/UDP/IP.
USB is not a serious control solution.
For remote TCP access to more than one instrument you need a switch and a firewall. And it is questionable whether your internet provider will allow that.
Greetings, Conrad
Quote from: conradelektro on 2020.06.23, 19:31:57
For remote TCP access to more than one instrument you need a switch and a firewall. And it is questionable whether your internet provider will allow that.
Not really, because of NAT-PNP, UPnP, IGDP and STUN, but I am sure Itsu has precise manual control over his own router with PAT/NAT and he has more than one device connected to his internal LAN, i.e. he is running them off some Ethernet switch. He's worked in the computer industry for a long time, after all.
Who in their right mind would give control of their LAN and router/gateway to the ISP, anyway?
Quote from: verpies on 2020.06.23, 19:03:43
It is different because it involves interfacing and synchronizing your FG, SA and scope at the same time, as they all have Ethernet ports. I could incorporate the AirSpy as the receiver, too but because it has only an USB port I'd have to find a linkable API for it. Not some damned C# library ! @&*$#
OK, but what is the benefit of being able to interfacing (synchronizing ok, i see that) to my FG, SA and scope or me being able to connect to someone else his?
Itsu
itsu
please remove this post if not on topic or miss understood ...question to Verpies on this topic
are you saying persons who do not have access to equipment but do have skill set to
use such
could remotely experiment or do actual useful work for this open source community ?
that would be remarkable IMO
we have equipment sitting dormant in some places [Ernies equipment ATM as example
sorry if I misunderstood [and interruption].
@ Chet
Yes, you understood correctly.
@Itsu
It would allow you to se, e.g. how differently my system behaves than yours and it would allow me the same with yours.
It is better than making movies according to individual requests, "zoom on this portion of the signal, measure the spectrum a little later after the stimulating pulse, space these pulses a little, further, oh no, no go back you missed that spike...."
Of course, it is not possible to connect the probes remotely, so they would have to be preset by you.
Right, so it would be on a one by one relation combined with a skype phone link or so to do some directly driven measurements, got it.
In my working environment i had an FTP server running using NAT, so something similar would be feasable now again.
Itsu
Quote from: Itsu on 2020.06.24, 10:03:17
In my working environment i had an FTP server running using NAT, so something similar would be feasable now again.
Configuring your router to automatically open random TCP ports for incoming FTP data connections in the PASV mode is much more difficult than allowing someone to communicate with your instruments using the TCP/IP protocol over their Ethernet interfaces. FTP is much more difficult to accommodate !
But the biggest advantage to you would be to automate and synchronize your
local measurementsa and signal generation. For example you could insert a
Varactor diode (https://youtu.be/icw8terKP-M?t=110) in series with the pancake coils forming a tunable LC circuit, by driving the varactor diode with DC from the 2nd channel of your FG in a synchronized manner via a precomputed lookup table, you will be able to have its LC resonance frequency follow the frequency output of the 1st FG channel. No OEM RAT software would be able to accomplish that...and that is just one example.
As for the remote collaboration, for example, I could remotely precompute the values in that lookup table (
varactor bias V vs. Ch1.freq (https://youtu.be/icw8terKP-M?t=110)), while you sleep.
Thanks, i see what you mean allthough we are talking about very specific equipment and procedures it could be a benefit for setting up measurement tests.
Seems that i can access OUR again normally after yesterdays lockup.
Building the 2th pancake coil but it takes severall trials to get it right with this 1mm thick wire and the present hot weather overhere does not help much.
I received the micro compass, so i hope to make some measurements with it tonight.
Regards Itsu
Quote from: Itsu on 2020.06.25, 08:59:20
Thanks, i see what you mean allthough we are talking about very specific equipment and procedures it could be a benefit for setting up measurement tests.
Yes but I have some of the same equipment that you have and having the source code at my disposal and proficiency to alter it at will to accommodate for the differences makes them inconsequential.
Quote from: Itsu on 2020.06.25, 08:59:20
Seems that i can access OUR again normally after yesterdays lockup.
I experienced the same problem. I have narrowed it down to the presence of the
percent character in messages.
IMO this forum's code does not escape this character properly and the HTTP server treats it as some kind of attack and bans the IP of the sender.
Quote from: Itsu on 2020.06.25, 08:59:20
Building the 2nd pancake coil but it takes several trials to get it right with this 1mm thick wire and the present hot weather over here does not help much.
Yes, heat is a show stopper for me, too. That is why I pump the cold water from my pond to the radiator in my workshop. This is cheap because the pond cools off naturally at night.
Quote from: Itsu on 2020.06.25, 08:59:20
I received the micro compass, so i hope to make some measurements with it tonight.
How small is it ?
OK, on the above, hopefully the forum's code can be debugged now.
The micro compass is 13mm diameter.
Itsu
Quote from: Itsu on 2020.06.25, 15:41:32
The micro compass is 13mm diameter.
Hmm, mine is 4mm in diameter and floats in oil.
QuoteHmm, mine is 4mm in diameter and floats in oil.
Well, 13mm is more then micro enough for my eyes :o
Would it matter to go even smaller?
Anyway, here the compass behaviour with 1A running through the pancake coil:
https://www.youtube.com/watch?v=a3RQ2Thx_b8
I also used my hall sensor to sweep by the coil (middle again) and capture the voltage trace on a slow time base setting to show the magnetic field, see screenshot (1A again).
Itsu
Quote from: Itsu on 2020.06.25, 19:57:21
Well, 13mm is more then micro enough for my eyes :o
Would it matter to go even smaller?
It matters if your two pancake coils are closer to each other than 13mm in your next measurement.
Also, these liquid filled bubble micro compasses rotate in 3D so they visualize all dimensions of the magnetic field.
Anyway, for NMR purposes the optimum placement of the water tube is where the compass has the North pointing left or right ( as per the directions in your video), when measuring between two coils.
Quote from: Itsu on 2020.06.25, 19:57:21
I also used my hall sensor to sweep by the coil (middle again) and capture the voltage trace on a slow time base setting to show the magnetic field, see screenshot (1A again).
That was pretty clever.
If you power two coils with any AC, then the hall sensor will visualize where between the two parallel pancake coils, the RADIAL AMPLITUDE of the magnetic field is the greatest. The 1D Hall sensor has to be oriented carefully to sense magnetic flux in the radial direction only.
For NMR we are interested in maximizing the radial component of the magnetic field where the water tube is.
On the diagram below, the radial direction is depicted as any line which is perpendicular to the red major symmetry axis and is also passing through the blue point. Two such blue lines on this diagram illustrate two of the infinitely many possible radial directions.
Imagine your sensor or compass sliding along one of these blue lines.
Thanks, will do the AC test lateron, first i have the 2 pancake coils positioned 15mm apart from each other both pointing the same way (coils to front).
Both pancake coils are similar wound meaning both CW and both measure 115.6uH @ 100Khz.
I have connected them in series aiding first.
Sweeping the compass shows north on the front, south on the back and flipping north to south inbetween the coils in the middle.
When connecting them series bucking, i have south on front and back and east/west inbetween on the right half swapping over west/east on the left half.
Guess that last one is the desired field.
Video here: https://www.youtube.com/watch?v=us5_F7dKsy0
Itsu
Quote from: Itsu on 2020.06.26, 20:00:19
When connecting them series bucking, i have south on front and back and east/west inbetween on the right half swapping over west/east on the left half.
Guess that last one is the desired field.
Yes
Quote from: Itsu on 2020.06.26, 20:00:19
Both pancake coils are similar wound meaning both CW and both measure 115.6uH @ 100Khz.
And what is their inductance when connected as series bucking ? What does your SA+TG think of their frequency response when they are 15mm apart ?
These coils take a lot of work to build, don't they? How many meters of wire did you use for them ?
I was lazy and cannibalized my coils out of an inductive cooker.
They are wound with Litz wire in these cookers, which I thought to be very handy because the skin effect depth in copper at 4MHz is 0.033mm :(
This means that a 1mm diameter solid copper wire has 8x more resistance to 4MHz AC, than to DC. That's not counting the proximity effect and inductive reactance, which I've already calculated before.
Itsu and Verpies,
Most excellent video!! This is a precise example of the
kind of research and documentation which is so very
important to the studies of electromagnetic phenomena!
Thanks to both of you for showing all how it should be
done! Your dedicated efforts are truly appreciated.
Quote from: verpies on 2020.06.26, 22:01:39
Yes
And what is their inductance when connected as series bucking ? What does your SA+TG think of their frequency response when they are 15mm apart ?
These coils take a lot of work to build, don't they? How many meters of wire did you use for them ?
I was lazy and cannibalized my coils out of an inductive cooker.
They are wound with Litz wire in these cookers, which I thought to be very handy because the skin effect depth in copper at 4MHz is 0.033mm :(
This means that a 1mm diameter solid copper wire has 8x more resistance to 4MHz AC, than to DC. That's not counting the proximity effect and inductive reactance, which I've already calculated before.
A quick measurement shows in series bucking at 15mm apart they measure 106uH @ 100Khz.
I will do some further tests tonight, when hopefully the workshop has cooled down somewhat.
The coils are more difficult to make then i expected, as the plexiglas molds are to weak when trying to tightly winds the 1mm magnet wire, they flex causing the windings to overlap if not carefull.
I did not measure the wire length, but at an average of 31cm circumference (@ 10cm diameter) it would be around 9.3m.
Good idea on the litz wire, i have some, but its to thick (2.5mm dia) and to short (10m), but i keep it in mind when building some better pancakes.
Itsu
Quote from: muDped on 2020.06.27, 02:25:00
Itsu and Verpies,
Most excellent video!! This is a precise example of the
kind of research and documentation which is so very
important to the studies of electromagnetic phenomena!
Thanks to both of you for showing all how it should be
done! Your dedicated efforts are truly appreciated.
Thanks MuDped,
you are welcome, and your comments are truly appreciated.
Itsu
Quote from: Itsu on 2020.06.27, 08:18:18
Good idea on the litz wire, i have some, but its to thick (2.5mm dia) and to short (10m), but i keep it in mind when building some better pancakes.
How much does the Litz wire cost in your area?
In mine, the 1mm diameter Litz wire (66 wires 0.08mm ea.) costs 0.72 EUR for 1m.
A quick look around:
Conrad shows €1,24 / m
https://www.conrad.nl/p/pack-litz-wire-koperdraad-gelakt-buitendiameter-incl-isolatielak1-mm-buitendiameter-excl-isolatielak-080-mm-40-1402140
Reichelt €0,93 / m
https://www.reichelt.nl/200g-koperen-litzedraad-op-spoel-litzedraad-120x0-1mm-cli-200-120-p57209.html?&trstct=pol_0&nbc=1
So it depends, but around €1 / m
Itsu
On ebay one can find quite good deals for speaker wires which have two cores but can be ripped apart into a single core wire:
100m 2x0.5mm Speaker Cable Transparent Twin Loudspeaker Wire Car Home Audio Hifi
https://www.ebay.nl/itm/100m-2x0-5mm-Speaker-Cable-Transparent-Twin-Loudspeaker-Wire-Car-Home-Audio-Hifi/302652032514
13, 42, 79 Strand Speaker Cable Twin Stranded CCA Figure 8
https://www.ebay.nl/itm/13-42-79-Strand-Speaker-Cable-Twin-Stranded-CCA-Figure-8/333606535701
Greetings, Conrad
Quote from: conradelektro on 2020.06.27, 14:53:34
On ebay one can find quite good deals for speaker wires which have two cores but can be ripped apart into a single core wire:
Yes, some speaker cables can be easily separated into two separate stranded wires.
Are the strands of one of these wires, isolated from each other?
The 2 pancake coils, 15mm distance parallel, in series bucking on the SA / TG, see screenshot.
Seems the selfresonance has shifted down somewhat to 4Mhz due to the lower combined bucking inductance.
Itsu
Quote from: verpies on 2020.06.25, 20:42:43
It matters if your two pancake coils are closer to each other than 13mm in your next measurement.
Also, these liquid filled bubble micro compasses rotate in 3D so they visualize all dimensions of the magnetic field.
Anyway, for NMR purposes the optimum placement of the water tube is where the compass has the North pointing left or right ( as per the directions in your video), when measuring between two coils.
That was pretty clever.
If you power two coils with any AC, then the hall sensor will visualize where between the two parallel pancake coils, the RADIAL AMPLITUDE of the magnetic field is the greatest. The 1D Hall sensor has to be oriented carefully to sense magnetic flux in the radial direction only.
For NMR we are interested in maximizing the radial component of the magnetic field where the water tube is.
On the diagram below, the radial direction is depicted as any line which is perpendicular to the red major symmetry axis and is also passing through the blue point. Two such blue lines on this diagram illustrate two of the infinitely many possible radial directions.
Imagine your sensor or compass sliding along one of these blue lines.
I tried to do that AC test mentioned in post #51, using a 220V:12V transformer feeding a 12V automotive bulb in series with the both pancake coils.
So there was 1.5A AC rms @ 50Hz running through the coils, but the hall sensor picked up only some small signals from the coils with probably a lot of 50Hz noise interference from my workshop
No steady signal was seen on the scope which made sense to me.
Itsu
Quote from: Itsu on 2020.06.27, 20:17:04
So there was 1.5A AC rms @ 50Hz running through the coils, but the hall sensor picked up only some small signals from the coils with probably a lot of 50Hz noise interference from my workshop
I was afraid of that.
Do you have a powerful source of higher frequency AC ...e.g. from one of these inductive metal heaters, or cookers, or an old project like
this one (https://www.youtube.com/watch?v=RU15rQEV-Rg&feature=youtu.be) ?
Quote from: Itsu on 2020.06.27, 16:11:48
Seems the selfresonance has shifted down somewhat to 4Mhz due to the lower combined bucking inductance.
How has the resonance frequency shifted down, when lower inductance shifts up this frequency, because it is in the denominator of f=1/2π√(LC) ??? The peak #1 does not seem like the LC resonance (even with the distributed capacitance). How do you have this connected up for measurement ?
Do you have a suitable diode to try how much you can "pull" the frequency as in
this simple experiment (https://youtu.be/icw8terKP-M?t=110) ?
QuoteI was afraid of that.
Do you have a powerful source of higher frequency AC ...e.g. from one of these inductive metal heaters, or cookers, or an old project like this one ?
Yes i still have that, so will see if i can use it.
QuoteHow has the resonance frequency shifted down, when lower inductance shifts up this frequency, because it is in the denominator of f=1/2π√(LC) ???
Yeah right, my brain was still in bucking mode i guess.
So why is with the lower inductance (106uH instead of the 115uH the LITTLE peak at #1 lower in frequency? Or is the new resonance peak the one at #2 (around 7Mhz), so indeed higher thus at 4.8pF distributed /selfcapacitance?
QuoteThe peak #1 does not seem like the LC resonance (even with the distributed capacitance).
How do you have this connected up for measurement ?
See diagram below, same as with the single coil measurement as shown in post #25
QuoteDo you have a suitable diode to try how much you can "pull" the frequency as in this simple experiment ?
I don't think i have a varactor diode, but any diode would work a little like that i understand; i will take a look / test.
I will try to wind the plastic tube coil as the severall layers boat varnish are dried enough by now to make it watertight.
Itsu
I did a quick FM sweep from 1Khz to 10Mhz with the 2 pancake coils 15mm apart parallel while in series bucking, see screenshot.
The loosely coupled FG loop was placed inbetween the 2 coils, but its position has big influence on
the amplitude of the both peaks and a lesser influence on the resonance frequencies.
So we see resonance peaks on 3.5 and 4.6Mhz.
I did some tests 3 years ago using bifilar pancake coils closely parallel to each other (3:40) here:
https://www.youtube.com/watch?v=AQEJ08TRNkU&t=245s
It showed that putting them to close their resonance points never merged.
Itsu
Quote from: Itsu on 2020.06.28, 08:58:45
So why is with the lower inductance (106uH instead of the 115uH the LITTLE peak at #1 lower in frequency? Or is the new resonance peak the one at #2 (around 7Mhz), so indeed higher thus at 4.8pF distributed /selfcapacitance?
See diagram below, same as with the single coil measurement as shown in post #25
The key to understanding this behavior is remembering that at the resonance frequency:
1) An ideal series LC circuit acts as a short circuit.
2) An ideal parallel LC circuit acts as an open circuit.
But these are not ideal LC circuits, thus at resonance:
a) The current flowing in a series LCR circuit is at the MAXIMUM.
b) The current flowing in a parallel LCR circuit is at the MINIMUM.
Because of the distributed intrawinding capacitance, these pancake coils act as two imperfect parallel LCR circuits...and when they are connected in series and are far apart, they act as two parallel LC circuits connected in series
When they are close together, their mutual inductance and and interwinding capacitance comes into play.
It is surprising that the two frequency peaks are so far apart for coils that are virtually identical.
Anyway, the SA essentially measures voltage at its input, but since its input has the 50Ω impedance, this voltage is proportional to the current flowing in this input impedance, too (V= i * 50Ω).
When you are measuring one coil and we treat it as a parallel LC circuit, we can expect it to act as on open circuit at resonance. Obviously, an open circuit doesn't allow any current through. No current flowing through the 50Ω input impedance, means no voltage appearing at the SA's input, because 0A * 50Ω = 0V.
With an imperfect LCR circuit, the voltage at the SA's input, would not be 0V, but it would be at the
minimum, which represents the dip denoted as dip #2. (it is noisy because the amplitude is low at that point - as it should be in this measurement setup).
If you want the parallel LC resonance to manifest itself as a high amplitude peak (not noisy), then connect it as on the diagram below. (calibrate it to a flat line after all the coil's cables and interconnects are in place ...but
before the coil itself is connected).
Thanks for the info / reminders / refreshing.
I put up some measurements as can be seen in the video.
Concerning your TG / SA setup diagram, this shows almost nothing as i had to put everything on its maximum power / sensitivity to see a very slight dip around 4.15Mhz, see screenshot below.
LCR meter measurements:
single coil 1: 116.2uH
single coil 2: 115.2uH
both coils 15mm apart, series aiding: 354uH
both coils 15mm apart, series bucking: 107.3uH
Sweep measurements:
single coil 1: 4.1Mhz resonance
single coil 2: 3.8Mhz resonance
both coils 15mm apart, measured separately: 2 peaks: 1st 2.9Mhz, second 6Mhz
both coils 15mm apart, series aiding: 2.2Mhz resonance
both coils 15mm apart, series bucking: 2 peaks: 3.5Mhz and 4.5Mhz.
TG / SA parallel measurement:
both coils 15mm apart, series bucking: 4.15Mhz
Video here: https://www.youtube.com/watch?v=gQivPoqBl0A
Itsu
Looks like the coil has low Q and is too weak* to short the TG's output to ground . The TG has output impedance of 50Ω.
Did you try to use the optional resistor to weaken the TG's output so the coil has a chance to "win" with the TG ?
* the coil's minimum impedance is too high
Yes, i tried 10K in series with the TG lead, but did not see any difference.
Let me try some different value's.
Quote from: Itsu on 2020.06.21, 19:07:44
First pancake coil is drying, it measures 115uH @ 100Khz.
Inner diameter is 7cm, outer diameter 13cm, 30 turns of 1mm wire
I cannibalized the other side of the cooker and wound a 30 turn pancake coil with 1mm Litz wire.
It measures 93.55µH (@100kHz) and 0.51Ω.
OD=127mm, ID= 59mm
Soon, I will measure its frequency params, too.
O0
Thats closer to the by Smudge calculated 100uH then my coils.
I wonder what its selfresonance frequency will be.
Itsu
I made the tube coil, it is 242 turns of 1mm wire, measuring 31uH @ 100Khz and 0.3 Ohm.
Itsu
Ok, that 4Mhz looks like the real one.
But i wonder how we get to 4.3Mhz resonance when the coils already selfresonante that low.
I tried your TG / SA setup with severall resistors between 1K and 120K Ohm, but it does not want to reveal its resonance point :(
itsu
Quote from: Itsu on 2020.06.28, 20:41:20
...
I tried your TG / SA setup with severall resistors between 1K and 120K Ohm, but it does not want to reveal its resonance point :(
...
Hi Itsu and Verpies,
Sorry to chime in,
Itsu, you may wish to test your TG / SA setup using a "normal" series LC circuit instead of the pancake coil?
To check the setup, use your toroid type T520-2 powdered iron core on which 6 turns gave you 24uH some time ago (this is from a last year post of yours) and make any coil on it in the range 110 -116 uH and resonate it with a series capacitor (18-22 pF) to get resonance around 3 - 3.5 MHZ range.
I mean using Verpies's drawing with the series resistor but replace the series L1 and L2 in it by the 'normal' series LC circuit to see how it behaves. This series LC circuit should give a well visible dip (several dB attenuation) at its resonant frequency because it should have higher overall Q factor than the pancake coil(s) have.
However, the question may arise, what self resonance a single pancake coil in itself may have: a series or a parallel resonance or both? In case it has a parallel resonance, then the pancake coil may be connected directly between the TG - SA 'center' pins where the optional resistance is shown in Verpies drawing. And a dip should appear at the parallel resonant frequency while a peak should manifest at a series resonant frequency. Do you agree?
Gyula
Quote from: Itsu on 2020.06.28, 18:43:28
I wonder what its selfresonance frequency will be.
That is hard to measure by SA+TG (actually I use VNA) because my coil rings for a very long time and if I do not put a long delay between sweeps, then when the next sweep comes, the coil is
still ringing from the previous one ...and I get garbage results (grass).
Actually, the coil needs a pause to settle not only between sweeps but also between acquisition of adjacent frequency points! ...this is an example where custom software gives greater control of the acquisition process.
Watch out for this gotcha, with DUTs that settle for a long time !
I've seen peaks around 3MHz, 6MHz and 35MHz. The grass is around the second one.
Quote from: gyula on 2020.06.28, 22:37:33
In case it has a parallel resonance, then the pancake coil may be connected directly between the TG - SA 'center' pins where the optional resistance is shown in Verpies drawing. And a dip should appear at the parallel resonant frequency while a peak should manifest at a series resonant frequency. Do you agree?
Itsu has already tried connecting the pancake coil between the TG - SA 'center' pins and he did get at dip,
see here (https://www.overunityresearch.com/index.php?topic=3924.msg82655#msg82655). That dip was noisy because in such setup, the pancake coil, which behaves as a parallel LC circuit, had the highest impedance and kept the signal amplitude low at the SA's input. This is to be expected with such connection.
The alternate connection is designed to yield the maximum amplitude at resonance. To get a peak (not dip) with a parallel LC circuit one can also measure the magnitude of its impedance at various frequencies (having an impedance bridge or VNA helps). Impedance will be highest at the resonance.
This is illustrated on the two plots of my coil below. The first one illustrates the magnitude of the impedance of the coil vs.frequency. Notice that near the peak the impedance reaches ~7kΩ !
The second one illustrates only the reactance of the coil vs frequency. Reactance is the imaginary component of impedance and is supposed to be ZERO at resonance (because at resonance, the inductive reactance and capacitive reactance cancel each other). ...and they do that at 6.3MHz.
The second plot also proves that the pancake coil behaves as an LC circuit up to 8MHz (anything above the 0 represents inductive reactance and below - capacitive reactance)
I am also planing a novel way of measuring the self resonance of the pancake coil.
I will short the pancake coil (cats like to be petted and coils like to be shorted C.C ) and I will put my non-contact current probe around this short.
Then I will discharge my HV stun gun into a 1 turn loop of Litz wire positioned somewhat far from the pancake coil.
This will allow me to observe the decaying self-oscillations of the pancake's current on the scope and independently from any disturbances, that the 1 turn loop may cause, because once the spark terminates, the 1 turn loop will become effectively open-circuited (effectively non-existent).
I think the same could be accomplished with a little less extreme HV source and an avalanching BJT transistor in place of the spark gap.
Quote from: gyula on 2020.06.28, 22:37:33
Hi Itsu and Verpies,
Sorry to chime in,
Itsu, you may wish to test your TG / SA setup using a "normal" series LC circuit instead of the pancake coil?
To check the setup, use your toroid type T520-2 powdered iron core on which 6 turns gave you 24uH some time ago (this is from a last year post of yours) and make any coil on it in the range 110 -116 uH and resonate it with a series capacitor (18-22 pF) to get resonance around 3 - 3.5 MHZ range.
I mean using Verpies's drawing with the series resistor but replace the series L1 and L2 in it by the 'normal' series LC circuit to see how it behaves. This series LC circuit should give a well visible dip (several dB attenuation) at its resonant frequency because it should have higher overall Q factor than the pancake coil(s) have.
However, the question may arise, what self resonance a single pancake coil in itself may have: a series or a parallel resonance or both? In case it has a parallel resonance, then the pancake coil may be connected directly between the TG - SA 'center' pins where the optional resistance is shown in Verpies drawing. And a dip should appear at the parallel resonant frequency while a peak should manifest at a series resonant frequency. Do you agree?
Gyula
Hi Gyula,
thanks to chime in, yes as verpies mentioned, i tried "the TG - SA 'center' pins" setup which i normally do as the other setup is kind of unnatural to me (shorting out the TG / SA), but of course works OK on different circuits as explained by verpies.
But this shows with the pancake coil some, for me, strange outcome (see post #61) which probably is caused by the specific series bucking mode setup of the both pancake coils.
Changing over to the verpies setup (TG / SA parallel) shows virtually nothing.
Regards Itsu
verpies,
thanks for the plots, nice stuff.
As max. impedance in a parallel LC is at resonance, so that would be at 7.8Mhz looking at plot #1.
But looking at plot #2, Xc and Xl cancel each other out at 6.3Mhz which also points to resonance.
I would have expected to have both resonances to be shown at the same frequency.
Again there is far more there then meets the eye :D
I will try to experiment with your novel way of measuring the self resonance of the pancake coil O0
Itsu
Quote from: Itsu on 2020.06.29, 10:01:22
As max. impedance in a parallel LC is at resonance, so that would be at 7.8Mhz looking at plot #1.
But looking at plot #2, Xc and Xl cancel each other out at 6.3Mhz which also points to resonance.
I would have expected to have both resonances to be shown at the same frequency.
Again there is far more there then meets the eye :D
Yes, apparently something increases the REAL part of the impedance after the IMAGINARY part (reactance) gets cancelled. There is more to the coil's impedance then mere imaginary reactance. For example, the real resistance of the coil matters, too (in mine it is 0.5Ω).
The skin effect works in the proper direction to explain this since as the frequency increases, it artificially increases the real resistance of the wire (8x in your case) but I don't think it can make such a large difference because 8x the real resistance of my coil is only 4Ω and not the 741Ω difference we see. So maybe some standing waves or the proximity effect are responsible for this increase.
But most likely I still have problems with my measurement technique, i.e the coil keeps ringing and does not settle between the acquisition of adjacent frequency points.
Ringing coils are also a big problem for the reception of the AC magnetic field form the Larmor precession in NMR. It is a problem without RF switching because they both occur at the same frequency and are only separated in time domain (yes, yes in spatial direction, too, but that does not matter practically).
All,
It may be that the SRF measurement is getting somewhat overly complicated. Here is an example of measuring the pancake coil shown below using a signal generator with a 50 ohm source impedance, a 100k ohm series resistor and the coil under test connected between the output of the 100k resistor and ground. A 50kHz pulse is fed to this network and the SRF is easily seen in the ring down on the scope trace.
Regards,
Pm
We've already tried it, see here (https://www.overunityresearch.com/index.php?topic=3924.msg82568#msg82568). We've gotten different results with a very similar coil.
So, I shorted the pancake coil with a 0.22Ω resistor. Stimulated it with a short magnetic pulse from a 1T coil (20cm away, parallel, on the same axis) powered by HV and spark gap ( Tesla style ! ).
Scoping across this 0.22Ω resistor gives me this result:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35361)
As you can see, the magnetic pulse induces approximately 2Apk-pk current pulse in the pancake coil and as soon as the spark terminates, the pancake coil is free to oscillate in isolation ( i.e. without any interference from the 1T coil )
Dead shorting the pancake coil and using a contactless current probe on the short, works too. I didn't post it because my current probe has 30MHz bandwidth limitation and doesn't look as sharp.
But the ringing signal looks distorted, so is this the 3MHz, 6MHz and 35MHz signals mentioned earlier all together?
It finally settles down to about 20Mhz (50ns between 2 peaks)?
Itsu
Quote from: Itsu on 2020.06.29, 15:01:15
But the ringing signal looks distorted, so is this the 3MHz, 6MHz and 35MHz signals mentioned earlier all together?
It is distorted while the arc is still going but as soon as the arc terminates it calms down. The arc has a 1ns rise time and 600ns fall time. Any ideas how to make it shorter without blowing on it ? Magnets ?
Anyway, when it calms down, counting the peaks over 5 divisions yields 21MHz. Before you ask: "Do you see it on the VNA impedance measurement ?", A: Yes, I do.
Ok, yes, blowing air over the arc i have seen Tinsel do that, magnets also suppose to work.
I put together the pancake coils (15mm apart), the tube coil inbetween and the 2 ring magnets on the outside for some resonance testing.
The input circuit was as shown below in the drawing as specified by Smudge.
I used a 100x higher value (10nF) then the 100pF trimmer.
The FG is sweeping again from 1Khz to 10Mhz, so each scope division is 1Mhz.
Resonance moves from 1.4Mhz (peak) to 2.8Mhz, so way to low for the needed 4.3Mhz NMR frequency.
So something has to be adjusted.
Video here: https://www.youtube.com/watch?v=QunSj9id3ME
Regards itsu
Quote from: Itsu on 2020.06.29, 16:20:46
The FG is sweeping again from 1Khz to 10Mhz, so each scope division is 1Mhz.
Resonance moves from 1.4Mhz (peak) to 2.8Mhz, so way to low for the needed 4.3Mhz NMR frequency.
So something has to be adjusted.
Well, the trimmer cap is certainly too big for the coils. I would be interested, how the removal of the toroidal coil alters that FM sweep. ...and what about the magnets?
The really bad news will be when you discover, that the signal injected into the pancake coils appears on the toroidal coil ...without water.
Hi Itsu and Verpies,
Thanks for your answers, and I should have read the earlier posts to collect all the info before my post. 8)
Itsu, I think in the present setup so far what you could do to increase the resonant frequency for the pancake coils is to reduce their number of turns to bring down the 115 uH to around 60-80 uH or so.
Gyula
Quote from: Itsu on 2020.06.29, 16:20:46
...
Resonance moves from 1.4Mhz (peak) to 2.8Mhz, so way to low for the needed 4.3Mhz NMR frequency.
So something has to be adjusted.
...
Quote from: verpies on 2020.06.29, 20:13:23
Well, the trimmer cap is certainly too big. I would be interested, how the removal of the toroidal coil alters that FM sweep. ...and what about the magnets?
The really bad news will be when you discover, that the signal injected into the pancake coils appears on the toroidal coil ...without water.
I can check if removing the magnets and the toroidal coil has any influence (probably yes) on the resonance point, but the bottom line is, they have to be there.
I did put a probe on the toroidal coil output and i did saw some signals there, but it depends on if there also is a probe on the input (grounding problem) or not at what frequency and amplitude.
I will do some further tests.
Itsu
Quote from: gyula on 2020.06.29, 21:25:49
Hi Itsu and Verpies,
Thanks for your answers, and I should have read the earlier posts to collect all the info before my post. 8)
Itsu, I think in the present setup so far what you could do to increase the resonant frequency for the pancake coils is to reduce their number of turns to bring down the 115 uH to around 60-80 uH or so.
Gyula
Hi Gyula,
i think reducing the number of turns of the pancake coils would be the only solution.
Originally, the design was for 11 turns each, see picture, but after some debate it was increased to the present 30 turns.
Also Smudge mentioned here in post #23 https://www.overunityresearch.com/index.php?topic=3924.msg82567#msg82567 that according to his FEMM calculations, the pancake coils in series opposing would be around 67uH, which match better with your 60 - 80 uH needed for 4.3Mhz.
Itsu.
Quote from: Itsu on 2020.06.30, 08:15:15
I can check if removing the magnets and the toroidal coil has any influence (probably yes) on the resonance point, but the bottom line is, they have to be there.
Yes they do, but judging whether their presence increases or decreases this frequency...and by how much, points the way forward. There are things that can be done, such as spacing, isolating, litzing, antilongitudal windings. There is even a difference whether the center of the signal coax is connected to the middle of the pancake coil vs. the outer diameter.
Quote from: Itsu on 2020.06.30, 08:15:15
I did put a probe on the toroidal coil output and i did saw some signals there, but it depends on if there also is a probe on the input (grounding problem) or not at what frequency and amplitude.
Yes, signal crosstalk through the ground path is a show stopper.
Quote from: Itsu on 2020.06.30, 08:24:42
i think reducing the number of turns of the pancake coils would be the only solution.
I told you so
here (https://www.overunityresearch.com/index.php?topic=3924.msg82539#msg82539).
OK, I still have some 1mm Litz wire to make smaller coils, which will match yours. I just have to buy more SMA panel connectors and CD spools for coil-formers. Because of this, my O.D. will be 127mm again.
Don't forget that the NMR frequency is determined by the static field within the water from the magnets. Has this field been measured yet? Moving the magnets further apart will reduce the field and the NMR frequency, but at the expense of the field not being parallel to the horizontal axis. So if winding new coils is a pain then perhaps moving the magnets will allow some NMR work to proceed at a lower frequency.
Smudge
Quote from: Smudge on 2020.06.30, 10:13:51
Don't forget that the NMR frequency is determined by the static field within the water from the magnets.
Moving the magnets further apart will reduce the field and the NMR frequency, but at the expense of the field not being parallel to the horizontal axis.
Yes, my dear doctor. This is elementary
Quote from: Smudge on 2020.06.30, 10:13:51
Has this field been measured yet?
I don't think Itsu has a calibrated gaussmeter.
Quote from: Smudge on 2020.06.30, 10:13:51
Don't forget that the NMR frequency is determined by the static field within the water from the magnets. Has this field been measured yet? Moving the magnets further apart will reduce the field and the NMR frequency, but at the expense of the field not being parallel to the horizontal axis. So if winding new coils is a pain then perhaps moving the magnets will allow some NMR work to proceed at a lower frequency.
Smudge
Technical specifications for the ring magnets grade Y30
Gauss (G) Tesla (T) kOe kA/m kOe kA/m MGOe kJ/m 3 °C
≥ 3900 ≥ 0.39 ≥ 2.3 ≥ 184 ≥ 2.3 ≥ 188 ≥ 3.4 ≥ 27.6 ≤ 250
from: https://www.magnet-shop.com/ferrite/ringmagnets/ringmagnet-140.0-x-60.0-x-20.0-mm-y30-ferrite
Looking for a decent gaussmeter app for my Iphone, most only handle uTesla's.
Itsu
Quote from: verpies on 2020.06.30, 08:43:03
Yes they do, but judging whether their presence increases or decreases this frequency...and by how much, points the way forward. There are things that can be done, such as spacing, isolating, litzing, antilongitudal windings. There is even a difference whether the center of the signal coax is connected to the middle of the pancake coil vs. the outer diameter.
Yes, signal crosstalk through the ground path is a show stopper.
Removing the magnets and/or the toroidal coil has no noticable influence on the pancake coils resonance frequency range.
So i guess i need to start removing some turns.....
Itsu
Quote from: Itsu on 2020.06.30, 15:14:22
Removing the magnets and/or the toroidal coil has no noticable influence on the pancake coils resonance frequency range.
That is excellent news !
I went in and removed some turns of the both pancake coils so i am left with 11turns each.
Sweeping with the FG now shows the 4.3Mhz resonance peak is somewhere in the middle of the 100pF trimmer cap.
I will do some further measurements tomorrow.
Regards Itsu
Quote from: Itsu on 2020.06.30, 20:40:01
I went in and removed some turns of the both pancake coils so i am left with 11turns each.
What's the O.D. and I.D. ?
O.D. = 112mm, I.D. = 89mm.
Itsu
Toroidal coil inserted inbetween the pancake coils and loaded with a 100pF trimmer cap.
Adjusted the pancake coils to resonate at 4.3Mhz, see yellow peak.
Then adjusted the toroidal coil trimmer cap to max output on 4.3Mhz, see red trace (differential probing).
Input trimmer cap about half way (50pF?), output trimmer cap barely active (20pF?).
The yellow trace seems to split, not sure what that 2th yellow and red peak / artifact is.
Itsu
I have just done some FEMM simulations for the pair of ring magnets. FEMM has data for ceramic 5 that closely matches the Y30 data, ie Br = 0.39T. I did two runs, (a) with the separation between the magnet faces of 14mm the field in the water region is 0.124T, and (b) with a separation of 28mm the field drops to 0.083T. From Itsu's videos it looks like the 28mm separation is about right. That 0.083T should give a NMR frequency of 3.53MHz. However this is only a guide because FEMM is not a 3D simulation, it gives the field between infinitely long strips, see image below.
Smudge
Thanks Smudge,
i measured the separation between the magnet faces to be 25mm in my present setup.
So what for an effect should i be looking for when sweeping the frequency and adjusting the trimmer pots for any NMR?
What would the range of the output load be, 100 Ohm?
Itsu
Quote from: Itsu on 2020.07.01, 10:54:43
What would the range of the output load be, 100 Ohm?
Please take out your SWR bridge, connect it to the SA and measure the magnitude of the complex impedance
* of the toroidal coil at the frequencies of interest.
This will give you the empirical impedance of the toroidal coil (...and its cap) providing better data for choosing a good load for this coil.
* √(R
2+X
2)
Quote from: Itsu on 2020.07.01, 10:54:43
Thanks Smudge,
i measured the separation between the magnet faces to be 25mm in my present setup.
OK, my FEMM gives a field of 0.0895T at that separation which equates to a NMR frequency of 3.8Mhz.
QuoteSo what for an effect should i be looking for when sweeping the frequency and adjusting the trimmer pots for any NMR?
What would the range of the output load be, 100 Ohm?
First off I would want to minimise any unwanted coupling to the toroidal output coil, so what you posted earlier today with red and yellow traces needs looking at. Ideally with no water present and both coils tuned to the NMR frequency we want zero coupling. That may not be achievable. That split or dip in the yellow trace is to be expected when there is coupling since the other coil is sucking energy from the pancake coil. It does show you that the other coil's resonance is at or about the same frequency. With the toroidal coil's field at right angles to the pancake coil's field there should be zero magnetic coupling, so it is probably down to capacitive coupling.
Although you correctly have the pancake coils in series opposing there are two ways this can be set up, either the outer connection of one pancake is connected to the inner connection of the other, or the connection is at both outers or both inners (the opposing field determined by which way round the pancakes are placed). With an unbalanced drive from the signal source I can't see any way to eliminate capacitive coupling for either configuration. Ideally it requires a balanced drive, then one configuration is better than the other. If the drive is at both outers while the series connection is at both inners (or vice versa), then the capacitive coupling is minimised. So I am sorry to say that you need a wide bandwidth balun, something you have already played with on another thread.
One you achieve minimum unwanted coupling, then you can add water and look for the NMR signal which by its gyratory action couples the 90 degree input and output fields together. While searching for this I would not necessarily deliberately load the output coil, I would go for a high load resistance if any. This keeps the output coil Q high which give you a better chance of finding a small NMR signal , but at the expense of having to tune both coils at each search frequency. Not sure using frequency sweep is the right way to do this, but I could be wrong there. I had envisaged a manual "sweep" in small steps.
Smudge
Quote from: verpies on 2020.07.01, 12:14:49
Please take out your SWR bridge, connect it to the SA and measure the magnitude of the complex impedance* of the toroidal coil at the frequencies of interest.
This will give you the empirical impedance of the toroidal coil (...and its cap) providing better data for choosing a good load for this coil.
* √(R2+X2)
Thanks, i will play with that tonight.
Itsu
Quote from: Smudge on 2020.07.01, 14:46:19
OK, my FEMM gives a field of 0.0895T at that separation which equates to a NMR frequency of 3.8Mhz.
First off I would want to minimise any unwanted coupling to the toroidal output coil, so what you posted earlier today with red and yellow traces needs looking at. Ideally with no water present and both coils tuned to the NMR frequency we want zero coupling. That may not be achievable. That split or dip in the yellow trace is to be expected when there is coupling since the other coil is sucking energy from the pancake coil. It does show you that the other coil's resonance is at or about the same frequency. With the toroidal coil's field at right angles to the pancake coil's field there should be zero magnetic coupling, so it is probably down to capacitive coupling.
Although you correctly have the pancake coils in series opposing there are two ways this can be set up, either the outer connection of one pancake is connected to the inner connection of the other, or the connection is at both outers or both inners (the opposing field determined by which way round the pancakes are placed). With an unbalanced drive from the signal source I can't see any way to eliminate capacitive coupling for either configuration. Ideally it requires a balanced drive, then one configuration is better than the other. If the drive is at both outers while the series connection is at both inners (or vice versa), then the capacitive coupling is minimised. So I am sorry to say that you need a wide bandwidth balun, something you have already played with on another thread.
One you achieve minimum unwanted coupling, then you can add water and look for the NMR signal which by its gyratory action couples the 90 degree input and output fields together. While searching for this I would not necessarily deliberately load the output coil, I would go for a high load resistance if any. This keeps the output coil Q high which give you a better chance of finding a small NMR signal , but at the expense of having to tune both coils at each search frequency. Not sure using frequency sweep is the right way to do this, but I could be wrong there. I had envisaged a manual "sweep" in small steps.
Smudge
Smudge, thanks.
I will readjust to around 3.8Mhz.
My pancake coils are connected this way: the inner connection of one pancake is connected to the outer connection of the other, but set up in parallel mirror mode (both coils facing each other).
So i drive the both: one inner, other outer.
I to had in mind the manual "sweep", not the FG sweep i used up till now.
Itsu
Quote from: Itsu on 2020.07.01, 15:24:48
Smudge, thanks.
I will readjust to around 3.8Mhz.
My pancake coils are connected this way: the inner connection of one pancake is connected to the outer connection of the other, but set up in parallel mirror mode (both coils facing each other).
So i drive the both: one inner, other outer.
I to had in mind the manual "sweep", not the FG sweep i used up till now.
Itsu
Then you will not be able to minimise the capacitive coupling between the pancakes and the toroid. I think you need to
not have both coils facing each other, feed at both inner or outer and series connect at the other end. When fed with a balanced input this puts the radial centre line between the pancake coils at zero electric potential.
Smudge
Ok, but when i flip over one pancake coil, the 4mm thick plexiglas will be inbetween the toroidal coil
and this flipped pancake coil causing some unbalance, see light brown substrate in below picture.
Guess i need to rewind 1 pancake coil opposite direction as the other one.
Itsu
Quote from: Smudge on 2020.07.01, 14:46:19
First off I would want to minimise any unwanted coupling to the toroidal output coil, so what you posted earlier today with red and yellow traces needs looking at.
Yes, we should determine what the ratio of this coupling is. e.g. the ratio of the current amplitude fed into the pancake coils vs. received by the toroidal coil.
For the calculation of this ratio the I/O impedance should be known.
A plot of the insertion loss vs. frequency (via two-port S21 measurement), would also give us information whether the crosstalk is capacitive or inductive in nature, because the impedance of acoustic and capacitive crosstalk decreases with frequency, and the impedance of inductive crosstalk increases with frequency.
Quote from: Smudge on 2020.07.01, 14:46:19
Ideally with no water present and both coils tuned to the NMR frequency we want zero coupling.
That would be ideal, indeed.
Quote from: Smudge on 2020.07.01, 14:46:19
With the toroidal coil's field at right angles to the pancake coil's field there should be zero magnetic coupling, so it is probably down to capacitive coupling.
No. You forgot that Itsu's toroidal winding has only one layer ...and that means that this coil has non-negligible sensitivity to the circumferential currents present in the pancake coils.
Another way to look at it is noticing that the wires in the pancake coil are not perpendicular to the wires in the toroidal coil. The latter has a winding pitch, which results in 1 turn circumferential component.
Of course, this can be cancelled with properly connected bifilar coil or even-layered coil.
Quote from: Smudge on 2020.07.01, 14:46:19
Although you correctly have the pancake coils in series opposing there are two ways this can be set up, either the outer connection of one pancake is connected to the inner connection of the other, or the connection is at both outers or both inners (the opposing field determined by which way round the pancakes are placed).
Winding pancake coils in opposite directions constitutes another option. This can also avoid the plexi coil-formers being on different sides of the coils.
Also, an effort should be made to cancel their collective radial current, which arises as the electric charges move from the outer to the inner radius of the pancake coil (or vice versa).
Quote from: Smudge on 2020.07.01, 14:46:19
With an unbalanced drive from the signal source I can't see any way to eliminate capacitive coupling for either configuration. Ideally it requires a balanced drive, then one configuration is better than the other. If the drive is at both outers while the series connection is at both inners (or vice versa), then the capacitive coupling is minimised. So I am sorry to say that you need a wide bandwidth balun, something you have already played with on another thread.
I agree that balanced drive and Rx will help a little.
Quote from: Smudge on 2020.07.01, 14:46:19
One you achieve minimum unwanted coupling, then you can add water and look for the NMR signal which by its gyratory action couples the 90 degree input and output fields together.
The unique property of this signal is that it persists after the stimulus from the pancake coils ceases, so temporal separation is another method to minimize the crosstalk.
P.S.
At the risk of sounding nauseating. Litz wire provides clear benefits besides presenting lower AC impedance at RF.
For example, I just noticed that an
unlooped piece of 1mm solid copper wire gets too hot to touch in the proximity of a 75W 100kHz inductive heater (w/ solenoidal coil).
Notice, that this is a straight piece of wire, which is
not connected in a loop, so circular eddy currents in it, can have the maximum diameter of 1mm.
Yet, significant energy transfer* happens anyway. It happens
regardless of this piece of this wire being oriented perpendicularly to the heater's coil or parallel to it. It also does not matter whether this piece of wire is
not inside the heater's solenoid - 5mm outside of heater's coil is enough (with perpendicular and parallel wire orientations, too).
An unlooped piece of a Litz wire (of the same diameter) does not experience any noticeable heating in any orientation and position.
* I'll let Smudge calculate whether eddy currents in this straight unlooped solid wire, which can raise its temperature by +25ºC, can affect the conductivity along this wire or induce electric potentials along this wire ...and how much quantitatively.
Quote from: Itsu on 2020.07.01, 16:14:02
Ok, but when i flip over one pancake coil, the 4mm thick plexiglas will be inbetween the toroidal coil
and this flipped pancake coil causing some unbalance, see light brown substrate in below picture.
Guess i need to rewind 1 pancake coil opposite direction as the other one.
Itsu
Or add another layer of plexiglas. Do the same to the other coil, then the magnet field and NMR frequency will be even lower.
Smudge
Quote from: Smudge on 2020.07.01, 18:10:25
Or add another layer of plexiglas. Do the same to the other coil, then the magnet field and NMR frequency will be even lower.
...and the fringing of the static magnetic field - higher.
Also, doesn't the S/N ratio of the NMR signal decrease with the magnetic flux density of the static field and the frequency?
Quote from: verpies on 2020.07.01, 16:46:18
No. You forgot that Itsu's toroidal winding has only one layer ...and that means that this coil has non-negligible sensitivity to the circumferential currents present in the pancake coils.
Yes I agree, but that can be minimised by returning the end of the toroidal coil via a single "turn" following the water ring.
QuoteAlso, an effort should be made to cancel their collective radial current, which arises as the electric charges move from the outer to the inner radius of the pancake coil (or vice versa).
I agree and if the radial currents in both coils are in the same direction then this should be achieved.
Quote
P.S.
At the risk of sounding nauseating. Litz wire provides clear benefits besides presenting lower AC impedance at RF.
For example, I just noticed that an unlooped piece of 1mm solid copper wire gets too hot to touch in the proximity of a 75W 100kHz inductive heater (w/ solenoidal coil).
Notice, that this is a straight piece of wire, which is not connected in a loop, so circular eddy currents in it, can have the maximum diameter of 1mm.
Yet, significant energy transfer* happens anyway. It happens regardless whether the piece of this wire is oriented perpendicularly to the heater's coil or parallel to it. It also does not matter whether this piece of wire is not inside the heater's solenoid - 5mm outside of heater's coil is enough (with perpendicular and parallel wire orientations, too).
An unlooped piece of a Litz wire (of the same diameter) does not experience any noticeable heating in any orientation and position.
* I'll let Smudge calculate whether eddy currents in this straight unlooped solid wire, which can raise its temperature by +25ºC, can affect the conductivity along this wire or induce electric potentials along this wire ...and how much quantitatively.
I will look into that.
Smudge
For the time being i have:
# flipped over 1 pancake coil, so both coils "face" the same direction, not mirrored.
# added a piece of 4mm thick plexiglass between the not flipped pancake coil and the toroidal coil.
# added a piece of 4mm thick plexiglass in front of the flipped over pancake coil.
# connected the pancake coils so that its driven to both outside connections.
This means that the both pancake coils are 4mm extra away from the toroidal coil and there is an extra 4mm between the pancake coils and the magnets.
The total distance between the magnets is now 33mm.
Tuning input for about 3.8Mhz resonance and matching the output to that shows the following signals:
Yellow is input pancake coils
red is differential probing output signal.
So we still have a considerate amount of crosstalk between input and output.
Itsu
Quote from: Itsu on 2020.07.01, 20:41:12
Yellow is input pancake coils
Is that the current flowing through the pancake coils ?
No yellow is the voltage across the pancake coils connection.
Below i have added in green the current through the pancakes.
Red still the differentiale probed output voltage signal
Itsu
Quote from: Itsu on 2020.07.02, 09:06:41
Below i have added in green the current through the pancakes.
Is it really the current flowing through the pancakes only ...or the current flowing to the pancakes
and caps from the AC source ?
Sorry for pestering...
Quote from: Itsu on 2020.07.02, 09:06:41
Red still the differential probed output voltage signal
Is the voltage output from the toroidal coil tweaked & peaked with the var.cap (trimmer) each time the test frequency changes ?
Quote from: Smudge on 2020.07.01, 14:46:19
...I would not necessarily deliberately load the output coil, I would go for a high load resistance if any.
I know that this advice was only for signal search but when trying to determine the I/O crosstalk ratio, I really don't like measuring the output voltage amplitude in the
unloaded LC circuit of the toroidal coil, because the
resonant rise can unpredictably inflate the signal amplitudes there and make it difficult to compare them at several test frequencies (after "peaking").
I think it is really important to determine the crosstalk vs. frequency relationship well, because such measurement allows us to determine whether this crosstalk is predominantly capacitive in nature ...or inductive. I really hope it is not acoustic via Ampere's forces.
Depending on the results, we can deal with it differently. For example capacitive interaction can be shielded with grounded Litz cloth, which does not affect HF magnetic fields (yes, capacitive coupling also falls off with the k of the dielectric and with distance, but distance has other detrimental effects on the entire system).
Undesirable inductive coupling can be managed by controlling the field geometries more precisely (e.g. cancelling longitudinal currents and generally "squaring" the fields).
For acoustic coupling, preventing the coils from touching each other or transferring ultrasonic vibrations through solid objects its usually enough (copper to air to copper ultrasound transmission is very inefficient)
Trying to answer, but i get:
Forbidden
You don't have permission to access this resource.
Additionally, a 403 Forbidden error was encountered while trying to use an ErrorDocument to handle the request.
Itsu
Send through the proxy......
QuoteIs it really the current flowing through the pancakes only ...or the current flowing to the pancakes and caps from the AC source ?
Sorry for pestering...
No problem, it is the current through the pancake coils at position shown in the diagram below.
QuoteIs the voltage output from the toroidal coil tweaked & peaked with the var.cap (trimmer) each time the test frequency changes ?
Yes, but its very touchy and i can only set it up correctly using the FM sweep setup where i trim the red output signal ontop of the input signal with the output trimmer cap.
But during the switch over to normal sine wave signal display i loose that ability to trim that output cap correctly ontop, so am kind of blind.
I also noticed a 10Vpp yellow signal loss between setups (see above 2 last screenshots, 1 = 28Vpp other 18Vpp yellow signal).
Seems that when setting up the FG trigger port (ground only) to the scopes green probe port it gains 10Vpp.
When removing that FG trigger port connection for doing the normal signal / current probe setup it looses that 10Vpp again.
Means to me we have some serious groundloops with severe impact.
Itsu
Quote from: Itsu on 2020.07.02, 11:12:40
Send through the proxy......
Gripe to Peterae ! He will need your IP number to investigate this (error 403 is different than the 406 error with the percent character).
Quote from: Itsu on 2020.07.02, 11:12:40
No problem, it is the current through the pancake coils at position shown in the diagram below.
OK, this signal is very immune to interference so you can rely on it. If it was not for the current flowing through the intrawinding capacitance, then this signal would reflect the magnetic field generated by the pancake coils precisely.
Quote from: Itsu on 2020.07.02, 11:12:40
Yes, but its very touchy and i can only set it up correctly using the FM sweep setup where i trim the red output signal ontop of the input signal with the output trimmer cap.
Touchy means very selective, which is good. Touchy can also mean that the trimmer cap is too big, and its small adjustments cause large its capacitance changes.
Do you have two trimmer caps? ..the big one and a small one, that you can connect in series ?
Quote from: Itsu on 2020.07.02, 11:12:40
I also noticed a 10Vpp yellow signal loss between setups (see above 2 last screenshots, 1 = 28Vpp other 18Vpp yellow signal).
Seems that when setting up the FG trigger port (ground only) to the scopes green probe port it gains 10Vpp.
When removing that FG trigger port connection for doing the normal signal / current probe setup it looses that 10Vpp again.
I had a problem like that once. I used an isolation transformer form an old Ethernet card to feed the trigger signal to the scope (I tried optocouplers too, but they require a PS and are low speed).
Also, I have 1:1 isolation transformers on the 240VAC power supplies of all my gear and I do not use the ground lead on them (unless I want to).
Quote from: Itsu on 2020.07.02, 11:12:40
Means to me we have some serious groundloops with severe impact.
You're not kidding!
QuoteGripe to Peterae ! He will need your IP number to investigate this (error 403 is different than the 406 error with the percent character).
Message send to Peterae
QuoteTouchy means very selective, which is good. Touchy can also mean that the trimmer cap is too big, and its small adjustments cause large its capacitance changes.
Do you have two trimmer caps? ..the big one and a small one, that you can connect in series ?
Yes i have som smaller trimmer caps which i can use in series.
QuoteI had a problem like that once. I used an isolation transformer form an old Ethernet card to feed the trigger signal to the scope (I tried optocouplers too, but they require a PS and are low speed).
Also, I have 1:1 isolation transformers on the 240VAC power supplies of all my gear and I do not use the ground lead on them (unless I want to).
Ok, so i could have had this problem earlier when using the FM sweep setup or when just having that FG trigger port connected (video shot here: https://www.youtube.com/watch?v=P0pTPVgrhFs )
I will try to think of that during these setups.
Itsu
OUR still only accessable via the proxy.
I made a wideband 1:1 balun, what is the best place to put it, between signal source and the caps or between the pancake coils and caps?
Itsu
Quote from: Itsu on 2020.07.02, 17:20:24
OUR still only accessable via the proxy.
Apparently the server "thinks" that you are a dangerous hacker.
Quote from: Itsu on 2020.07.02, 17:20:24
I made a wideband 1:1 balun, what is the best place to put it, between signal source and the caps or between the pancake coils and caps?
If it is a galvanically isolated balun with 1:1 ratio then my answer is "between signal source and the caps".
If the balun ratio was something else than 1:1 then you could be tempted to match the impedance of the pancake coils to the impedance of the signal source transformed by such balun.
QuoteApparently the server "thinks" that you are a dangerous hacker.
Yes, that would be twice this week then.
QuoteIf it is a galvanically isolated balun with 1:1 ratio then my answer is "between signal source and the caps".
If the balun ratio was something else than 1:1 then you could be tempted to match the impedance of the pancake coils to the impedance of the signal source transformed by such balun.
Its this one: https://vk6ysf.com/balun_1-1.htm using a T106-2 with 16 turns.
So not galvanically isolated.
Itsu
Quote from: Itsu on 2020.07.02, 18:36:25
So not galvanically isolated.
So it is not suitable for interrupting ground loops. :(
These galvanically isolated 1:1 baluns are hard to find on the web.
In this case its just needed to balans the input signals into the pancake coils to minimize the crosstalk.
So should we really need to be able to completly avoid any signals on the output without any water in the tube?
I cannot imagine that being possible with coils that close to each other.
Itsu
Quote from: Itsu on 2020.07.02, 19:28:50
These galvanically isolated 1:1 baluns are hard to find on the web.
..but they are easy to make with a small ferrite toroid and non overlapped windings. Also, old Ethernet cards are a cheap source of high bandwidth signal transformers which are galvanically isolated.
Quote from: Itsu on 2020.07.02, 19:28:50
In this case its just needed to balance the input signals into the pancake coils to minimize the crosstalk.
You are welcome to try balancing the sides and see how much that helps. We are here to learn as much as invent, after all.
Quote from: Itsu on 2020.07.02, 19:28:50
So should we really need to be able to completely avoid any signals on the output without any water in the tube?
Not completely but as much as possible. If the S/N ratio of the NMR signal will be too small to observe over the crosstalk, then there is always TDM to the rescue.
Quote from: Itsu on 2020.07.02, 19:28:50
I cannot imagine that being possible with coils that close to each other.
They should not touch - that's for sure.
Right now you don't even know whether the crosstalk is caused predominantly by inductive or capacitive coupling (or acoustic).
If the coupling is predominantly capacitive, a grounded shield between coils, (made out of Litz cloth so it does not impede AC magnetic fields) will make a big difference.
Quote..but they are easy to make with a small ferrite toroid and non overlapped windings. Also, old Ethernet cards are a cheap source of high bandwidth signal transformers which are galvanically isolated.
You mean these things, see picture? I got some out of a router.
I was doing some tests with TG/SA like you suggested, but allthough the input trimmer cap is working fine to tune the resonance, the output cap did not have any noticable influence so from that perspective it seems there is no visible crosstalk.
By the way, my old Tek scope does not show this 10Vpp gain/lose when connecting the FG trigger port to scope port 4.
Seems the new one is more sensitive in that area.
Itsu
Quote from: Itsu on 2020.07.02, 20:43:03
You mean these things, see picture? I got some out of a router.
Yes, but the ones I found were much bigger. Does yours transfer the synch pulse from the FG, even if it is only the edge ?
Quote from: Itsu on 2020.07.02, 20:43:03
I was doing some tests with TG/SA like you suggested, but allthough the input trimmer cap is working fine to tune the resonance, the output cap did not have any noticable influence so from that perspective it seems there is no visible crosstalk.
That's good news but I'd like to order a hookup schematic and a movie, please ;)
Quote from: Itsu on 2020.07.02, 20:43:03
By the way, my old Tek scope does not show this 10Vpp gain/lose when connecting the FG trigger port to scope port 4.
Hmm, I wonder where the difference comes from. These scopes should have very similar port impedances and capacitances. Maybe their internal grounding is different. It's worth investigating but that's for another time.
Ok, update seen, i will make that TG/SA diagram/ movie tonight
Typing this on my iphone and 4G as now also the proxy does not work
Tried severall but none worked, not even on google etc.
I hope peter can take a look soon
Itsu
Quote from: Itsu on 2020.07.03, 08:24:59
...now also the proxy does not work
I sent you 2 new ones by PM
Thanks, i am back in using the UK proxy, but its slow and unreliabe.
Still locked out (timing out) of OUR.com when using no proxy while all other internet traffic is working ok.
Itsu
Hi Itsu,
On my Magnetic Delay Transformer thread you made a twisted wire balun as reported here
https://www.overunityresearch.com/index.php?topic=3847.msg78156#msg78156 (https://www.overunityresearch.com/index.php?topic=3847.msg78156#msg78156)
This should provide the wanted unbalanced to balanced input for the pancake coils.
Smudge
Thanks Smudge, i still have it somewhere, and it has galvanic isolation O0
I will do some testing with it too.
Itsu
Quote from: verpies on 2020.07.02, 21:48:37
Yes, but the ones I found were much bigger. Does yours transfer the synch pulse from the FG, even if it is only the edge ?
That's good news but I'd like to order a hookup schematic and a movie, please ;)
Hmm, I wonder where the difference comes from. These scopes should have very similar port impedances and capacitances. Maybe their internal grounding is different. It's worth investigating but that's for another time.
I tried 3 different baluns now, that small one, the one i made yesterday and the one made earlier as Smudge mentioned, but all 3 do NOT transfer the synch pulse from the FG :(
Itsu
Can I post a % don't try it yourself yet though
So through cpanel I can turn mod security off, but it says this is very unwise for security purposes.
Quote from: Peterae on 2020.07.03, 20:19:11
Can I post a % don't try it yourself yet though
Yes you can. Did you tame the mod_security ?
Video showing the resonance influence of the both coils (pancakes v toroidal coil) on first the scope when sweeping and then using the TG and SA.
I did a normalisation before the actual test by shorting the TG SA leads using a BNC female-female plug at the end of the cables.
The TG SA does not show any influence of the output trimmer cap on the resonance trace not in the input to output situation, nor in the output to input situation.
Guess the 50 Ohm impedances are causing this.
Diagram on how the setup was done below, video here: https://www.youtube.com/watch?v=_sZsrii2bhY
Itsu
I found a switch in cpanel to turn mod security off, trouble was that when I turned mod security back on and viewed the page my IP got banned, and I think anyone else that viewed it for the few seconds it was on the page also got banned oops, had to quickly change my IP turn mod off , go edit my post to say percentage lol.
Now how dangerous is it for the forum to run with mod security turned off I wonder.
Quote from: Itsu on 2020.07.03, 20:17:31
I tried 3 different baluns now, that small one, the one i made yesterday and the one made earlier as Smudge mentioned, but all 3 do NOT transfer the synch pulse from the FG :(
Even when driven through a small cap to eliminate the DC component ?
Quote from: Peterae on 2020.07.03, 20:28:01
Now how dangerous is it for the forum to run with mod security turned off I wonder.
The are mentions of the impact of this solution on some security forums.
Google for the full text of the 406 error message and add "mod_security".
... the real solution though would be for the SMF code to properly escape the percent character, so the server does not frown upon it.
Quote from: verpies on 2020.07.03, 20:28:19
Even when driven through a small cap to eliminate the DC component ?
Yes, for all 3 baluns, again nothing shows on the scope.
Quote from: Itsu on 2020.07.03, 20:38:33
Yes, for all 3 baluns, again nothing shows on the scope.
Let me try on mine.
Which output jack on the Rigol FG do you use to obtain the signal to trigger the scope? The one the front panel or rear ?
Its the one on the back for CH2
Quote from: Itsu on 2020.07.03, 20:48:34
Its the one on the back for CH2
With my 1:1 telephone balun, the signal from the Trig output on the rear panel of the FG transfers as a 10ms, 3V voltage pulse on the secondary. This pulse coincides with the edge of the 1Hz square wave from the FG's Trig output.
The inductance of its open primary is 200mH. Its core is a toroidal ferrite 20mm O.D. Its windings do not touch (they are 5mm apart).
With the Ethernet balun, the edge of a 1Hz square wave appears as a 1µS and 3V pulse on the secondary, so you cannot see it with a slow timebase*.
P.S.
The Trig output of the FG should be protected with two back to back 5V zener diodes connected in series. These diodes short the FG's Trig output to protect it from balun's flyback voltage when its secondary is not loaded. If the secondary is always loaded then these diodes are not needed. Because of this I keep mine loaded with 1kΩ resistor (and 50Ω resistor with other signal sources)*
Some scopes cannot trigger on short pulses when their timebase is slow and the sample rate is low.
Better scopes, that have dedicated Trig inputs, can do it because they have a hardware edge detector at these inputs ...and do not depend on the sampling subsystem to capture the triggering pulse.
If your scope cannot trigger on short pulses with slow time bases, then you can try this passive pulse stretcher.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35415)
- Setting the scope probe to 10x is important so the probe does not drain the cap too quickly.
- The Schottky diode D1 increases the amplitude of the stretched pulse. (this can be a small-signal Schottky diode)
- The Zener diode D2 terminates the stretched pulse earlier (optional).
...but eventually consider making an active trigger-pulse stretcher for your scope with a flip flop, e.g. made with 74HC02 chip (or a faster one). Its disadvantage is that it needs a 3V-5V power supply on the secondary side, though.
Quote from: verpies on 2020.07.03, 22:59:12
With my 1:1 telephone balun, the signal from the Trig output on the rear panel of the FG transfers as a 10ms, 3V voltage pulse on the secondary. This pulse coincides with the edge of the 1Hz square wave from the FG's Trig output.
The inductance of its open primary is 200mH. Its core is a toroidal ferrite 20mm O.D. Its windings do not touch (they are 5mm apart).
With the Ethernet balun, the edge of a 1Hz square wave appears as a 1µS and 3V pulse on the secondary, so you cannot see it with a slow timebase*.
P.S.
The Trig output of the FG should be protected with two back to back 5V zener diodes connected in series. These diodes short the FG's Trig output to protect it from balun's flyback voltage when its secondary is not loaded. If the secondary is always loaded then these diodes are not needed. Because of this I keep mine loaded with 1kΩ resistor (and 50Ω resistor with other signal sources)
* Some scopes cannot trigger on short pulses when their timebase is slow and the sample rate is low.
Better scopes, that have dedicated Trig inputs, can do it because they have a hardware edge detector at these inputs ...and do not depend on the sampling subsystem to capture the triggering pulse.
Ok, i redid the synch pulse testing and i see its there, but indeed not visible with the normal timebase.
With the small balun i have a 3V 1.6us pulse on the secondary.
The Smudge balun is >3V at 20us even
Itsu
Ok, so what would be the preferred balun for the NMR input setup, the small one or the large Smudge one?
Both galvanically isolate, but does the small one also change from balanced to unbalanced?
Quote from: Itsu on 2020.07.04, 08:53:18
With the small balun i have a 3V 1.6µs pulse on the secondary.
A large inductance transformer is needed only for transferring pulses, which have a looong rise time (or fall time), i.e.: slow analog signals.
With digital signals, the short pulse appearing across the secondary winding of small transformers affects only the output amplitude of the passive pulse stretcher. It does
not affect the output of the active pulse stretcher.
Quote from: Itsu on 2020.07.04, 11:13:11
Ok, so what would be the preferred balun for the NMR input setup, the small one or the large Smudge one?
Both galvanically isolate, but does the small one also change from balanced to unbalanced?
I would go with the small one because it has a smaller interwinding capacitance and the signal frequency is high enough that the low inductance of the small one does not matter. Also, the current is so low now, that there is no chance of saturating the core. If you ever go to higher current levels then a larger core will be needed (but not a larger inductance).
For the isolation of digital signals I would also use the small one but with a pulse stretcher so I can see the signal clearly on the scope when the sweep times are long (I have an ext.Trig input, so I could do without the stretcher).
P.S.
If I were you, I would rewind the coils (with Litz wire) so they are symmetric in themselves and do not need to be concerned with the hot/cold side issue and don't have any unbalanced longitudinal currents ...but that is only what I would do. The author might disagree.
Ok, thanks, i put up my balun made yesterday (https://vk6ysf.com/balun_1-1.htm) which i know is not isolating, but does unbalance/balance on the input side before the caps.
It minimzes the input signal on the pancake coils though.
It went from 24Vpp to now about 10Vpp while i increased the FG signal from 10Vpp to 20Vpp.
Sweeping from 3 to 5Mhz still.
So we lose a lot of signal using the balun.
Concerning your P.S., sounds sound to me, but what do you mean by "so they are symmetric in themselves"?
How can i wind a pancake coil to be symmetric in itselve?
Itsu
Quote from: Itsu on 2020.07.04, 12:04:23
How can i wind a pancake coil to be symmetric in itself?
For example like this:
Hmmm, that is neat, never saw that one before O0
Seems i am unbanned as i have fast access again to OUR without using the proxy.
I have tested the severall baluns i have, and the best one (least signal loss) is the big one Smudge was refering to earlier.
But even this one has about 66% signal loss compared to the no balun situation (12V v 39V).
Also there is no difference percentage wise in input v output signal in resonance.
With balun i have 12V input and 8V output (33%).
Without balun i have 39V input and 25.6V output (33%) see screenshot.
So it seems to me that using a balun is not helping the avoid the crosstalk and only adds losses to the system.
Itsu
Quote from: Itsu on 2020.07.04, 19:17:01
Seems i am unbanned as i have fast access again to OUR without using the proxy.
It is my experience that the ban times-out when there is not activity from the banned IP.
Quote from: Itsu on 2020.07.04, 19:17:01
But even this one has about 66% signal loss compared to the no balun situation (12V v 39V).
This is too much loss for a 1:1 transformer. Perhaps its ratio is not 1:1 and the impedance is getting transformed or the transformer introduces leakage inductance into the primary LC circuit and detunes it.
How does the transmission loss vs. frequency plot appear for this transformer, when you connect the TG's output to the primary and SA's input to the secondary ?
Quote from: Itsu on 2020.07.04, 19:17:01
Also there is no difference percentage wise in input v output signal in resonance.
How do you measure that?
Quote from: Itsu on 2020.07.04, 19:17:01
With balun i have 12V input and 8V output (33%).
Without balun i have 39V input and 25.6V output (33%) see screenshot.
What about the current flowing in the pancake coils? The voltage at the caps is not very representative of the magnetic flux amplitude generated by the coils...
Quote from: Itsu on 2020.07.04, 19:17:01
So it seems to me that using a balun is not helping the avoid the crosstalk and only adds losses to the system.
I am surprised, symmetrizing the input to the primary LC circuit should help a little. Does galvanic isolation make any difference ?
QuoteIt is my experience that the ban times-out when there is not activity from the banned IP.
Yes, could be, but now it toke 2 days, the last time only 12 hours or so.
QuoteThis is too much loss for a 1:1 transformer. Perhaps its ratio is not 1:1 and the impedance is getting transformed or the transformer introduces leakage inductance into the primary LC circuit and detunes it.
How does the transmission loss vs. frequency plot appear for this transformer, when you connect the TG's output to the primary and SA's input to the secondary ?
Its soldered to a little pcb now, so will do that tomorrow.
QuoteHow do you measure that?
By looking at the FM sweep plots like the one in my last post, it shows the 39V in / 25.6V out = 33%
Another one not shown but with the balun shows in 12V / out 8V =33% too
QuoteWhat about the current flowing in the pancake coils? The voltage at the caps is not very representative of the magnetic flux amplitude generated by the coils...
I will do a current pancake coil versus output signal (unloaded) per frequency step tomorrow.
QuoteI am surprised, symmetrizing the input to the primary LC circuit should help a little. Does galvanic isolation make any difference ?
The Smudge balun is galvanically isolated, it has trifilar windings (2 layers) on a big toroid from which only 2 out of the 3 wires are used as isolation transformer, see here:
https://www.overunityresearch.com/index.php?topic=3847.msg78156#msg78156
its the first picture, the isolation xformer is the one on the left.
Itsu
Quote from: Itsu on 2020.07.04, 20:41:39
I will do a current pancake coil versus output signal (unloaded) per frequency step tomorrow.
Your secondary circuit is closed by the capacitance and resistance of the wire, so it is loaded ...albeit not deliberately.
It should be frequency-selective as any LCR circuit is. Did you ever do a FM sweep of the secondary or just observed its frequency response through its reflection back to the primary ?
No i did not sweep the secondary, will do that too.
The current through the pancakes is weak, around 5mA rms and very sensitive to tuning the mostly input trimmer cap.
It also is not "in sync" with the output voltage, so i will go for the max. current each time.
I will try to make a plot between 3 and 5Mhz.
Itsu
Quote from: Itsu on 2020.07.05, 08:46:42
It also is not "in sync" with the output voltage, so i will go for the max. current each time.
In theory, the voltage induced in an open-circuited secondary is proportional to the 1
st derivative of current flowing in the primary (di/dt).
IOW: 90º off, because this is where the sine wave changes the fastest ( is the steepest ).
Ok, i will check on that 90° off situation.
Sweeping the secondary shows a flat response up till 60Mhz.
The 100pF parallel output cap seems to have no influence, also the input cap not.
The below screenshot shows a 1Khz to 10Mhz sweep of the secondary (toroidal) coil.
Itsu
Quote from: Itsu on 2020.07.05, 09:21:07
Sweeping the secondary shows a flat response up till 60Mhz.
The 100pF parallel output cap seems to have no influence, also the input cap not.
This is really weird! Where do you apply and sense the signal during the sweep ?
Please remind me what is the inductance of the secondary.
Yes, i found that weird too.
See picture for connections.
The toroidal coil measured 31uH
Itsu
QuoteQuoteQuote from: Itsu on 2020-07-04, 20:17:01
But even this one has about 66% signal loss compared to the no balun situation (12V v 39V).
This is too much loss for a 1:1 transformer. Perhaps its ratio is not 1:1 and the impedance is getting transformed or the transformer introduces leakage inductance into the primary LC circuit and detunes it.
How does the transmission loss vs. frequency plot appear for this transformer, when you connect the TG's output to the primary and SA's input to the secondary ?
Here first below the TG SA plot of the Smudge balun ALONE, 0 to 50Mhz.
Its fairly flat the first 10Mhz in which we want to use it.
Second plot is when TG's output to the primary and SA's input to the secondary including the Smudge balun (also 0 to 50Mhz).
Itsu
Quote from: Itsu on 2020.07.05, 11:26:05
Here first below the TG SA plot of the Smudge balun ALONE, 0 to 50Mhz. Its fairly flat the first 10Mhz in which we want to use it.
That's to be expected of a transformer which windings are in such intimate contact. It has 53.5% maximum transmission loss of the signal's amplitude up to 13.5MHz if you'd done the calibration apriori.
Why? Because -6.65dB = 46.5% of the amplitude applied to the primary winding.
Let's see how it performs from 0Hz to 500kHz so I can correlate it with the other plot. It is impossible that it has 0db transmission loss at 0Hz.
Quote from: Itsu on 2020.07.05, 11:26:05
Second plot is when TG's output to the primary and SA's input to the secondary including the Smudge balun (also 0 to 50Mhz).
Did you do the SOLT calibration (flattening) with the balun and other interconnects in place before connecting the pancake and toroidal coils ? Why do you have the 10dB attenuation at the SA's input on ?
Anyway, I'd like to see it zoomed in the frequency span between 0Hz and 4MHz. I think there is a -65dB dip to at 400kHz but I cannot see what's before that.
To have a complete picture, I'd also need a plot where the TG is connected to the input caps (the same as before), but the SA's input is connected to a good
* CSR in series with the pancake coils (with the toroidal coil not connected and opened). Sorry to be so demanding - I just care more about the current flowing in the pancake coils
** than the voltage applied to their input caps.
P.S.
Please use long sweep times so the coils have the time to settle. Mine were ringing for milliseconds, so the adjacent frequency points were affected by the coils still ringing from the stimulus at the previous f.point.*
"good" means low resistance and non-inductive. Its resistance should be as small as possible but not so small that the signal level across it approaches the internal noise level of the SA.**
I cannot write the same about the opened secondary side (the toroidal coil). Sensing the induced voltage across it, is representative of the magnetic flux variations it experiences.
QuoteThat's to be expected of a transformer which windings are in such intimate contact. It has 53.5% maximum transmission loss of the signal's amplitude up to 13.5MHz if you'd done the calibration apriori.
Why? Because -6.65dB = 46.5% of the amplitude applied to the primary winding.
Let's see how it performs from 0Hz to 500kHz so I can correlate it with the other plot. It is impossible that it has 0db transmission loss at 0Hz.
Ok, 1st screenshot is the Smudge balun from 0 (in fact from 9Khz = minimum) to 500Khz.
The 10dB attenuation at the SA's input is default, i now have removed it, so now at 0dB.
QuoteDid you do the SOLT calibration (flattening) with the balun and other interconnects in place before connecting the pancake and toroidal coils ? Why do you have the 10dB attenuation at the SA's input on ?
Anyway, I'd like to see it zoomed in the frequency span between 0Hz and 4MHz. I think there is a -65dB dip to at 400kHz but I cannot see what's before that.
I normally always first before any measurement do the "normalisation" procedure using as much connections needed before hooking up the DUT, but i did NOT included the balun in this normalisation process, see 2th screenshot.
0 (9Khz) to 4Mhz, SA input 0dB.
I will redo this one including the balun in the normalisation procedure next.
The csr plot for the whole picture comes then......
Itsu
This plot is with the Smudge balun included in the normalisation (flattened), 0 (9Khz) to 4Mhz and 0dB SA input.
Itsu
This is a plot with the TG connected to the input caps (the same as before), but the SA's input is connected to a good* CSR in series with the pancake coils (with the toroidal coil not connected and opened).
csr is 10 Ohm 1% inductionfree in the pancake coils return lead.
SA connected across this csr, TG SA grounds together.
0 (9Khz) to 10Mhz, not shure how the afjust the sweeping times (RBW?) EDIT increased the sweep time from 11ms in the first screenshot to 10s in the 2th
Itsu
Quote from: Itsu on 2020.07.05, 16:29:32
0 (9Khz) to 10Mhz, not shure how the afjust the sweeping times (RBW?)
Yes, "RBW" or "TrigSetup/Points". Some SAs also have an option "Delay Between Points".
Quote from: Itsu on 2020.07.05, 16:29:32
This is a plot with the TG connected to the input caps (the same as before), but the SA's input is connected to a good* CSR in series with the pancake coils (with the toroidal coil not connected and opened).
...
EDIT increased the sweep time from 11ms in the first screenshot to 10s in the 2th
OK, this should illustrate to you that despite the TG injecting a CONSTANT VOLTAGE AMPLITUDE signal into the primary circuit, the current amplitude in the pancake coils is NOT constant.
Now, this function of pancake current amplitude vs. input voltage amplitude, must be used to modify the voltage Transmission Loss measurement, which you've already done. Can your SA export it in numeric form (e.g. csv) ...or do I have to count pixels ?
All of this is illustrated on the diagram below:
Our goal is to obtain the frequency characteristics of the Crosstalk Factor (red) in order to determine what is causing it (electric coupling, magnetic coupling or both...and how much of each).
Voltage does not create the magnetic field of the pancake coils - the current does.
But we can measure only the voltage Transmission Loss (blue), because our Signal Source (TG or FG) is a voltage source and it is located before the Input Matching Circuit, which has its own frequency characteristic.
Note, that I am disregarding the frequency characteristic of the output circuit at the toroidal winding - it is not even present on the diagram.
QuoteOK, this should illustrate to you that despite the TG injecting a CONSTANT VOLTAGE AMPLITUDE signal into the primary circuit, the current amplitude in the pancake coils is NOT constant.
Well, it illustrates to me that we have resonance in the pancake coils around 4.1Mhz now viewed through that peak (being current).
Do you say that the current trace should be flat as it is now NOT flat (NOT constant)
Anyway, here attached a CSV file from that trace. (its not zipped, just remove the .zip and there you have the csv file)
Itsu
Quote from: Itsu on 2020.07.05, 18:47:21
Do you say that the current trace should be flat as it is now NOT flat (NOT constant)
No, no, the pancake current amplitude of input voltage amplitude vs. frequency should not be flat in the input matching circuit. It just means that the voltage induced across the secondary toroidal coil must be scaled down by the inverse of Pancake.i (Input.v)
Quote from: Itsu on 2020.07.05, 18:47:21
Anyway, here attached a CSV file from that trace. (its not zipped, just remove the .zip and there you have the csv file)
Got it, but any scaling process involves at least two inputs...so I also need the "blue" measurement in the same frequency span and RBW as "fuchsia" measurement, which I just received from you in 1.csv.zip.
I used the fuchsia color to mark the pancake.i(input.v) measurement in the diagram below.
OK, here the "blue" measurement in the same frequency span and RBW:
QuoteOur goal is to obtain the frequency characteristics of the Crosstalk Factor (red) in order to determine what is causing it (electric coupling, magnetic coupling or both...and how much of each).
Ok, so by knowing the blue and fuchsia color data you can calculate the red data.
Itsu
Quote from: Itsu on 2020.07.05, 19:50:18
...in the same frequency span and RBW:
2.csv.zip
Nope, this one is with a different span and RBW. e.g. it starts from 0Hz and the fuchsia one started from 50kHz. I could interpolate to fit them, but why introduce a numerical error...
QuoteNope, this one is with a different span and RBW. e.g. it starts from 0Hz and the fuchsia one started from 50kHz. I could interpolate to fit them, but why introduce a numerical error...
Hmmm, then that first one i mist that it started at 50Khz.
Anyway, here again the blue trace starting from 50Khz now:
Itsu
So far the highest crosstalk seems to be at 7.46MHz and EM...or the toroidal secondary circuit is just the most sensitive at this frequency because it resonates then. So please check if the trimmer at the secondary circuit affects this 7.46MHz peak.
However, it looks like between measurements, something was disturbed and the pancake primary resonance frequency has shifted from 3.8MHz to 4.1MHz, so lets repeat these "blue" and "fuchsia" measurements tomorrow according to this diagram (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35486) and including the balun, but calibrate/normalize with no balun nor toroidal nor pancake coils for both measurements.
Let's try the linear display of the amplitude on the SA this time (not logarithmic). I am hoping for more precision this way and that the red lower peak disappears completely after scaling.
Ok, i noticed that when attaching the SA to the 10 Ohm resistor it influenced the resonance peak and went from 3.8Mhz to 4.1Mhz.
Could be the extra capacitance of the SA connection caused this.
Will check tomorrow.
Thanks, itsu
Quote from: Itsu on 2020.07.05, 23:13:40
Ok, i noticed that when attaching the SA to the 10 Ohm resistor it influenced the resonance peak and went from 3.8Mhz to 4.1Mhz.
Could be the extra capacitance of the SA connection caused this.
I was afraid of that.
- Maybe that balun will help* , or...
- a lower resistance CSR, or...
- changing the side of the CSR where the SA's ground lead/sheield is connected, or...
- adding a high resistance in the ground lead/shield going to the SA's input, or...
- the contactless current probe connected to SA's input
Anyway, leave the CSR in for both measurements unless you decide to use the contactless current probe.
*
I have doubts about this particular balun and you will have them too, once you measure the capacitance between its primary and secondary winding.
OK, my thoughts were to use your last option, the contactless current probe connected to SA's input.
I will try that first.
Concerning your doubts about the balun, the capacitance between prim/sec is shown in the below picture.
Itsu
That is almost 2nF between the primary and secondary.
At 4MHz that capacitance presents a 20Ω path for the signal !
Is there a way you can push these windings apart, so there is a 5mm gap between the primary and secondary ?
Right, no its 2 layers, so i have to remove many windings to get that.
The contactless current probe option does not work as its connected to the scope, not the SA.
Using a 1 Ohm csr also does not work, still there is a difference of 3.8 to 4.1 Mhz, even when i leave the connection in place during both measurements.
The peak is less with 1 Ohm.
I have a smaller balun like the Smudge balun, also trifilar, but with 1 layer, but inside the toroid the windings touch each other which i can try.
But the capacitance comes from the intertwined trifilar wires, not from the trifilar windings themselve i think, so would spacing them help?
Itsu
Quote from: Itsu on 2020.07.06, 09:07:24
The contactless current probe option does not work as its connected to the scope, not the SA.
I've seen in some of your videos, that you have a separate amplifier for the current probe. Can't the output of this amplifier be connected to SA's input ? Of course, the amplitude of the signal coming out of the amplifier cannot exceed the max input level of the SA.
Quote from: Itsu on 2020.07.06, 09:07:24
But the capacitance comes from the intertwined trifilar wires, not from the trifilar windings themselve i think, so would spacing them help?
I am not tracking well the winding anatomies of these transformers, but if the primary and secondary windings are intertwined then their interwinding capacitance will remain high.
I also have a P6021 AC current probe which i have hooked up to the SA port and it shows a trace with a peak still at 4.16Mhz.
The question is how to measure with this probe the whole setup?
Just attach a single loop wire to the output / load connector and measure the current there?
Itsu
The contactless AC current probe is useful only for the fuchsia measurement.
For the blue measurement just use the regular voltage-sensing connections.
..but I have a feeling I do not understand the question :(
Because i still have the frequency difference (3.8 - 4.1Mhz) between using the current probe (4.1Mhz peak) and the regular voltage-sensing connections (3.8Mhz peak),
i want to use the same setup between the 2 measurements (fuchsia and blue), so i want to use the current probe also for the blue measurement.
But that means i have to measure the current somewhere across the whole setup, or is that not possible?
EDIT:
Concerning the frequency shift, i think the 3.8 to 4.1 Mhz frequency shift is not caused by the
csr / current measurements, but its the other way around, the "regular voltage-sensing connections"
are causing the frequency to shift from its normal 4.1Mhz to 3.8Mhz.
Quote from: Itsu on 2020.07.06, 10:48:39
Because i still have the frequency difference (3.8 - 4.1Mhz) between using the current probe (4.1Mhz peak) and the regular voltage-sensing connections (3.8Mhz peak),
Do you experience this frequency difference between the fuchsia measurement with the voltage-sensing connection on the CSR vs. fuchsia measurement with the P6021 current probe ?
Also, did you calibrate/normalize the frequency response of the P6021 current probe by shorting the TG and puting the P6021 current probe on that short ?
Quote from: Itsu on 2020.07.06, 10:48:39
i want to use the same setup between the 2 measurements (fuchsia and blue), so i want to use the current probe also for the blue measurement.
You could just short the secondary toroidal coil with a wire (or with a trimmer) cap and put the P6021 current probe on that short. I think this is what your photo shows but I am not sure where the trimmer is in that secondary circuit.
However, it is not necessary to use the same setup for the fuchsia and blue measurements, because the fuchsia measurement is just a
relative one that tells us how much the current amplitude in the pancake coils gets attenuated vs. frequency, when the TG drives the primary input circuit with a constant voltage amplitude stimulus. For example, at 3MHz the current's amplitude is only 0.8% of the amplitude at 4.16MHz, despite the TG providing the same voltage stimulus.
Shorting the secondary circuit will allow large currents to flow in the toroidal coil and this current will reflect back to the primary pancake coils via the crosstalk flux and Lenz law. Thus you will be able to see the secondary current influencing the primary current.
See my EDIT in above post.
QuoteDo you experience this frequency difference between the fuchsia measurement with the voltage-sensing connection on the CSR vs. fuchsia measurement with the P6021 current probe ?
No, the frequency peak at both those measurements are the same around 4.1Mhz.
QuoteAlso, did you calibrate/normalize the frequency response of the P6021 current probe by shorting the TG and puting the P6021 current probe on that short ?
yes, i always (try) to normalize (flatten) the connections before attaching the DUT also in this case.
QuoteYou could just short the secondary toroidal coil with a wire (or with a trimmer) cap and put the P6021 current probe on that short. I think this is what your photo shows but I am not sure where the trimmer is in that circuit.
The trimmer is parallel between the toroidal coil and the blue shorting loop, so i can remove the short, use the trimmer as short and measure the current in one of the toroidal connection leads.
QuoteHowever, this it is not necessary to use the same setup for the fuchsia and blue measurements, because the fuchsia measurement is just a relative one that tells us how much current flows in the pancake coils for a constant voltage amplitude stimulus from the TG vs. frequency.
But but, we did that yesterday, only the problem was that the peaks are not at the same frequency when doing so (3.8 v 4.1 Mhz).
Quote from: Itsu on 2020.07.06, 10:48:39
EDIT:
Concerning the frequency shift, i think the 3.8 to 4.1 Mhz frequency shift is not caused by the
csr / current measurements, but its the other way around, the "regular voltage-sensing connections"
are causing the frequency to shift from its normal 4.1Mhz to 3.8Mhz.
I think so, too. The current measurements are less invasive with coils.
Ok, using the current probe for both fuchsia as blue, had to increase the TG level from -20dB to 0dB and set the amplitude from 10dB/div to 6dB/div.
Any use?
Itsu
Is the fuchsia measurement made with the secondary toroidal coil opened ?
I am asking because shorting the secondary circuit allows large currents to flow in the toroidal coil and this current will reflect back to the primary pancake coils via the crosstalk flux and Lenz law. Thus you can see the secondary current influencing the primary current.
Also, I just noticed that your vertical units are in dBm ...as in Decibel Milliwatts or Decibel Millivolts ?
Yes, the fuchsia measurement are made with the secondary toroidal coil opened, see above diagram.
The vertical units are in dBm as in Decibel Milliwatts.
Decibel Millivolts would be dBmV with Rigol, you want that data?
Grounding the toroidal coil like doing in the "regular voltage-sensing connections" setup makes the peaks shift from 4.1 to 3.8Mhz.
So i could try to use the csr measurement again, making sure the toroidal coil is grounded during both measurements.
Itsu
Quote from: Itsu on 2020.07.06, 12:26:00
Yes, the fuchsia measurement are made with the secondary toroidal coil opened, see above diagram.
...but in
this diagram (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35503), the toroidal coil is shorted through a trimmer cap
Quote from: Itsu on 2020.07.06, 12:26:00
The vertical units are in dBm as in Decibel Milliwatts.
Decibel Millivolts would be dBmV with Rigol, you want that data?
Yes, because with dBm the SA could be using a power detector internally and we need an envelope detector (amplitude detector as in AM radio). There is a square relation between these detectors.
Quote from: Itsu on 2020.07.06, 12:26:00
Grounding the toroidal coil like doing in the "regular voltage-sensing connections" setup makes the peaks shift from 4.1 to 3.8Mhz.
Most likely because of the ADDED capacitance between the DUT and the lab environment.
Quote from: Itsu on 2020.07.06, 12:26:00
So i could try to use the csr measurement again, making sure the toroidal coil is grounded during both measurements.
Not yet
Quote...but in this diagram above the toroidal coil is shorted through a trimmer cap
Yes, as you mentioned this above in post #186:
QuoteYou could just short the secondary toroidal coil with a wire (or with a trimmer) cap and put the P6021 current probe on that short.
You want no trimmer cap too?
QuoteYes, because with dBm the SA could be using a power detector and we need an envelope detector (amplitude detector as in AM radio).
Ok, making some new measurements without the secondary shorted by anything and in dBmV
Below some measurements in dBmV, but still with the trimmer cap as short on the secondary.
Quote from: Itsu on 2020.07.06, 12:42:21
Yes, as you mentioned this above in post #186:
QuoteYou could just short the secondary toroidal coil with a wire (or with a trimmer) cap and put the P6021 current probe on that short.
That was in reference to the blue measurement
only.
Quote from: Itsu on 2020.07.06, 12:42:21
You want no trimmer cap too?
Correct. I want the secondary toroidal coil to be completely open when making the fuchsia measurement, so there is as little as possible reflected signal back to the primary pancake coils.
Ok, new measurements, fuchsia with open toroidal coil, blue with trimmer cap short on toroidal coil both in dBmV:
Itsu
OK so scaling the blue measurement by the inverse of fuchsia measurement, yields this plot of Crosstalk factor vs. Frequency attached below.
As you can see, this time the measurements are made well and the influence of LC circuit's frequency response at the primary winding (pancakes) is cancelled completely (i.e. no peaks around 4MHz).
Up to 2.5MHz the crosstalk is much smaller than at higher frequencies, which suggests capacitive coupling.
From 6MHz to 7.5MHz the slight downward slope could be taken for inductive coupling but I think this is caused by the intrawinding capacitance of the pancake coils stealing current from their windings at higher frequencies (IOW: current flowing between the windings, instead through the windings). If it was not for the frequency selectivity of the secondary LC circuit, I would expect this downward trend to continue.
The large peak at 8.5MHz is most likely an artifact of the frequency selectivity of the secondary LC circuit, which "amplifies" the crosstalk. This should be verified by watching the smaller 8.5MHz peak during the same blue measurement in real time while tweaking the trimmer cap connected to the secondary toroidal coil.
IMO the remedy for the undesirable coupling is the E-field symmetrization of the system like Smudge has suggested originally. Also, I would suggest grounded electric shielding*, which does not attenuate HF H-fields strongly. This shielding should be placed between the pancake coils and the toroidal coil (or just wrapped around the toroidal coil). Last, but not least, the H-field symmetrization of all coils and liberal use of Litz wire in their construction, per my earlier suggestions, will also minimize the I/O crosstalk and increase the quality of the signals.
* Such shelding is regularly used even in low-frequency power transformers and shielded loop antennas. See here (http://www.angelfire.com/planet/sft_accommodation/magnetic_b_field_antenna.htm) and here (https://electrical-engineering-portal.com/purpose-of-shielded-isolation-transformer)
Thanks, i am glad the data was OK and you could make this great analysis.
For the large peak at 8.5MHz, i think i remember that the 100pF output trimmer cap had no influence on it, only on the 4Mhz peak, but i will do some tests to verify.
So to go forward, a few things should be addressed first like:
# Perfecting the symmetry of the system (pancake coils v totoidal coil v magnets).
# electric shielding by using grounded Litz cloth between pancakes and toroidal coil.
# the use of Litz wire in ALL 3 coils.
# fixing the system so it can not vibrate ultrasonically.
your earlier suggestions i think are here:
https://www.overunityresearch.com/index.php?topic=3924.msg82795#msg82795
QuoteDepending on the results, we can deal with it differently.
For example capacitive interaction can be shielded with grounded Litz cloth, which does not affect HF magnetic fields (yes, capacitive coupling also falls off with the k of the dielectric and with distance, but distance has other detrimental effects on the entire system).
Undesirable inductive coupling can be managed by controlling the field geometries more precisely (e.g. cancelling longitudinal currents and generally "squaring" the fields).
For acoustic coupling, preventing the coils from touching each other or transferring ultrasonic vibrations through solid objects its usually enough (copper to air to copper ultrasound transmission is very inefficient)
Regards Itsu
Ok, will be ordering some stuff.
Besides the checkout on the 8.5Mhz peak, i want to check why the toroidal coil with parallel 100pF cap produced a flat response during a 0 to 60Mhz sweep.
https://www.overunityresearch.com/index.php?topic=3924.msg82872#msg82872
I would expect to see a dip around 4Mhz as the coil measures 31uH.
Also what would be a good balun for this setup so it gives minimum attenuation and does the unbalans to balans conversion.
Itsu
I made this graph from the 9.csv file from yesterday.
I notice some differences between your graphs, can you explain how you modified that data so it gives similar output as yours?
Thanks, itsu
If you cannot buy Litz cloth easily or cheaply, don't bother waving it by hand - it's just too much work.
You can get away with winding several layers of very thin wire (or cheap frayable Litz wire consisting of many thin enameled wires) on some kind of a flexible hose, that is not a part of the system, but has the diameter >= to O.D. of the toroidal winding's crossection. You can wind it using a motorized drill with a stiff rod inside the hose*, because it will easily have 10 thousand turns. After winding it, bend the hose into the circle of the same average diameter as the toroidal coil ...thus creating another toroidal coil that would cover it. Make sure you cannot see the hose through the wires (IOW: the hose is completely covered with wires after bending) - this consideration determines how many layers you should wind on the hose.
Varnish the winding on the bent hose to fix its shape.
After the varnish cures, cut the winding in half circumferentially, using a knife or "dremel" micro saw, so you obtain two separate concave circles (a top one and a bottom one). Don't fray the edges of the cut, too much.
Remove the hose. It will be easy to separate because the two halves of winding will not hug it closely anymore.
Coat with solder one circular edge of on top winding half, and the circular edge of the bottom winding half. Do not coat the edges deeply, only the very ends of the wires.
Solder the two halves together with a circumferential solder bridge over the RX toroidal coil (insulating kapton tape between them). Do not complete the bridge all the way around - you can't do it anyway because you have the water tube inlet in the way.
In other project you would go only 359º around and leave a small gap, or make a small circumferential overlap (better).
* Here (https://www.overunityresearch.com/index.php?topic=3924.msg83015#msg83015) is a calculation how long it would take to wind the thin wire on a hose using a motorized drill at 300rpm.
Ok, i see what you mean.
So its not enough to just put some fabric between the pancake and the toroidal coil, it has to fully encapsulate the toroidal coil.
There are some shielding woven fabrics out there, but not so much litz cloth (quick look).
What about using the outer shielding of thick coax cable, then follow your modification?
Itsu
Quote from: Itsu on 2020.07.07, 14:38:00
So its not enough to just put some fabric between the pancake and the toroidal coil, it has to fully encapsulate the toroidal coil.
Yes, it is enough if the flat shielding Litz fabric extends well beyond the pancake coils, but encapsulating the toroidal coil is more effective and uses less of this shielding fabric ...and easier to do if you cannot buy this fabric pre-made.
I got mine on Alibaba, from a company that was making all kinds of fabrics...mostly cotton ones for T-shirts ;) They must've just substituted a thin enameled copper wire for the usual cotton yarn in their loom.
Quote from: Itsu on 2020.07.07, 14:38:00
There are some shielding woven fabrics out there, but not so much Litz cloth (quick look).
The fabric must be made out of individually insulated wires because if it is not, then circular eddy currents will flow and oppose the HF magnetic fields. We don't want that!
We want to shield only the E-field, not the H-field.
Quote from: Itsu on 2020.07.07, 14:38:00
What about using the outer shielding of thick coax cable, then follow your modification?
If this particular coax has its shield made with individually insulated strands, then you could use it.
..but the goal of the shield in the common coax is to confine both the E-fields and HF H-fields, so I doubt that it is using a shield that lets the high-frequency H-Field through.
Ok, i got it, individually insulated strands is key.
Itsu
When injecting a sweeping signal (3 turns) in the toroidal coil (pancakes fysically there, but open) plus 100pF parallel trimmer cap, i have a nice single resonance peak varied between 2.6 and 4Mhz.
Using the pancakes as injection point (bucking), i have a similar resonance peak as above (slightly lower), but also a 2th resonance peak around 7 to 7.8Mhz, so shorter range.
EDIT, Using the pancakes as injection point (aiding), i have a similar resonance peak as above, but also a 2th resonance peak but around 8.8 to 9.8Mhz.
Using 1 pancake coil at a time then there is 1 resonance peak again, around 4Mhz.
So the 2th resonance peak seems to come (reflection?) from the both pancake coils when in series.
I did not measure the new reduced pancake coils for inductance, all @ 100Khz:
PC1: 22.1uH
PC2: 22.3uH
series bucking: 30.6uH
series aiding: 57.7uH
Itsu
Quote from: Itsu on 2020.07.07, 15:26:05
When injecting a sweeping signal (3 turns) in the toroidal coil (pancakes physically there, but open)...
Tell me more about the orientation of these 3 turns with respect to the toroidal coil.
Quote from: Itsu on 2020.07.07, 15:26:05
So the 2th resonance peak seems to come (reflection?) from the both pancake coils when in series.
What about these 2 pancake coils having slightly different self-resonance frequencies and them beating again each other ?
QuoteTell me more about the orientation of these 3 turns with respect to the toroidal coil.
3 turn loop directly over the toroidal coil as second layer, so max coupling.
QuoteWhat about these 2 pancake coils having slightly different self-resonance frequencies and them beating again each other ?
yes, ( (f1+f2) and (f1-f2) ) i was thinking about that and it probably has something to do with it, but then the 2th range ( (f1+f2) ) would not be shorter.
Itsu
Quote from: Itsu on 2020.07.07, 11:02:24
I made this graph from the 9.csv file from yesterday.
I notice some differences between your graphs, can you explain how you modified that data so it gives similar output as yours?
Basically I normalize the fuchsia data to the max peak, assuming that this peak is as close to the zero impedance as this gets. Anything below this peak I treat as an attenuation of the primary LC circuit.
Next, I divide the blue measurement by the primary attenuation just computed.
The above applies to linear magnitudes.
To do the same with logarithmic magnitudes (dBmV), I normalize the fuchsia peak to 0dB and subtract it from the blue instead of dividing by it, because subtracting logarithmic magnitudes is equivalent to dividing linear magnitudes. See
here (https://www.intmath.com/exponential-logarithmic-functions/3-logarithm-laws.php).
Also, note that the magnitudes in the red plot, which I have posted, were linear.
There is a way to do all this without the csv math by using some fake calibration of the SA.
To do that, just normalize/calibrate it using the fuchsia setup and then do the blue measurement with this fake calibration*. This way, the frequency response of the primary LC circuit will be automatically cancelled out by the SA's correction/calibration algorithm and the result will be the red measurement.
I preferred to calculate the csv data by hand to see what was happening and to have the current probe's frequency response calibrated out, too.
A much better way would be to put a 10MHz Hall magnetic sensor between the pancake and toroidal coil, because the data from such sensor (when properly oriented) would not include the effects of current flowing between the turns of the pancake coil (through the interwinding capacitance) but such HF magnetic sensors are very hard to obtain. The usual ones top out at 1MHz.
*
This assumes that the current probe has a constant amplitude vs. frequency response.
Quote from: Itsu on 2020.07.07, 19:20:10
yes, ( (f1+f2) and (f1-f2) ) i was thinking about that and it probably has something to do with it, but then the 2th range ( (f1+f2) ) would not be shorter.
The resonances we are measuring most likely can be explained by loop antenna theory.
This article (https://pdfs.semanticscholar.org/4736/4a3b4a94109482271f3ff3d249d780acac23.pdf) is a helpful refresher. The spacing between the pancake coils would be responsible for the higher frequency peak we are observing.
Anyway, I just remembered that you are an old Ham and you should have some ham receivers sitting around.
I wish it would be be possible to experiment with the half-dipole receiving antenna (for sensing the E-field) and full loop antennas (for sensing the H-field) in the near & far field, when the pancake coil acts as the transmitter. This would allow you to easily evaluate E/M ratio and the effectiveness of the E-Field shielding.
Unfortunately the antennas needed for this are not practical since the wavelengths we are dealing with are on the order of 100m. Maybe you could stretch two 25m wires needed for the half-wave dipole into your friendly neighbor's garden ...but building a 30m diameter full-wavelength loop, would be just too much.
Thanks for the csv data explanation, i could try that "fake calibration of the SA" as it would give me a quick look for improvement or not when doing changes.
Ok about the 2th resonance point, so its more going into loop antenna theory direction (nice article).
I could experiment with the distance between the 2 pancakes to see how they react.
When i moved to my present house i knew i would not have the possibility to erect any antennas, so i kind of abandoned the HAM radio stuff.
Both the lack of space for antennas and the annoyance one causes by RFI at your neighbours are some reasons, so that path is closed.
Itsu
Looking for litz cloth or litz fabric or woven non-conductive shielding etc. etc. but the only things i find on the web is conductive stuff (so non litz i presume).
This makes it hard to easily shield the toroidal coil from the pancake coils and the only way forwards would be your way of "winding a several layers of very thin Litz wire on some kind of a flexible hose" etc.
Together with the rewinding of both the pancake coils and the toroidal coil in this symmetrical arrangement using litz wire makes this project one not for the faint hearted.
Itsu
Quote from: Itsu on 2020.07.08, 08:52:05
Looking for litz cloth or litz fabric or woven non-conductive shielding etc. etc. but the only things i find on the web is conductive stuff (so non litz i presume).
I got mine on Alibaba, from a chinese company that was making all kinds of fabrics...mostly cotton ones for T-shirts ;) It was advertised as non-tarnishing copper fabric. Initially I thought that they just painted it, but no - they used enameled wire so it would not change the color AND be flexible. They must've just substituted a thin enameled copper wire for the usual cotton yarn in their loom.
i tried that "fake calibration of the SA" procedure and got the below result (NOT using the current probe).
Itsu
"fake calibration of the SA" procedure using the current probe (TG output cranked up from -20 to 0dBm or from 26.99 to 46.99dBmV)
Quote from: Itsu on 2020.07.08, 14:17:20
"fake calibration of the SA" procedure using the current probe (TG output cranked up from -20 to 0dBm or from 26.99 to 46.99dBmV)
Below is the data from the 11.csv and 12.csv files presented in linear form (not dBmV).
You can see that the crosstalk increases with frequency, so the E-field shielding should attenuate it dramatically.
The peak #1 in 11.csv is caused by incomplete cancellation of the primary LC circuit's frequency response, which is the result of the shift in its resonance frequency. This shift is due to this type of probing, which disturbs the primary resonant circuit during the calibration.
The 12.csv includes the current probe frequency response error. I would be curious to see this error measured separately by:
- calibrating when the TG is shorted by the CSR and normalizing with the voltage signal appearing across this CSR
- next, measuring the signal from the current probe placed over this CSR
(this is only for my curiosity and not really needed for the project).
Ok, good to know the E-field shielding is needed.
Concerning the current probe frequency response error, when using the current probe i do not use the csr.
So i don't understand the calibration process to tackle that error.
You want me to use different leads on the SA input side, first the normalizing with the voltage signal appearing across this CSR then the current probe attached and measuring the current through the csr?
Quote from: Itsu on 2020.07.08, 15:55:05
...first the normalizing with the voltage signal appearing across this CSR then the current probe attached and measuring the current through the csr?
Yes, putting the current probe ON the CSR (or its leg) makes the measurement more comparable to the calibration, because in both situations the TG will have to source the same amount of current.
If you short the TG only for the measurement with the current probe, then the TG will have to source a little more current, which is less fair and less comparable.
P.S.
I did not write that I want you to do it - I wrote that I would be curious to see it. It just would be nice...
I gladly will do any test, no problem, i can only learn from it O0
I am not sure i understand the test protocol.
Below i made a test using the csr (10 Ohm) on the TG and measuring with an equal length of coax cable on the SA side the voltage across that csr and then normalizing (flatten) its trace.
Then removing that equal length of coax on the SA, replace it with the current probe in one of the csr legs and logging the response.
It turns out to be flat across the 10Mhz.
Itsu
Quote from: Itsu on 2020.07.08, 18:42:52
I gladly will do any test, no problem, i can only learn from it O0
Thank you, that was educational to me, too. I did not expect the current probe to be so good.
Does it stay flat all the way up to its rated frequency ?
Yes, seems so, this 60Mhz rated AC probe drops off around 70Mhz, see 100Mhz scan.
Concerning this E-field shielding, is it possible to use pieces of litz wire, spread it out vertically
and horizontally across the pancake coil former (so 2 layers) and coat it with varnish to form a shield?
Its about 2cm wide, so to cover 15cm i need 8 pieces x 2 for 2 layers.
For both pancake coils that would mean 32 pieces.
Itsu
Quote from: Itsu on 2020.07.08, 19:34:19
Concerning this E-field shielding, is it possible to use pieces of litz wire, spread it out vertically
and horizontally across the pancake coil former (so 2 layers) and coat it with varnish to form a shield?
If the strands are individually insulated then yes.
...but wouldn't winding these 10 thousand turns on a hose with a motorized drill be easier ?
Yes, these strands come from 2mm diameter thick litz wire from which i have 10m of.
I don't know what would be easier, but both would be hours of fun :o
I am afraid that after bending the hose into shape, i end up with a build up of wires on the inside and minimum coverage at the outside.
Guess its just a matter of trying.
By the way, the water tube has both an inlet and an outlet.
Itsu
Quote from: Itsu on 2020.07.09, 08:40:35
I don't know what would be easier, but both would be hours of fun :o
Just winding it with a motorized drill (or lathe) takes under 15min, ...and it doesn't have to be a very precise winding.
You can estimate the amount of turns needed to cover the outer circumference by dividing it by the thickness of the thin wire.
For example for a 120mm O.D. the outer circumference is 377mm and when you divide that by e.g. 0.1mm diameter of the thin wire, the result is 3770 turns.
At 300rpm the winding time for 1 layer is a little over 12 minutes.
What is the I.D. and O.D of your water tube anyway ?
Quote from: Itsu on 2020.07.09, 08:40:35
I am afraid that after bending the hose into shape, i end up with a build up of wires on the inside and minimum coverage at the outside.
Guess its just a matter of trying.
Yes, the wires will bunch up on the inner circumference, but because
the difference between the inside circumference and outside circumference of your water tube is small, that bunching effect will also be small...and it will not hurt the shield's performance.
This does not need to be a precise job !
Quote from: Itsu on 2020.07.09, 08:40:35
By the way, the water tube has both an inlet and an outlet.
So these spots are excellent candidates for the coverage by your flattened Litz wire idea. Just cut the soldered end short (shallow solder penetration) because any bulk metal allows circular eddy current to develop in HF magnetic field.
QuoteWhat is the I.D. and O.D of your water tube anyway ?
with 1mm thick wire on the toroidal coil around it and some varnish it measures:
I.D. = 85mm
O.D. = 114mm
Itsu
Quote from: Itsu on 2020.07.09, 11:11:06
with 1mm thick wire on the toroidal coil around it and some varnish it measures:
I.D. = 85mm
O.D. = 114mm
So with these diameters you will have 35% more wires on the inner circumference after bending the hose. With 1 layer on the outer circumference, the depth of these bunched up wires on the inner one would be 0.2mm (when using a 0.1mm diameter wire).
Practically, I would make the bent hose's O.D. 118mm and I.D. 81mm, in order to give the shield a 2mm air gap from the toroidal coil and allow space for the overlap of the unconnected edges of the shield and for wire crossings of the toroidal coil, if you rewind it using the special winding technique, which cancels the circumferential current (...and the coil's associated sensitivity in the undesired direction).
Thanks, it all seems so simple when you write it down that way O0
The problem with the special winding technique for the water tube is that the tube with its toroidal coil is solidly covered with varnish / glue, so probably hard to unwind without damage.
Perhaps i should put up my smily face and ask Conrad for another one C.C
Itsu
Quote from: Itsu on 2020.07.09, 13:26:34
Perhaps i should put up my smily face and ask Conrad for another one C.C
Get two but make the water inlets much smaller. I'll take one off you hands.
I can replicate your setup because I have these 140mm magnets and 112/90mm pancake coils.
OK, PM send, not sure the smaller inlets would be possible easily, but Conrad will tell.
Quote from: Itsu on 2020.07.09, 13:26:34
Perhaps i should put up my smily face and ask Conrad for another one C.C
Itsu
In principle it is no problem to send a new tube, but I am in the middle of a move from one apartment to another and my 3D printer will be one of the last items to be transported. It can well be up to a month till I am ready to 3D-print again.
There are so many other things to do till everything is set up in the new apartment. And the plague does not help at all when trying to get stuff I need to make the new apartment ready. The home improvement stores and garden centers are full of people which makes me hesitant to go shopping for the many little things I need to set up the apartment. At the moment I try to get plants and big pots to make my balconies green.
The move from the old apartment to the new was delayed for four months because of the Coronavirus. The craftsmen had many excuses for not keeping promises.
Greetings, Conrad
Conrad,
no problem, your move is far more of importance now.
There must be more 3D printers around which can be used to print something similar.
We see/hear you again when settled in your new apartment.
Regards Itsu
I put everything together (pancakes, toroidal coil, magnets, double distilled water in the tube, 5W power amplifier etc.) for seeing if there are some unforeseen problems like leakage etc.
No leakage has been seen, so thats OK, and the used power amplifier (100Khz to 60Mhz @ 5W) also behaves OK when driven into a 50 Ohm dummy load on the 4Mhz test frequency with a 1:1 SWR.
But when hooking it up to the input circuit via the 1:1 balun i get not enough power to effectively use the SWR meter, so it seems to me i have a very poor SWR no matter how to tune the 100pF input trimmer cap and the PA regulates back.
I will have to do more tests there to be able to put more power into the system or at least have a good SWR so the PA won't regulate back or get damaged.
Video here: https://www.youtube.com/watch?v=57KryMiFsG8
Itsu
It is not surprising, that the large cap in parallel with the signal source reflects so much energy back and causes a high SWR. Simulate it in PSpice and see...
The goal is to drive a high current through the pancake coils.
In cases when the frequency is low or the coil has a low inductance (or both), it is straightforward to drive high AC current through this coil using a low-impedance high-current amplifier, such as the EL2009.
In such cases, the coil's impedance is low enough whereby it can be driven by an amplifier directly.
However, the impedance of a coil increases with frequency as |Z|=√(R2 + (2πfL)2), so at high-frequency, the coil's impedance is assumed to be very high. See the impedance of my coil here (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35349) as it reaches over 7kΩ.
Thus, a high-voltage driver is needed to drive high current into this coil. These capacitors try to do that by swinging the front-end of your RF amplifier to high voltages. That's why you are measuring almost 200V at these coils' terminals (the current would be more interesting though, because it is the current which generates the magnetic field - the goal).
Anyway, what SWR do you get when the secondary of that 1:1 balun is loaded only with a 50Ω dummy load ?
Below is more info on pri/sec electric shielding in transformers from Linear Circuit Design Handbook. This is about LF transformers, but it is also applicable to HF, except the shield's anatomy changes.
(https://cdn.head-fi.org/a/9990382_thumb.jpg)
Thanks verpies,
Your coil impedance around 4Mhz looks to be about 4800 Ohm, so indeed high compared to the low 50 Ohm of the PA.
According to the author Smudge, this is to be expected as he writes in his PDF (first one in post #1 of this thread) on page 11:
QuoteFigure 9 shows the basic circuit diagram where the input is assumed to be a low impedance signal generator that needs to be matched to the high impedance of the input resonant circuit.
This is simply achieved with the capacitor network shown.
The drive coils will have RF voltage across them that is several hundred volts.
But the PA needs to be matched correctly and the present capacitor network i have (10nF / 100pF) is not doing that.
Might be a good idea to use a simulator to find out what works better.
I will check out today what SWR i have behind the 1:1 balun when loaded with a 50 Ohm dummy.
Thanks for the info on the screening, i will try to download that handbook.
Itsu
Have you all considered driving the pancake coils in series self-resonance. This will give you the maximum available current if the PA can supply it. The input matching would then be done with an additional series inductor of say ~2uH at 4MHz.
Just a thot.
regards,
Pm
Quote from: partzman on 2020.07.10, 13:40:32
Have you all considered driving the pancake coils in series self-resonance.
Yes, of course but I am waiting for Smudge to alter his design if this does not work.
verpies,
QuoteAnyway, what SWR do you get when the secondary of that 1:1 balun is loaded only with a 50Ω dummy load ?
I measured the SWR when the secondary of that 1:1 balun is loaded only with a 50Ω dummy load to be 1:1.2, so a good match.
Partzman,
QuoteHave you all considered driving the pancake coils in series self-resonance. This will give you the maximum available current if the PA can supply it. The input matching would then be done with an additional series inductor of say ~2uH at 4MHz.
thanks for the input, not sure where your extra inductance has to go, could you specify?
Itsu
Quote from: Itsu on 2020.07.10, 19:48:34
I measured the SWR when the secondary of that 1:1 balun is loaded only with a 50Ω dummy load to be 1:1.2, so a good match.
So we cannot blame the balun for this.
On another subject - In a project like this it would be nice to have a high frequency magnetic field sensor.
Can you try if a miniature loop of wire
* inserted into your current probe works as such magnetic sensor ?
*
For best results use that 2mm Litz wire to make that loop (the solder junction should be as short as possible).
Quote from: Itsu on 2020.07.10, 19:48:34
verpies,
I measured the SWR when the secondary of that 1:1 balun is loaded only with a 50Ω dummy load to be 1:1.2, so a good match.
Partzman,
thanks for the input, not sure where your extra inductance has to go, could you specify?
Itsu
Itsu,
Ignoring the present requirement for an isolating balun, the inductor would simply connect between the PA and the series pancake coils and a coupling cap would be used to remove any DC offset. The basic difference is now the tap between the pancake coils is moving symmetrically above an below ground instead of remaining at ~0 volts with the balanced input.
If the balun is still required to eliminate this single ended ac voltage, then the inductor would connect in one of the two balanced lines or could be split into two inductors, one in each input line.
The small inductor would produce a small phase shift and perhaps change the measured SRF of the pancake coils but this could be trimmed out.
Regards,
Pm
Measurments of my precision pancake coil:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns* of 1mm Litz wire (68 strand), 3.2m long, antiradial weave (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35585).
When the break is opened:
Intrawinding capacitance: 174.75pF (@100kHz)
ESR: 218Ω (@100kHz)
When the break is closed:
Inductance: 19.65µH (@100kHz)
DC resistance: 0.22Ω
The linear plots of impedance magnitude |Z| vs. frequency, are attached below:
* because for cancelling the radial current it must be an even number
Notice that this 45MHz peak, which appears when my coil is opened, is very close to a reflection from an unterminated transmission line, which has the length approximately equal to the length of the wire in the coil.
I wonder what the characteristic impedance of that coil is, ...when it is treated as a transmission line.
P.S.
From the lumped LC formula (https://en.wikipedia.org/wiki/LC_circuit#Resonance_effect), the resonance frequency calculated with the capacitance and inductance measured @100kHz by the LCR meter, comes out to 2.7MHz.
Quote from: partzman on 2020.07.10, 22:04:13
Itsu,
Ignoring the present requirement for an isolating balun, the inductor would simply connect between the PA and the series pancake coils and a coupling cap would be used to remove any DC offset. The basic difference is now the tap between the pancake coils is moving symmetrically above an below ground instead of remaining at ~0 volts with the balanced input.
If the balun is still required to eliminate this single ended ac voltage, then the inductor would connect in one of the two balanced lines or could be split into two inductors, one in each input line.
The small inductor would produce a small phase shift and perhaps change the measured SRF of the pancake coils but this could be trimmed out.
Regards,
Pm
PM,
thanks for the specification, the balun used is not an isolating one, its just there to balans (symmetrize) the input to the both pancake coils.
Itsu
verpies,
QuoteOn another subject - In a project like this it would be nice to have a high frequency magnetic field sensor.
Can you try if a miniature loop of wire* inserted into your current probe works as such magnetic sensor ?
* For best results use that 2mm Litz wire to make that loop (the solder junction should be as short as possible).
I can try that, i have 0.94mm litz wire coming next week, so i guess it would be better to use that.
QuoteMeasurments of my precision pancake coil:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns* of 1mm Litz wire (68 strand), 3.2m long, antiradial weave.
Wow, that is a beauty, nicely done O0
I was wondering how to start such a coil, like from in the middle, but just using a split and a bridge lateron is an easy solution.
So the plots show that the impedance around 5Mhz is still high (3198 Ohm) so the need of impedance matching is still there.
Itsu
Quote from: Itsu on 2020.07.11, 08:40:03
I can try that, i have 0.94mm litz wire coming next week, so i guess it would be better to use that.
Why wait?
This is only a test whether it will be sensitive enough.
Quote from: Itsu on 2020.07.11, 08:40:03
I was wondering how to start such a coil, like from in the middle, but just using a split and a bridge lateron is an easy solution.
I did the bridge in the middle because I wanted to access the middle, so I could investigate this coil as a transmission line, measure its self-capacitance, etc.
You can start from the middle if you plan to always have a short there.
Quote from: Itsu on 2020.07.11, 08:40:03
So the plots show that the impedance around 5Mhz is still high (3198 Ohm) so the need of impedance matching is still there.
Yes, the coil's impedance maginitude |Z| crosses 50Ω at 405kHz and 16.75MHz, but the former one is due to the pure inductive reactance and the latter one is due to pure capacitive reactance... and the SWR never goes down below 1000.
QuoteWhy wait?
This is only a test whether it will be sensitive enough.
OK, will try something today.
QuoteI did the bridge in the middle because I wanted to access the middle, so I could investigate this coil as a transmission line, measure its self-capacitance, etc.
You can start from the middle if you plan to always have a short there.
OK, guess i will do that then as i now have a ballpark length (3.2m) to go allthough i think i will aim for some more coverage of the tube by adding 2 or 4 more turns (12 or 14 turns).
QuoteYes, the coil's impedance maginitude |Z| crosses 50Ω at 405kHz and 16.75MHz, but the former one is due to the pure inductive reactance and the latter one is due to pure capacitive reactance... and the SWR never goes down below 1000.
I don't see that (crosses 50Ω at 405kHz and 16.75MHz) in the plot, but i have no experience with this kind of equipment at all.
Seems very nice / handy to have though a VNA, would a HP 8753D or so be any good?
Itsu
Quote from: Itsu on 2020.07.11, 11:53:39
I don't see that (crosses 50Ω at 405kHz and 16.75MHz) in the plot, but i have no experience with this kind of equipment at all.
That's because the vertical scale of
this plot (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35581) is not zoomed-in, but you can extrapolate where the 50Ω horizontal line would be on these plots ...it's in the dips.
If I plot the inverse of the impedance's magnitude Y=1/|Z| (a.k.a. magnitude of Admittance) then it becomes more visible because Admittance is a more natural plot for shorted inductors.
The blue markers, in the plot attached below, show where the Admittance crosses the 20 milliSiemens (mS) line, which is equivalent to 50Ω.
Quote from: Itsu on 2020.07.11, 11:53:39
Seems very nice / handy to have though a VNA, would a HP 8753D or so be any good?
That's expensive. At the frequencies we are experimenting with,
this (https://youtu.be/Sb3q8f0NBZc?t=900) is a good & inexpensive choice.
BTW: He makes more experiments with this VNA in his subsequent videos.
Also, remember that your SA can make these |Z| vs. f plots (...and |Y| vs. f plots) with the "impedance bridge" already.
It cannot make the Smith chart, though, because it does not capture the phase information and thus the imaginary component of the impedance (reactance) is not distinguished.
Quote from: Itsu on 2020.07.10, 08:41:02
But the PA needs to be matched correctly and the present capacitor network i have (10nF / 100pF) is not doing that.
Low SWR is a general indicator of efficient energy transfer from the source to the load* (meaning: "low energy reflection coefficient"), but it is
not an indicator of the maximum current flow in the coil.
Purely reactive components are perfect energy reflectors (by definition they have the reflection coefficient equal = 1 , the highest possible). They absorb the energy from the source and in the next quarter-cycle, they give it all back to the source, ...so the net energy transfer is zero over the entire cycle, but these circulating currents can be high, nonetheless.
Quote from: Itsu on 2020.07.10, 08:41:02
Might be a good idea to use a simulator to find out what works better.
Watch a guy using a sim to match a loop antenna using Smudge's capacitor network.
https://youtu.be/weNwZ0D3txg
*
The pancake coils are not the load. If you put a solid copper sheet next to them, then the induced eddy currents will heat this sheet and present a load to the source. ...the precessing protons will, too.
QuoteOn another subject - In a project like this it would be nice to have a high frequency magnetic field sensor.
Can you try if a miniature loop of wire* inserted into your current probe works as such magnetic sensor ?
* For best results use that 2mm Litz wire to make that loop (the solder junction should be as short as possible).
I made the below shown mini loop of litz wire and measured the magnetic field of the shown coil when fed by the FG with a 3.9Mhz AC sine wave.
Putting the probe close with the loop in the same direction as the coils windings, it picks up some signal, see screenshot.
Without the loop there is almost no signal.
Is this what you mean?
Itsu
Quote from: verpies on 2020.07.11, 14:15:02
That's because the vertical scale of this plot (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35581) is not zoomed-in, but you can extrapolate where the 50Ω horizontal line would be on these plots ...it's in the dips.
If I plot the inverse of the impedance's magnitude Y=1/|Z| (a.k.a. magnitude of Admittance) then it becomes more visible because Admittance is a more natural plot for shorted inductors.
The blue markers, in the plot attached below, show where the Admittance crosses the 20 milliSiemens (mS) line, which is equivalent to 50Ω.
That's expensive. At the frequencies we are experimenting with, this (https://youtu.be/Sb3q8f0NBZc?t=900) is a good & inexpensive choice.
BTW: He makes more experiments with this VNA in his subsequent videos.
Also, remember that your SA can make these |Z| vs. f plots (...and |Y| vs. f plots) with the "impedance bridge" already.
It cannot make the Smith chart, though, because it does not capture the phase information and thus the imaginary component of the impedance (reactance) is not distinguished.
Thanks for the additional info, seems it needs some experience to interpret the results.
That nanoVNA seems interesting, one can start to get the hang of it against very limited costs.
Seems there is already a V2 up to 3Ghz.
Those smith charts look very nice to have onboard.
I will seriously consider to buy one.
Itsu
Quote from: verpies on 2020.07.11, 18:55:16
Low SWR is a general indicator of efficient energy transfer from the source to the load* (meaning: "low energy reflection coefficient"), but it is not an indicator of the maximum current flow in the coil.
Purely reactive components are perfect energy reflectors (by definition they have the reflection coefficient equal = 1 , the highest possible). They absorb the energy from the source and in the next quarter-cycle, they give it all back to the source, ...so the net energy transfer is zero over the entire cycle, but these circulating currents can be high, nonetheless.
Watch a guy using a sim to match a loop antenna using Smudge's capacitor network.
https://youtu.be/weNwZ0D3txg
* The pancake coils are not the load. If you put a solid copper sheet next to them, then the induced eddy currents will heat this sheet and present a load to the source. ...the precessing protons will, too.
OK, so the best way to tune is to measure the current in the pancake coils and tune for max.
I tried something earlier, but with the magnets in place they influence the current probe, so will need to remove them.
That simsmith program looks great, will have to take a closer look at it.
Itsu
Quote from: Itsu on 2020.07.11, 19:05:26
Is this what you mean?
Yes, this is exactly what I had in mind.
If this Litz microloop behaves well below its resonance frequency (which should be pretty high because of its small size) and is linear, then you have a directional sensor of high frequency magnetic field.
This means that your measurements of the current flowing through the pancake coils (by CSR or i-probe) will not be skewed anymore by the current flowing in their intrawinding capacitances.
This way you will obtain a more realistic picture what is going on in these coils.
Quote from: Itsu on 2020.07.11, 19:11:50
I will seriously consider to buy one.
In that case, beware of the many fakes/clones out there and also watch this review of that NanoVNA:
https://youtu.be/fSzY5q-QTpw
Quote from: Itsu on 2020.07.11, 19:22:58
OK, so the best way to tune is to measure the current in the pancake coils and tune for max.
Better yet, measure the amplitude of the magnetic field generated by the pancake coils using your new Litzed i-Probe (no pun intended) with the proper orientation.
This way the current flowing in the pancake's intrawinding capacitance will not deceive you.
The only way you can go wrong with this magnetic method is if:
- the Litz loop is nonlinear (e.g. skin effect)
- the Litz loop is too close to its SRF (but that frequency should be very high for such a small loop)
- the i-Probe's internal DC compensation circuit gets overloaded
Quote from: Itsu on 2020.07.11, 19:22:58
I tried something earlier, but with the magnets in place they influence the current probe, so will need to remove them.
Yeah, the magnets saturate the magnetic core in the probe and the internal DC compensation circuit cannot cancel that level of external magnetic flux density from the magnets.
My i-Probe even has a red LED that lights up when that happens.
Bonus rant below (not necessarily for Itsu):
In most daily Ham experiences the AC electric energy provided by the signal source (FG, SG, oscillator, exciter, RF Amplifier) gets permanently converted to:
- heat
- mechanical kinetic energy
- radio waves.
The heat leaves the system and transfers to the environment. See:
Joule Heating (https://en.wikipedia.org/wiki/Joule_heating) and
Inductive Heating (https://en.wikipedia.org/wiki/Induction_heating) and
Electric/Dielectric Heating (https://en.wikipedia.org/wiki/Dielectric_heating).
Ampere forces between wires and electric forces as well as piezoelectrcs can cause acoustic vibrations of wires and other components.
The radio waves separate from the material radiator (antenna) and move outward away from it at the speed of light*. See:
Radiation Resistance (https://en.wikipedia.org/wiki/Radiation_resistance).
This conversion of energy is necessary for us to be able to define a load and the energy transfer to that load.
An ideal reactive component (capacitor, inductor) does not convert the energy from the source permanently. It merely stores it temporarily and then gives it all back to the source. So, an ideal reactive component cannot be a "load" unless it radiates radio waves into the far-field ( near-field E or H fields do not separate from the "radiator", thus their energy is recoverable and no permanent energy conversion occurs. IOW: The generation of Near Fields by reactive components
does not count as permanent energy conversion ! )
In this entire endeavor (this thread) we are trying to convert the AC electric energy from the source into a
fourth form of energy - the energy of proton's precession**, and to detect this energy.
This is a different form of energy from the three listed above. This difference is an exciting stuff in itself.
*
For Centralflow: ...or matter moves inward away from the photons at the speed of light.**
Of course the proton precession energy eventually gets converted to heat through spin relaxation.
QuoteIn that case, beware of the many fakes/clones out there and also watch this review of that NanoVNA:
https://youtu.be/fSzY5q-QTpw
That nanoVNA-F (10Khz-1.5Ghz) with the bigger screen looks perfect for me considering my age.
Quote from: verpies on 2020.07.11, 19:35:08
Better yet, measure the amplitude of the magnetic field generated by the pancake coils using your new Litzed i-Probe (no pun intended - I hate Apple Inc.) with the proper orientation.
This way the current flowing in the pancake's intrawinding capacitance will not deceive you.
The only way you can go wrong with this magnetic method is if:
- the Litz loop is nonlinear (e.g. skin effect)
- the Litz loop is too close to its SRF (but that frequency should be very high for such a small loop)
- the i-Probe's internal DC compensation circuit gets overloaded
Yeah, the magnets saturate the magnetic core in the probe and the internal DC compensation circuit cannot cancel that level of external magnetic flux density from the magnets.
My i-Probe even has a red LED that lights up when that happens.
OK, i will try that, but lets take one step back.
you said earlier:
QuoteLow SWR is a general indicator of efficient energy transfer from the source to the load* (meaning: "low energy reflection coefficient"), but it is not an indicator of the maximum current flow in the coil.
Surely there must be a way to have the best of both worlds, meaning a good match from the PA to the pancake input circuit (resistive) and max. current in the pancake coils itselve (reactive).
I found a good article covering the feedline coupling too here:
https://www.nonstopsystems.com/radio/frank_radio_antenna_magloop.htm
Thanks for the rant, it contains great info like on Radiation Resistance.
Itsu
Itsu
Don't you have an antenna matching unit for HF, a small ferrite Toroid, pass the wire which connects the two pancake coils through the Toroid and wind 2 turns of wire through the Toroid and connect to the matching unit. You should get full power into the coils and no reflection back to source, just my 2 pence worth.
Regards
Mike
In reply to Verpies rant, may I offer my own rant.
Quote
.........................In this entire endeavor (this thread) we are trying to convert the AC electric energy from the source into a third form of energy - the energy of proton's precession**, and to detect this energy.
I would not use the word "convert". Conversion of one form of energy into another brings in the whole thorny concept of CoE. I would say that in this endeavour we are trying to use the AC electrical energy of the source to
enable energy to be extracted from the perpetual precession motions of the protons. I think this "conversion" aspect comes from the quantum mechanical approach where you define the splitting of energy levels (see image below) which doesn't convey the actual precession motion of the spinning nucleus. I come from the old school where I believe the nucleus does actually spin and precess, that precession motion exists and is perpetual but is so random between atoms that it does not create observable signals. Yes you need to apply energy in order to see a signal, and certainly in the usual pulsed NMR experiments the quantum mechanical approach supplies the answer for the needed energies. But if the precession motions are perpetual and our energy input has merely taken away the randomness, why can't we keep this cohered state going forever and continually extract energy from those precessions? Ramsey's observations some 70 years ago suggests this may be possible. That then begs the question, why in the last 70 years has no one followed this up or found out how to do it? I think this may be due to the classical NMR experimental approach using solenoidal coils where the material sample containing the nuclei under investigation is within that coil. The scientists use "filling factor" to define the volume of the sample in relation to the volume of the solenoid. At 100% filling factor it is assumed that this gives maximum coupling between the sample and the coil. However those of us skilled in the art know that, if we treat the sample as a form of magnetic core, that core will suffer a demagnetization effect. The only way to eliminate such demagnetization is by use of very long thin solenoid-plus-cores, or the use of ring cores with toroidal windings. The former leads to problems in maintaining the uniformity of the crossed static field along the length of the solenoid, and as far as I know no one has attempted the use of the ring core approach. The cylindrical symmetry of the static field from disc magnets offers uniformity of the crossed field throughout the ring sample, and the toroidal coil offers maximum coupling with no demagnetization effects.
QuoteThis is a different form of energy from the two listed above. This difference is an exciting stuff in itself.
Yes, and if we achieve OU we know where the anomalous energy comes from
Quote** Of course the proton precession energy eventually gets converted to heat through spin relaxation.
That is true for pulsed NMR, but note that there are two relaxation times. The thermal relaxation is much longer than the dephasing relaxation. I think that CW operation offers the possibility that, after the initial switch-on absorption-energy has got the precessions cohered, the continued absorption energy rate needed to keep the phasing intact is lower than the radiated power extracted.
Smudge
Quote from: Smudge on 2020.07.12, 10:48:05
I would not use the word "convert". Conversion of one form of energy into another brings in the whole thorny concept of CoE.
I could agree with that when it comes to nuclear dynamic but when doing RF engineering, the notion of permanently converting the AC energy into something else is important, because if that conversion is not done, then the transmitting coil is just going to store this energy for the quarter of the AC cycle and return it all back to the source later.
When this happens, a large energy reflection is perceived by the source ( the "reflection coefficient" (Γ) approaching 1 ...and SWR approaching infinity ). This upsets the RF amplifier and its owner.
Of course, if the H-field generated between the coils does some work, i.e. on the nuclear spins, then the part of the energy used to do that work, does not get reflected back to the source ...and the reflection coefficient (Γ) improves ( as does the SWR, which is just equal to 1+│Γ│ / 1-│Γ│ ).
However, when the nuclei do not absorb the energy coherently because the owner of the RF amplifier is tuning the system off nuclear resonance, he can expect to have a "good match" (low Γ)
only when the wiring's resistance
* dissipates the AC energy as heat or the pancake coils efficiently convert this energy into far-field EM radiation.
Do we really want that ?
Quote from: Smudge on 2020.07.12, 10:48:05
I come from the old school where I believe the nucleus does actually spin and precess, that precession motion exists and is perpetual but is so random between atoms that it does not create observable signals.
I applaud your mechanistic approach to nuclear dynamics. I also think the contemporary approach is too abstract.
Quote from: Smudge on 2020.07.12, 10:48:05
But if the precession motions are perpetual and our energy input has merely taken away the randomness,
Very well
Quote from: Smudge on 2020.07.12, 10:48:05
why can't we keep this cohered state going forever and continually extract energy from those precessions? Ramsey's observations some 70 years ago suggests this may be possible.
Perhaps because nucleus cannot exist without these spins
** and if you cancel them, it will fall apart into simpler particles/motions: proton into muon, muon into electron, electron into photon. But no worry, that is still "energy extraction".
My unorthodox view is, that it is possible to have
motion without an object*** (since motion is merely a relation between space and time [s/t] ). Thus an object/particle is just a spun up space and these particles, I've just mentioned, as well as nuclei, are just mere motions, which can be cancelled by other opposing motions.
Quote from: Smudge on 2020.07.12, 10:48:05
That then begs the question, why in the last 70 years has no one followed this up or found out how to do it?
Because not many try what is though to be impossible.
*
and RF amplifier's internal resistance.**
because the nucleus consists of these spins***
Mainstream science treats objects axiomatically. On the basic level it does not define what makes an object an object. Sometimes it attempts to define its properties, such as: a definite spatial boundary or spatial coordinate (but not always, see: "wave-particle duality") or mass (but only for physical objects such as a collection of gravitating matter or single gravitating massive particles - not for photons though), or just "anything that exists" and is "separate from the background". And when you ask: "What "background" ? - the mainstream science replies: "Well, the spatial background, of course". Then you ask: "So is this `spatial background` another object ?", and the usual reply is: "No, of course not". Then you ask: "So if the `spatial background` is not an object then how can it exist?", then they usually flip and change their mind: "No, no, spatial background is an object", then you ask "How can it be an object if the prerequisites for an object existence is a spatial boundary and coordinate or mass or the general separateness from the background ?"...and round and round it goes.
Quote from: Centraflow on 2020.07.12, 10:39:34
...an antenna matching unit for HF, a small ferrite Toroid, pass the wire which connects the two pancake coils through the Toroid and wind 2 turns of wire through the Toroid and connect to the matching unit. You should get full power into the coils and no reflection back to source, just my 2 pence worth.
What is going to happen to the energy provided by the RF amplifier? Is it going to get absorbed by the "antenna matching unit" ?
Quote from: Itsu on 2020.07.12, 09:50:41
Surely there must be a way to have the best of both worlds, meaning a good match from the PA to the pancake input circuit (resistive) and max. current in the pancake coils itselve (reactive).
But "good match" in Ham terms means no energy reflection back to the source (i.e. low "reflection coefficient" (Γ) and low VSWR). This also means, that an efficient and permanent energy conversion must take place from the AC energy to something else. If the near H-field generated by the coils does not perform any work and does not create any far-field EM waves (which permanently detach from the coil), then how are you going to get that permanent conversion? ...and if you do not convert this AC energy to something else, the coil is just going to store it and feed it back to the source in the next quarter cycle, (because energy cannot be destroyed).
Quote from: Itsu on 2020.07.12, 09:50:41
I found a good article covering the feedline coupling too here:
https://www.nonstopsystems.com/radio/frank_radio_antenna_magloop.htm
That is a good article that aims to maximize far-field EM radiation from a loop antenna.
But our goal is to create the maximum near-field magnetic field (H-field) between two loops - not to create a maximum far-field EM radiation.
Do these two maxima coincide ?
Ok, so i understand that we are pumping RF into the pancake coils, but it cannot do anything with it:
QuoteIf the near H-field generated by the coils does not perform any work and does not create any far-field EM waves (which permanently detach from the coil), then how are you going to get that permanent conversion? ...and if you do not convert this AC energy to something else, the coil is just going to store it and feed it back to the source in the next quarter cycle, (because energy cannot be destroyed).
I think my mind cannot grasp this, as to me it is then pointless to try.
This "maximum near-field magnetic field (H-field) between two loops" surely needs some power to be build up, won't that power need to be fed in by the PA as efficiently as possible?
Edit:reading somewhat further back you say:
QuoteOf course, if the H-field generated between the coils does some work, i.e. on the nuclear spins, then the part of the energy used to do that work, does not get reflected back to the source ...and the reflection coefficient (Γ) improves ( as does the SWR, which is just equal to 1+│Γ│ / 1-│Γ│ ).
So does this mean that ONLY at the moment "the H-field generated between the coils does some work etc...." this will be seen as a dip in the reflection coefficient and SWR?
If so, then there needs to be the focus on during the tuning i think.
Itsu
Quote from: Itsu on 2020.07.12, 15:34:41
Ok, so i understand that we are pumping RF into the pancake coils, but it cannot do anything with it:
I think my mind cannot grasp this, as to me it is then pointless to try.
This "maximum near-field magnetic field (H-field) between two loops" surely needs some power to be build up, won't that power need to be fed in by the PA as efficiently as possible?
Itsu
In my mind we need a certain value of H field or B field and that has an energy density. Thus we must supply the total energy of that field integrated overall space, which essentially means the space occupied by the near field. Our sample of protons does not absorb that total energy (it does not kill the field), so we get back most of that energy each half cycle. That is why we need a tuning capacitor so that the returned energy then gets stored and recycled. The matching network has to isolate the amplifier from that back and forth flow of energy between L and C while allowing the amplifier to supply only the energy lost each cycle. I don't see that as a problem, but you must remember that my formative years were before the semiconductor revolution, I was brought up using vacuum tubes. Yes, if you got a mismatch the anodes glowed red hot (a good warning sign!), but the tubes survived. I am sure there must be ways to prevent poor SWR from damaging modern PA's.
Smudge
Quote from: verpies on 2020.07.12, 11:33:44
I could agree with that when it comes to nuclear dynamic but when doing RF engineering, the notion of permanently converting the AC energy into something else is important, because if that conversion is not done, then the transmitting coil is just going to store this energy for the quarter of the AC cycle and return it all back to the source later.
When this happens, a large energy reflection is perceived by the source ( the "reflection coefficient" (Γ) approaching 1 ...and SWR approaching infinity ). This upsets the RF amplifier and its owner.
Of course, if the H-field generated between the coils does some work, i.e. on the nuclear spins, then the part of the energy used to do that work, does not get reflected back to the source ...and the reflection coefficient (Γ) improves ( as does the SWR, which is just equal to 1+│Γ│ / 1-│Γ│ ).
However, when the nuclei do not absorb the energy coherently because the owner of the RF amplifier is tuning the system off nuclear resonance, he can expect to have a "good match" (low Γ) only when the wiring's resistance* dissipates the AC energy as heat or the pancake coils efficiently convert this energy into far-field EM radiation.
As I see it the at nuclear resonance the nuclei absorb only a small portion of the total field energy, so we need LC resonance to recycle the large quantity of unused energy. And I think it likely that at this early stage the small portion of energy absorbed will be over-shadowed by a larger portion of energy lost to the coil resistance. Thus the system will be well matched to that resistive loss which does not have an absorption peak, so I don't see the problem in the same way that you do.
Smudge
Quote from: verpies on 2020.07.12, 12:41:58
What is going to happen to the energy provided by the RF amplifier? Is it going to get absorbed by the "antenna matching unit" ?
Well you know it is not going to do that, so why ask!!!!!!!!!!
It matches the complex impedance of the transmitter to that of the input end of the feedline. The input impedance of the transmission line will be different from the characteristic impedance of the feedline if the impedance of the antenna on the other end of the line does not match the line's characteristic impedance. The consequence of the mismatch is that the line's impedance (voltage to current ratio and phase) will oscillate along the line, or equivalently, raise out-of-phase voltage standing waves and current standing waves along the feedline.
I can see you don't want me here so I am gone, I have more important things to do with what time I have left in my life, you are a very
condescending person and I think you need to stand back and look into that, your views are not everyone's as are not mine.
By the way of a compliment, your English is very good for eastern Europe.
Regards
Mike 8)
QuoteIn my mind we need a certain value of H field or B field and that has an energy density.
Thus we must supply the total energy of that field integrated overall space, which essentially means
the space occupied by the near field.
Our sample of protons does not absorb that total energy (it does not kill the field), so we get back most of that energy each half cycle.
That is why we need a tuning capacitor so that the returned energy then gets stored and recycled.
The matching network has to isolate the amplifier from that back and forth flow of energy between L and C while allowing the amplifier to supply only the energy lost each cycle.
I don't see that as a problem, but you must remember that my formative years were before the semiconductor revolution, I was brought up using vacuum tubes.
Yes, if you got a mismatch the anodes glowed red hot (a good warning sign!), but the tubes survived.
I am sure there must be ways to prevent poor SWR from damaging modern PA's.
Smudge
Thanks Smudge,
so i take that as that you agree with what verpies is saying that energy is sloshed back and forth ("so we get back most of that energy each half cycle.")
These modern PA's are able to cope with bad SWR as their power is being regulated back.
Thanks Mike,
it seems you, like me, have a hard time grasping that the PA is unable to deliver any power to the circuit as nothing is radiated or consumed in the "no tuned" situation.
No "matching unit" is able to change that, so the PA gets overloaded if no precausions are followed.
So caution is needed in the no tune situation untill the effect we are after is revealing itselve.
Itsu
Quote from: Centraflow on 2020.07.12, 16:20:16
Well you know it is not going to do that, so why ask!!!!!!!!!!
To provoke a discussion about the fate of energy provided by the source.
Quote from: Centraflow on 2020.07.12, 16:20:16
You are a very condescending person and I think you need to stand back and look into that
Yes, I do not have the best social skills but notice, that I never use Ad Hominem remarks.
Also, I know what I am talking about technically and my mind is flexible enough that you can change it with logical arguments.
Quote from: Centraflow on 2020.07.12, 16:20:16
By the way of a compliment, your English is very good for eastern Europe.
Thank you, but that is also an Ad Hominem remark, ...even if a positive one.
Quote from: Centraflow on 2020.07.12, 16:20:16
I can see you don't want me here so I am gone,
You are mistaken, I do want you here (minus the Ad Hominem remarks).
Quote from: Centraflow on 2020.07.12, 16:20:16
It matches the complex impedance of the transmitter to that of the input end of the feedline. The input impedance of the transmission line will be different from the characteristic impedance of the feedline if the impedance of the antenna on the other end of the line does not match the line's characteristic impedance. The consequence of the mismatch is that the line's impedance (voltage to current ratio and phase) will oscillate along the line, or equivalently, raise out-of-phase voltage standing waves and current standing waves along the feedline.
What you wrote above is all true, but you did not answer the question about the fate of energy supplied by the PA.
I made some tests using the I-probe.
Seems the best pickup is with the curved loop parallel to the pancake coil windings.
Max signal in the middle of the 11 turns.
Video here: https://www.youtube.com/watch?v=ijSGoCanuEM
Itsu
Quote from: Itsu on 2020.07.12, 15:34:41
I think my mind cannot grasp this, as to me it is then pointless to try.
Oh no, no - don't you give up.
Quote from: Itsu on 2020.07.12, 15:34:41
This "maximum near-field magnetic field (H-field) between two loops" surely needs some power to be build up, won't that power need to be fed in by the PA as efficiently as possible?
Your statement is correct* but only in reference to a constant level of magnetic flux or a coil supplied with DC.
The answer to your question at the end is "yes" but only for a constant level of magnetic flux or for the 1st quarter of the AC cycle.
What you are failing to consider is what happens AFTER the magnetic field has been built up to its maximum level in the 1st quarter of the AC cycle.
When you do that, it will become apparent that in order to maintain the oscillation, this magnetic energy needs to be decreased back to zero. This can be done either by converting it back to electric current and dissipating its energy as heat in some resistance or radiating it as a far-field EM wave or recycling it through a capacitor. This resistance and/or capacitance can be outside or inside the PA.
In case of energy recycling through a capacitance, the capacitance then becomes the sink of energy and for the next quarter of the AC cycle - the source of energy for the coil. And so, the energy sloshes back and forth between the capacitor and coil in the familiar parallel LC circuit manner. When this "sloshing" is lossless then the PA does not need to replenish that AC energy, so the
energy transfer between the PA and LC tank stops ...and the PA "sees" an open circuit. That apparent "open circuit" presents very poor match to the PA's output and this is what your SWR meter shows.
In realistic LC circuits, the energy "sloshing" between the coil and capacitance is being lost continuously in the resistance (as heat) and through EM radiation (because of radiation resistance), ultrasonics, ..and hopefully to proton's precession. The PA must then replenish this lost energy. The more needs to get replenished, the more energy flows form the PA to the LC circuit. Eventually this reaches an equilibrium, where the PA supplies as much energy as is being lost. Maximum energy transfer happens when the impedance of the PA is equal to the impedance of the load....albeit not the max efficiency. See:
MPTP (https://en.wikipedia.org/wiki/Maximum_power_transfer_theorem#Maximizing_power_transfer_versus_power_efficiency).
Quote from: Itsu on 2020.07.12, 15:34:41
reading somewhat further back you say:
So does this mean that ONLY at the moment "the H-field generated between the coils does some work etc...." this will be seen as a dip in the reflection coefficient and SWR?
Yes, otherwise no energy will flow from the PA to the ideal parallel LC circuit.
However, realistic parallel LC circuits always have resistive and radiative losses and these will need to be replenished by the PA, causing some net energy transfer and dipping of the reflection coefficient and SWR. My point here is that the maximization of the radiative losses (radio wave generation) is not synonymous with the maximization of near H-field amplitude because:
- the radiative losses (generation of far-field EM waves) are proportional to the product of the radiation resistance and the square of the current amplitude.
- the near H-field amplitude is proportional to the product of current amplitude and coil turns (ampturns),
The situation is a little different when the PA is connected in series with a series LC circuit, because then all the energy sloshing goes through the PA, so the PA's internal resistance and reactances participate in it.
Quote from: Itsu on 2020.07.12, 15:34:41
If so, then there needs to be the focus on during the tuning i think.
Yes, the goal is still to obtain the maximum magnetic flux amplitude and that means the maximum ampturns.
If you think that your HF new magnetic field sensor is trustworthy (the one made out of the Litzed loop in the i-Probe), then this is the best indicator for tuning for the maximum amplitude of the magnetic flux.
*
except you should have used the word "energy" instead of "power"
Quote from: Itsu on 2020.07.12, 19:05:17
I made some tests using the I-probe.
Seems the best pickup is with the curved loop parallel to the pancake coil windings.
Max signal in the middle of the 11 turns.
It looks like this magnetic probe is not sensitive only at its tip, bummer. ..but it is still useful for relative peaking when tuning the trimmer.
Also, due to the nature of the magnetic coupling the angle of the Litz loop must influence its sensitivity in the given direction, so always rotate it in 3D to find the field's orientation.
I am not surprised that the SWR & Power meter readings do not correspond to the peak sensed by the magnetic probe, because not the entire current flowing into the pancake circuit is responsible for generating the desired H-field. Some of this current flows between the coil's turns (via interwinding capacitance) and does not generate the H-field in the desired direction.
This is why the magnetic field probe measuring method is superior - it ignores all the electronic parasitics!
QuoteOh no, no - don't you give up.
I did not say i give up, its just that my mind says that what i am trying is pointless, but i have tried pointless things many times before :P
Ok, i am starting to grasp it, thanks for your persistence.
Quote* except you should have used the word "energy" instead of "power"
yes i knew that when i read it back, but i thought nobody will notice, damn...
Itsu
If you fall in love with you new HF magnetic probe and want to maximize its performance, then bring out your microscope and solder each individual strand of the Litz wire loop individually and varnish it so the solder joints do not touch. Stagger the joints if you do not want a bulge.
I know, I know it's a lot of tedious work - but consider it a part of your test gear collection and that it does not need to be repeated.
Too bad, you cannot calibrate this new magnetic probe because you have no way of generating a constant amplitude H-field of varying frequency.
Quote from: verpies on 2020.07.12, 20:13:34
It looks like this magnetic probe is not sensitive only at its tip, bummer. ..but it is still useful for relative peaking when tuning the trimmer.
Also, due to the nature of the magnetic coupling the angle of the Litz loop must influence its sensitivity in the given direction.
I am not surprised that the SWR & Power meter readings do not correspond to the peak sensed by the magnetic probe, because not the entire current flowing into the pancake circuit is responsible for generating the desired H-field. Some of this current flows between the coil's turns (via interwinding capacitance) and does not generate the H-field in the desired direction.
This is why the magnetic field probe measuring method is superior - it ignores all the electronic parasitics!
Yes, that relative peaking when tuning the trimmer is what i thought was the main purpose, this works nicely i think.
Still a question on my mind, these 2 pancake coils are in series (bucking), but we have technically not a series capacitor, so does that make them a parallel LC circuit due to their interwinding (parallel) capacitance?
Itsu
Quote from: Itsu on 2020.07.12, 20:28:59
Still a question on my mind, these 2 pancake coils are in series (bucking), but we have technically not a series capacitor, so does that make them a parallel LC circuit due to their interwinding (parallel) capacitance?
I think so, but I am not completely sure. It sometimes looks like a mixed series/parallel circuit.
I would like others to analyze this question too: Partzman, Smudge, Centralflow ?
Anyway, the H-field's direction between the pancake coils will be different than outside them. Amplitude too, especially if they are not identical.
Quote from: verpies on 2020.07.12, 20:18:25
If you fall in love with you new HF magnetic probe and want to maximize its performance, then bring out your microscope and solder each individual strand of the Litz wire loop individually and varnish it so the solder joints do not touch. Stagger the joints if you do not want a bulge.
I know, I know it's a lot of tedious work - but consider it a part of your test gear collection and that it does not need to be repeated.
Too bad, you cannot calibrate this new magnetic probe because you have no way of generating a constant amplitude H-field of varying frequency.
Pfff, you must be joking, right?
Perhaps with the smaller litz wire coming, it has only 120 separate strands :D
I forgot to mention that while measuring a coil by VNA, the near H-field which surrounds the coil and stores its reactive energy* manifests itself as the imaginary component of its impedance (reactance) while the far-field EM radiation resistance manifests itself as the real part of its impedance (resistance) ...just like the ohmic resistance of the wire.
As usual, the energy stored in the near reactive field (magnetic or electric) does not dissipate and is recoverable*, while the EM energy dissipates permanently as it is radiated away - thus it is not recoverable.
* Later in the AC cycle, this energy will be returned to its source (i.e. to the parallel capacitor ...or to the serial capacitor through the PA).
QuoteQuote from: Itsu on 2020-07-12, 21:28:59
Still a question on my mind, these 2 pancake coils are in series (bucking), but we have technically not a series capacitor, so does that make them a parallel LC circuit due to their interwinding (parallel) capacitance?
Quote from: verpies on 2020.07.12, 20:34:23
I think so, but I am not completely sure. It sometimes looks like a mixed series/parallel circuit.
I would like others to analyze this question too: Partzman, Smudge, Centralflow ?
Anyway, the H-field's direction between the pancake coils will be different than outside them. Amplitude too, especially if they are not identical.
I am asking because the high impedance (3KOhm) you measured during resonance (4.9Mhz) on your new pancake coil (also seen on my coils) using your VNA, points to parallel LC resonance (high impedance, low current and high voltage) i think.
If we do include the 100pF trimmer cap in this series pancake coil circuit it is a series LC circuit which in resonance would mean Low impedance, high current and low voltage.
So does it depends on where you measure (pancakes only or trimmer plus pancakes etc.) on what outcome impedance wise it has?
Itsu
Quote from: Itsu on 2020.07.13, 10:05:40
So does it depends on where you measure (pancakes only or trimmer plus pancakes etc.) on what outcome impedance wise it has?
Of course it does.
All of the plots of my coil, that I've posted, were of the naked coil without any external capacitors connected.
The attached plot also shows the naked coil.
Note that its maximum inductive reactance (blue trace above the 0 line) does not coincide with the peak of the orange trace.
The orange trace is the real component
* of the impedance which contains the radiation resistance, so it can be expected that at the peak of this trace, the coil radiates the most radio waves.
The blue trace is the imaginary component of the impedance (inductive reactance above the zero line, and capacitive below it), so it can be expected that at the peak of this trace, the coil builds up the highest near H-field.
**The green trace are the real and imaginary components combined ...or the magnitude of the complex impedance. Note that this magnitude is sometimes smaller than the real component :o
*
it is a sum of the DC resistance, radiation resistance and additional resistances due to the skin-effect and proximity-effect.**
This is what you new Litzed loop magnetic probe would show.
Quote from: verpies on 2020.07.13, 10:13:58
Of course it does.
The blue trace is the imaginary component of the impedance (inductive reactance above the zero line, and capacitive below it), so it can be expected that at the peak of this trace, the coil builds up the highest near H-field.
Are you sure about this? Wouldn't the highest current and therefore the highest H field occur at the zero crossing point of the impedance component?
Also typically, inter-winding capacitance would model as a parallel capacitance and inter-coil capacitance along with any parasitics would model as series capacitance.
Quote from: partzman on 2020.07.13, 13:41:59
Are you sure about this? Wouldn't the highest current and therefore the highest H field occur at the zero crossing point of the impedance component?
Thanks for catching this ambiguity. I meant it when all other things are being equal (including the current).
The way I wrote it suggests, that a higher current flows with higher magnitude of impedance when driven with a constant voltage - this is not true.
In the attached plots I changed the green plot to show the Admittance and the orange plot to show the Conductance. The reciprocals of what was shown before.
Now the peaks of these traces show frequencies where more current would flow when driven by a constant voltage.
And the green markers 1, 2, 4, 5, 7 illustrate where it approaches the standard impedance of 50Ω and maximal power transfer would occur.
Quote from: partzman on 2020.07.13, 13:41:59
Also typically, inter-winding capacitance would model as a parallel capacitance
Yes, so any added external capacitance will absorb this internal capacitance.
Quote from: partzman on 2020.07.13, 13:41:59
...and inter-coil capacitance along with any parasitics would model as series capacitance.
Which would explain that 2nd peak Itsu was observing with two coils.
Quote from: verpies on 2020.07.13, 10:13:58
Of course it does.
All of the plots of my coil, that I've posted, were of the naked coil without any external capacitors connected.
The attached plot also shows the naked coil.
Note that its maximum inductive reactance (blue trace above the 0 line) does not coincide with the peak of the orange trace.
The orange trace is the real component* of the impedance which is contains the radiation resistance, so it can be expected that at the peak of this trace, the coil radiates the most radio waves.
The blue trace is the imaginary component of the impedance (inductive reactance above the zero line, and capacitive below it), so it can be expected that at the peak of this trace, the coil builds up the highest near H-field.**
The green trace are the real and imaginary components combined ...or the magnitude of the complex impedance. Note that this magnitude is sometimes smaller than the real component :o
* it is a sum of the DC resistance, radiation resistance and additional resistances due to the skin-effect and proximity-effect.
** This is what you new Litzed loop magnetic probe would show.
Great, stuff O0
QuoteThe attached plot also shows the naked coil.
Note that its maximum inductive reactance (blue trace above the 0 line) does not coincide with the peak of the orange trace.
Thats the difference i saw with the peak of my I-probe (blue) and the peak of the Power/SWR meter (orange).
Why do we have a second array of peaks around 29Mhz? Does that mean 2 resonance points?
Itsu
Quote from: Itsu on 2020.07.13, 16:01:56
Why do we have a second array of peaks around 29Mhz? Does that mean 2 resonance points?
2 ?!
We have many more!
See the plot below with the white background. It was done with the closed coil up to 500MHz (warning: the VNA was not calibrated for this plot).
A badly terminated coax also has many diminishing│Z│peaks (see the plot with the black background and only 1 green trace). The white plot resembles that, except the green peaks are not equally spaced.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3691.0;attach=33443)
So to demonstrate that the other peaks are caused by reflections. I have treated the coil as a coiled transmission line and have terminated it in the middle break with a 200Ω resistor.
Below is a TDR plot from a PulseGen+Scope with the white trace showing the reflected pulse from the terminated pancake coil and the yellow trace showing reflected pulses from the unterminated coil (open break).
Also, below is a 500MHz span VNA plot after the termination. Compare it to the 500MHz plot from the previous message (warning: the vertical scales are not the same).
OK, the 50Mhz plot only showed 2 "resonance points", so therefor my remark about 2.
So those extra peaks are reflections not resonance points, in accordance with the theory which describes that a coil could have only 1 resonance point, at the frequency where Xc and Xl are equal O0
Below 2 SWR scans (0 to 500Mhz) of my 2 pancake coils with an open toroidal coil inbetween.
First is without input circuit, so naked coils,
second is with input circuit (balun, caps).
Both show a flat first 50Mhz with a high SWR
Now, the naked plot is very different considering that your coil has almost the same amount of turns as mine (11 vs. 10) wound with the same wire diameter (but mine is Litzed) and almost the same OD and ID (mine is 112mm and 91mm, respectively).
There are three differences, though:
- Number of coils
- Interconnects
- Winding scheme
I think the first point might be the elephant in the room.
*
I am assuming that you are making these measurements with your DSA815 + the VB1020 1-port "VSWR Bridge".
Yes, i agree, must be mainly the 2 coils in bucking series mode.
Correct on the setup
Below 2 FM sweeps (1Khz to 10Mhz) made with the i-probe close up to the front pancake, middle of turns.
First is without input circuit, so naked coils,
second is with input circuit (balun, caps).
Finally, i made a sweep (1Mhz to 20Mhz in 10s) using the PA and looking at the power meter response.
Seems we have a varying input power into the circuit from 0.5W to 8W over this 20Mhz range.
Be aware there are many variables here like the linearity of the PA across this range, the possiblity of the PA to regulate power back in case of mismatch, the accuracy of the used power meter etc.
Video here: https://www.youtube.com/watch?v=ZKsKVZ5d38Q
Itsu
Quote from: Itsu on 2020.07.14, 08:38:32
Below 2 FM sweeps (1Khz to 10Mhz) made with the i-probe close up to the front pancake, middle of turns.
Lets call it an "H-probe" when used to sense external magnetic fields generated by other coils.
I like the idea. Actually I was just thinking of suggesting it to you but with the SA doing the sweeping and sensing.
I didn't watch the video yet - I'm mobile now on slow connection.
Questions about the new build coils i will be making with 0.93mm litz wire.
Of these 2 pancake coils, one will be CW, the other CCW so they can be placed facing each other and be in bucking mode.
This means both pancake coils will be very close to the toroidal coil, only separated by the varnish and litz wire cloth and to each other (~14mm).
Also i will use thin (3mm) plexiglass as a former, so the magnets will be that close to the pancake coils and thus close to each other (~21mm)
How does this effect the setup, is it good or bad?
If bad is there an optimum distance between the severall layers?
What will be the approximate resonance frequency of the NMR?
Anything else?
Itsu
Quote from: verpies on 2020.07.14, 09:08:48
Lets call it an "H-probe" when used to sense external magnetic fields generated by other coils.
I like the idea. Actually I was just thinking of suggesting it to you but with the SA doing the sweeping and sensing.
I didn't watch the video yet - I'm mobile now o slow connection.
Ok, H-probe it is O0
I can try the SA for doing the sweeping and sensing.
Nothing heard from Conrad om my repeated request for the water tube 3D file, so he must have forgotten to send it and now is "on the move".
Is there someone else out there who had requested and received the 3D file when Conrad offered it and is willing to send it to me?
You can PM me here or via Overunity.com, thanks.
Itsu
Quote from: Itsu on 2020.07.14, 09:15:28
Questions about the new build coils i will be making with 0.93mm litz wire.
Of these 2 pancake coils, one will be CW, the other CCW so they can be placed facing each other and be in bucking mode.
Don't wind them yet, because I am making an experiment with a different winding method to investigate if the reflections, which I illustrated yesterday, will be different and better. The difference in this second symmetrical winding method is which wires cross on top, ( see the attached diagrams for the difference ).
Also, when pancake coils are wound in this symmetric way, the monikers CW and CCW lose sense, because the direction of the circumferential current can be reversed anytime be reversing the connection to the coil.
Finally, even if the coil was not wound in the symmetrical manner (just as an Archimedes spiral), then it seems to me that winding BOTH of them the same way and then flipping one over to face the other, would produce a bucking field.
Quote from: Itsu on 2020.07.14, 09:15:28
This means both pancake coils will be very close to the toroidal coil, only separated by the varnish and litz wire cloth and to each other (~14mm).
Also i will use thin (3mm) plexiglass as a former, so the magnets will be that close to the pancake coils and thus close to each other (~21mm)
Anyway, I am attaching a simulation of the aiding and bucking fields of coaxial parallel pancake coils. The magnitudes of both magnetic and electric fields are included (opposing ones are on the right (b) ).
This should illustrate to you how far the electric field extends. The purpose of the Litz cloth is to ground these electric fields so the capacitive coupling does not allow them to reach the toroidal coil.
P.S.
If you want the e-book where these sims come from, PM me an FTP info where I can upload it.
QuoteDon't wind them yet, because I am making an experiment with a different winding method to investigate if the reflections, which I illustrated yesterday, will be different and better.
The difference in this second symmetrical winding method is which wires cross on top, ( see the attached diagrams for the difference ).
Ok, i will wait, i still do not have the new litz, so have to wait anyway.
QuoteAlso, when pancake coils are wound in this symmetric way, the monikers CW and CCW lose sense, because the direction of the circumferential current can be reversed anytime be reversing the connection to the coil.
Ok makes sense :)
QuoteFinally, even if the coil was not wound in the symmetrical manner (just as an Archimedes spiral), then it seems to me that winding BOTH of them the same way and then flipping one over to face the other, would produce a bucking field.
well, i agree, but Smudge adviced here:
https://www.overunityresearch.com/index.php?topic=3924.msg82765#msg82765
to use symmetry to overcome the capacitive coupling between pancakes and toroidal coil by connecting the pancakes either from their outside connections (series connect them by their inner wires) or vv.
So then i had to flip one over.
QuoteAnyway, I am attaching a simulation of the aiding and bucking fields of coaxial parallel pancake coils. The magnitudes of both magnetic and electric fields are included (opposing ones are on the right (b) ).
This should illustrate to you how far the electric field extends. The purpose of the Litz cloth is to ground these electric fields so the capacitive coupling does not allow them to reach the toroidal coil.
P.S.
If you want the e-book where these sims come from, PM me an FTP info where I can upload it.
I have no FTP at the moment (no more fixed IP), but will set one up soon.
Itsu
Quote from: Itsu on 2020.07.14, 08:38:32
Video here: https://www.youtube.com/watch?v=ZKsKVZ5d38Q
I watched the video.
Does your power meter measure the real power or the apparent power ?
i.e. does your power meter read zero when you connect only a high Q capacitor at its output ?
Apparently apparent power (reactive power and true power) as with a 624nF capacitor at its output
and sweeping from 1 to 20Mhz, i see a similar output pattern as yesterday.
Itsu
Quote from: Itsu on 2020.07.14, 20:17:55
Apparently apparent power (reactive power and true power) as with a 624nF capacitor at its output
and sweeping from 1 to 20Mhz, i see a similar output pattern as yesterday.
So that explains it.
The readings shown in the video do not necessarily mean that the energy provided by the PA goes into the pancakes and does useful/useless
* work there. These readings can mean, that most of the energy provided by the PA is returned to it.
Do you have a recirculator so you can separate the forward power from the reflected power ?
*
Joule heating, ultrasonics, radio wave emmission into the far-field.
QuoteLets call it an "H-probe" when used to sense external magnetic fields generated by other coils.
I like the idea. Actually I was just thinking of suggesting it to you but with the SA doing the sweeping and sensing.
H-probe measurement (0 to 50Mhz) on the front pancake middle of windings using the TG and SA
Itsu
Quote from: verpies on 2020.07.10, 23:18:24
Measurments of my precision pancake coil:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns* of 1mm Litz wire (68 strand), 3.2m long, antiradial weave 1.
When the break is opened:
Intrawinding capacitance: 174.75pF (@100kHz)
ESR: 218Ω (@100kHz)
When the break is closed:
Inductance: 19.65µH (@100kHz)
DC resistance: 0.22Ω
Measurments of my 2nd precision pancake coil:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns
* of 1mm Litz wire (68 strand), 3.2m long, antiradial weave 2.
When the break is opened:
Intrawinding capacitance: 173.85pF (@100kHz)
ESR: 195Ω (@100kHz)
When the break is closed:
Inductance: 19.72µH (@100kHz)
DC resistance: 0.22Ω
QuoteSo that explains it.
The readings shown in the video do not necessarily mean that the energy provided by the PA goes into the pancakes and does useful/useless* work there. These readings can mean, that most of the energy provided by the PA is returned to it.
Do you have a recirculator so you can separate the forward power from the reflected power ?
* Joule heating, ultrasonics, radio wave emmission into the far-field.
Ok, no, i do not have a recirculator, and it seems most are designed for high to very high frequencies.
By the way, i found this informative website: https://www.qsl.net/va3iul/Homebrew_RF_Circuit_Design_Ideas/Homebrew_RF_Circuit_Design_Ideas.htm
Quote from: verpies on 2020.07.14, 22:14:09
Measurments of my 2nd precision pancake coil:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns* of 1mm Litz wire (68 strand), 3.2m long, antiradial weave 2.
When the break is opened:
Intrawinding capacitance: 173.85pF (@100kHz)
ESR: 195Ω (@100kHz)
When the break is closed:
Inductance: 19.72µH (@100kHz)
DC resistance: 0.22Ω
So the Weave 2 design does not show any significant difference from the weave 1 design, at least looking at these parameters.
Itsu
Quote from: Itsu on 2020.07.15, 07:04:58
So the Weave 2 design does not show any significant difference from the weave 1 design, at least looking at these parameters.
Yes, but there is a small difference in reflection. I'll post scopeshots later.
Also, take a look at this video for a guy making and using a directional coupler for quite a low frequency.
https://youtu.be/byF1FLdbUiA
https://youtu.be/iBK9ZIx9YaY
Your VSWR Bridge is a similar device, but it has 3 terminals and might burn out under high power.
I have some power meters (Bird 43 and a Gotting), but both are for higher frequency (Gotting) and/or for higher power (500W) at 4Mhz (Bird 43), so won't show much on the ±5W on 4Mhz we are dealing with.
But what i see on them is that in our system we have almost as much reflected power as power going in.
Got my litz wire and made the 2x 3mm thick plexiglas formers (150mm by 150mm square), so i am ready to wind them.
Aiming for 14 turns (0.93mm litz wire), so will cover an area of OD: 113mm, ID:87mm.
Just need to know the winding scheme, Weave 1 or Weave 2.
Regards Itsu
Quote from: Itsu on 2020.07.17, 08:18:13
Just need to know the winding scheme, Weave 1 or Weave 2.
They are very similar but Weave1 should generate a more symmetric H-field. I did not confirm it experimentally, though, so please get a 2nd opinion on that from Smudge or another expert.
I did not post the TDR scopeshots of the picopulse yet because I blew up my only 2N2369 transistor and I have only the following high-frequency BJTs in stock:
BF199
BC547B
2N3904
CX237B
MPS2907A
BC556B
Ok, so i would go for Weave1, so please Smudge or any other expert, can you confirm?
Here a link to the both coils, Coil2.png = weave2, coil.png = weave1:
https://www.overunityresearch.com/index.php?topic=3924.msg83193#msg83193
Bummer that you blew up your 2N2369, Alltransistors.com gives some equivalents, but none in your list.
2N3227 Si NPN 0.36 40 20 4 0.5 200 500 4 100 TO18
2N4137 Si NPN 0.36 40 20 4 0.2 200 500 4 40 TO18
2N4873 Si NPN 0.36 40 15 0.2 200 600 4 110 TO18
2N5272 Si NPN 0.36 40 20 0.2 200 500 4 100 TO18
2N915A Si NPN 0.36 70 50 5 0.2 200 600 4 50 TO18
BSX20 Si NPN 0.36 40 15 5 0.5 200 500 4 40 TO18
BSX93 Si NPN 0.36 40 15 4 0.5 200 500 4 40 TO18
CV7555 Si NPN 1.2 40 15 4.5 0.2 200 500 4 40 TO18
FT2369 Si NPN 0.36 40 15 4 0.2 200 500 4 40 TO18
Itsu
Quote from: Itsu on 2020.07.17, 15:45:33
Bummer that you blew up your 2N2369, Alltransistors.com gives some equivalents, but none in your list.
2N3227 Si NPN 0.36 40 20 4 0.5 200 500 4 100 TO18
2N4137 Si NPN 0.36 40 20 4 0.2 200 500 4 40 TO18
2N4873 Si NPN 0.36 40 15 0.2 200 600 4 110 TO18
2N5272 Si NPN 0.36 40 20 0.2 200 500 4 100 TO18
2N915A Si NPN 0.36 70 50 5 0.2 200 600 4 50 TO18
BSX20 Si NPN 0.36 40 15 5 0.5 200 500 4 40 TO18
BSX93 Si NPN 0.36 40 15 4 0.5 200 500 4 40 TO18
CV7555 Si NPN 1.2 40 15 4.5 0.2 200 500 4 40 TO18
FT2369 Si NPN 0.36 40 15 4 0.2 200 500 4 40 TO18
That list helped O0
It jogged my memory that I have seen a BSX transistor in my Mom's old CRT TV ~20 years ago. Don't ask me how I remember that.
She never throws anything away, so I will have it tonight.
Anyway, I found the attached Pushing Tool very helpful with winding these pancake coils.
Also, using a low-viscosity SuperGlue (cyanoacrylate) in an insulin syringe w/ integrated needle (from a pharmacy), is very useful for securing the Litz wire around the wire crossings, because this glue polymerizes instantly when subjected to pressure from the pushing tool.
How did you remember that! ;D
Yes, people seem to store the strangest things in their heads, now this comes in handy.
Now where did your mom store that old CRT TV......
Ok on the winding tool, i don't like superglue on plexiglas as it tends to give smears and dirty up the whole thing, but in this case i probably need to.
Itsu
Quote from: Itsu on 2020.07.18, 07:55:38
How did you remember that! ;D
Yes, people seem to store the strangest things in their heads, now this comes in handy.
Now where did your mom store that old CRT TV......
I got the BSX20! It was in the attic ..with everything else
Quote from: Itsu on 2020.07.18, 07:55:38
Ok on the winding tool, i don't like superglue on plexiglas as it tends to give smears and dirty up the whole thing, but in this case i probably need to.
Yup, but the insulin syringe gives you so much control, with such small amounts that it doesn't spread around.
So I reconstructed the Picopulser with the BSX20 transistor and 3pF cap ( actually 2x1.5pF ).
With a 100VDC supply, it generates 7.3VP-P, 492ps pulse (FWHM (https://en.wikipedia.org/wiki/Full_width_at_half_maximum)).
Thats a very clean pulse, so the BSX20 might be performing better then the original 2N2369 as the diagram for the LT1073 calls for a 350ps risetime pulse O0
Itsu
This was with a 50Ω coax directly to a 50Ω input of LowZ scope.
It wasn't so clean when I was probing it with a 1MΩ probe set to 10x. The probe's capacitance took time getting charged ...and distorted the pulse.
Quote from: Itsu on 2020.07.17, 15:45:33
Ok, so i would go for Weave1, so please Smudge or any other expert, can you confirm?
Here a link to the both coils, Coil2.png = weave2, coil.png = weave1:
https://www.overunityresearch.com/index.php?topic=3924.msg83193#msg83193
Seems we are not getting a second opinion on your Weave* coils from Smudge or any other coil expert which is disappointing.
So i will go for Weave1 next week as currently i am trying to repair a P6302 current probe which suffer from broken wires in its plug which is a pain to get open.
Itsu
Quote from: Itsu on 2020.07.19, 09:33:54
I am trying to repair a P6302 current probe which suffer from broken wires in its plug which is a pain to get open.
...and I wasted the entire day yesterday fixing a broken firmware in the scope.
Broken firmware!? So you went in, debugged the firmware then reflashed it?
Sounds like the manufacturer could be interested in it.
Itsu
Quote from: Itsu on 2020.07.19, 10:32:25
Broken firmware!? So you went in, debugged the firmware then reflashed it?
Yeah, with IDAPro and JTAG flasher.
Anyway, below are the waveforms of these two pancake coils (with the middle break open, although the photo shows the break bridged by a pot).
The upper trace is of Weave1, and the lower trace is of Weave2.
Notes:
These are not unprocessed traces. First, the pulse response of all the interconnects was saved to a reference waveform designated as M1.
Next, the opened Weave1 pancake coil was connected to Ch1 and the difference of Ch1-M1 was saved to a upper reference waveform M2.
Finally, the opened Weave2 pancake coil was connected to Ch1 and the difference of Ch1-M1 was displayed as the lower trace F1.
And below is an illustration of the difference that a termination makes.
As before, the upper trace (M2) depicts the pulse response of a Weave1 pancake coil with the middle break opened, and the the lower trace (F1) illustrates the same coil with a termination of 200Ω at the middle break.
As you can see, the largest difference is around +10ns, which is approximately the round trip delay of ½ of the wire length used to wind this coil (...or the length of the wire up to the middle break).
And below is an illustration of the difference that a termination makes...but this time with Weave2
The upper trace (M2) depicts the pulse response of a Weave2 pancake coil with the middle break opened, and the the lower trace (F1) illustrates the same coil with a termination of 200Ω at the middle break.
As you can see, the largest difference is still around +10ns, which is approximately the round trip delay of ½ of the wire length used to wind this coil (...or the length of the wire up to the middle break).
Nice screenshots, they show almost no difference between weave1 or weave2.
Is one coil more easy to build then the other, or are they similar difficult?
Quote from: Itsu on 2020.07.19, 15:42:55
Is one coil more easy to build then the other, or are they similar difficult?
There is a difference but it is only mental. Apparently it is easier for my brain to keep track of Weave 2. There is no mechanical difference.
There should be a difference in their H-field symmetry, though.
Ok, is one more symmetric (better) then the other?
Yes
Quote from: Itsu on 2020.07.19, 09:33:54
Seems we are not getting a second opinion on your Weave* coils from Smudge or any other coil expert which is disappointing.
Sorry about that. Having just recently moved house my time has been spent laying paving slabs and moving furniture around, I haven't really paid much attention to OUR. I have to say that I am not sure that such finesse in coil design is needed, but I am no expert. I am pleased that you and verpies are doing this work and long for the day when you actually discover the proton spin absorption peak or trough. I understand verpies' concerns that the signal could be too difficult to find until you have mastered the problems of getting high current into the pancake coils and have eliminated (as far as possible) unwanted magnetic and electric coupling from the pancake coils to the toroidal coil. I think my investigative approach would have been different, I would have spent more time just searching for the proton resonance then, having found it, I would work at that fixed frequency fine-tuning the system. But that may well have ended in failure in not finding the proton resonance.
You asked earlier what proton frequency would occur in your latest build. If we had some way of measuring the permanent magnet field between the pancakes we could give you an accurate answer. My FEMM result for your magnet separation of 21mm is a field very close to 0.1 Tesla that gives a frequency of 4.258 MHz, but FEMM is not a 3D simulator.
Smudge
Quote from: Itsu on 2020.07.14, 20:37:46
H-probe measurement (0 to 50Mhz) on the front pancake middle of windings using the TG and SA
Could you do it again but with the the H-probe positioned between the 2 pancake coils (connected in series opposing) ?
Prior to the sweeps, please find the one H-probe orientation and position, which gives you the strongest signal between the coils and mechanically fix the H-probe in that orientation and position for all sweeps. (a photo of this position would be nice to see, too).
Anyway, if you decide to do this, please make these 6 sweeps:
- pancake coils powerd by the PA, through a symmetrizing 1:1 balun and the parallel cap (Smudge's style).
- pancake coils powerd by the PA, and the parallel cap (Smudge's style).
- naked pancake coils powerd by the PA through a symmetrizing 1:1 balun.
- naked pancake coils powerd by the PA.
- pancake coils powered by the PA and a cap in series (Partzman's style).
- pancake coils powered by the PA, through a symmetrizing 1:1 balun and a cap in series (Partzman's style but w/balun).
Use long sweep times so the ringing resonant systems have the time to settle.
Quote from: Smudge on 2020.07.20, 14:18:57
Sorry about that. Having just recently moved house my time has been spent laying paving slabs and moving furniture around, I haven't really paid much attention to OUR. I have to say that I am not sure that such finesse in coil design is needed, but I am no expert. I am pleased that you and verpies are doing this work and long for the day when you actually discover the proton spin absorption peak or trough. I understand verpies' concerns that the signal could be too difficult to find until you have mastered the problems of getting high current into the pancake coils and have eliminated (as far as possible) unwanted magnetic and electric coupling from the pancake coils to the toroidal coil. I think my investigative approach would have been different, I would have spent more time just searching for the proton resonance then, having found it, I would work at that fixed frequency fine-tuning the system. But that may well have ended in failure in not finding the proton resonance.
You asked earlier what proton frequency would occur in your latest build. If we had some way of measuring the permanent magnet field between the pancakes we could give you an accurate answer. My FEMM result for your magnet separation of 21mm is a field very close to 0.1 Tesla that gives a frequency of 4.258 MHz, but FEMM is not a 3D simulator.
Smudge
Smudge,
no problem there, sorry for being impatient, you have enough on your head right now, take your time.
Thanks for the new data, its an approximate which we can work with.
Itsu
Quote from: verpies on 2020.07.20, 14:23:45
Could you do it again but with the the H-probe positioned between the 2 pancake coils (connected in series opposing) ?
Prior to the sweeps, please find the one H-probe orientation and position, which gives you the strongest signal between the coils and mechanically fix the H-probe in that orientation and position for all sweeps. (a photo of this position would be nice to see, too).
Anyway, if you decide to do this, please make these 6 sweeps:
- pancake coils powerd by the PA, through a symmetrizing 1:1 balun and the parallel cap (Smudge's style).
- pancake coils powerd by the PA, and the parallel cap (Smudge's style).
- naked pancake coils powerd by the PA through a symmetrizing 1:1 balun.
- naked pancake coils powerd by the PA.
- pancake coils powered by the PA and a cap in series (Partzman's style).
- pancake coils powered by the PA, through a symmetrizing 1:1 balun and a cap in series (Partzman's style but w/balun).
Use long sweep times so the ringing resonant systems have the time to settle.
Ok, i can do that, do you mean between pancake coils with the toroidal coil removed?
Quote from: Itsu on 2020.07.20, 15:45:45
...do you mean between pancake coils with the toroidal coil removed?
Yes
Quote from: verpies on 2020.07.20, 14:23:45
Prior to the sweeps, please find the one H-probe orientation and position, which gives you the strongest signal between the coils
Because the probe is sensitive not only at its end and maybe also to E-fields, I am concerned about the probe picking up something else than just the magnetic flux spanned by the Litz loop.
Thus, I have to revise my phrase in the sentence above, from "
the strongest signal" to:
"
the largest ratio of the signal amplitude with the Litz loop inside the probe vs. with the Litz loop removed (empty probe)" - for the same probe position and orientation, of course.
This should eliminate any false readings not contributed to by the Litz loop.
Ok, but the problem is that i do not have that much room to manoeuvre as the gap between the pancakes is small.
For maximum pickup i need to have the litz loop on the opposite side / curve of the pancakes, but it won't fit.
Will see what i can do....
Itsu
some test runs of: 1) naked pancake coils powerd by the PA.
2) pancake coils powerd by the PA, through a symmetrizing 1:1 balun and the parallel cap (Smudge's style).
3) pancake coils powerd by the PA, and the parallel cap (Smudge's style).
Position of the H-probe as shown in the above picture.
Be aware, PA can only handle from 100Khz, so sweep is from 100Khz to 50Mhz.
Is this what you expect?
Quote from: Itsu on 2020.07.20, 16:24:53
For maximum pickup i need to have the litz loop on the opposite side / curve of the pancakes, but it won't fit.
It is a silly problem and a serious problem at the same time.
BTW: Don't you get a better signal when the plane of the Litz loop is perpendicular to the planes of the pancake coils ?
Yes, silly and serious, i agree, but with how the H-probe is now, not easily fixed.
The signal is the strongest with the loop as parallel to the pancake coil turns as possible, but due to the distance of the loop from the pancake coil a 45° angle is optimum (60ma) while at 90° (perpendicular and touching) it drops to 52mA.
Also the curves of the loop and the pancake coil turns need to be the same for best pickup.
Measuring the 4 outer turns of the both coils (front left, front right, rear left, read right) shows great difference in signals with front coil left max.
Measuring inside the pancake coils is very limited and only possible the way i showed in the above picture.
I put the H-probe loop inbetween the both front and rear coils, but as they show different signals strengths (on the outsides) i do not know what the correct balanced position is (max i guess, but due to its
bucking positions it might be min. too).
Itsu
Quote from: Itsu on 2020.07.20, 19:28:39
some test runs of: 1) naked pancake coils powerd by the PA.
2) pancake coils powerd by the PA, through a symmetrizing 1:1 balun and the parallel cap (Smudge's style).
3) pancake coils powerd by the PA, and the parallel cap (Smudge's style).
Are these the new coils or old ones ?
Quote from: Itsu on 2020.07.20, 19:28:39
Be aware, PA can only handle from 100Khz, so sweep is from 100Khz to 50Mhz.
That's no problem. Noting interesting happens below 400kHz anyway.
Quote from: Itsu on 2020.07.20, 19:28:39
Is this what you expect?
For the naked coils without balun - yes. In this case it can be seen that the amplitude of the H-fields decreases with frequency until 5.3MHz as it should with inductive reactance. After that, the coil's SRF takes over and peaks at 8.7MHz.
In the sweep without the balun and with caps (Smudge style) the trend is reversed. The H-filed's amplitude
increases almost monotonically up to 8.7MHz as if it was determined by capacitive reactance, but it stays at
low amplitude and before 8.7MHz it does not reach the -10dB level that the naked coils reached (I assume the H-probe has not moved and the amplitudes are comparable).
There is a little disturbance around 600kHz (I'll ignore for now). The LC resonance is
not visible at 4MHz but the coil's SRF is still prominent at 8.7MHz. This is just wrong. The monotonic increase and low amplitude, too.
The sweep with the balun and with the caps clearly shows the LC resonance at 4MHz and coil's SRF at 8.7MHz at high amplitudes, which I expected, but the only difference from the previous case is the presence of the balun. The balun should not make this difference !
I am missing the sweep with naked coils powered by the balun to understand why that balun made such a difference.
Quote from: Itsu on 2020.07.21, 08:18:58
The signal is the strongest with the loop as parallel to the pancake coil turns as possible
The H-field generated by the pancake coils (between them) should be perpendicular to the magnetic field from the magnets
*. That's the whole idea of this opposing pancake coils arrangement !
This pancake H-field should also be perpendicular to the H-field that the toroidal coil would make, if it was powered.
I am very concerned if it is not so.
Quote from: Itsu on 2020.07.21, 08:18:58
Also the curves of the loop and the pancake coil turns need to be the same for best pickup.
I do not understand that. Do you mean, their radii of curvature need to be the same ?
Quote from: Itsu on 2020.07.21, 08:18:58
Measuring the 4 outer turns of the both coils (front left, front right, rear left, read right) shows great difference in signals with front coil left max.
Oops. I do not understand again what front-rear means in this context, as well as left-right. This is probably because my pancake coils are laying on a wooden table and yours are standing up.
Quote from: Itsu on 2020.07.21, 08:18:58
Measuring inside the pancake coils is very limited and only possible the way i showed in the above picture.
So the only solution is to use a smaller sensor or spread them apart.
Quote from: Itsu on 2020.07.21, 08:18:58
I put the H-probe loop inbetween the both front and rear coils
Did you mean to write "front and rear coil" ? The plural suggests four coils in total.
Quote from: Itsu on 2020.07.21, 08:18:58
but as they show different signals strengths (on the outsides) i do not know what the correct balanced position is (max i guess, but due to its bucking positions it might be min. too).
Take a look at Smudge's FEMM simulation. The angles that the flux makes with respect to the planes of the pancake coils, are very important.
Watch out, as far as I remember, his sim was rotated counterintuitively and used current strips. Ask him for clarification if you cannot visualize the pancake wires on his sim.
*
...and parallel to the planes of the pancake coils. Thus the H-probe's loop should encompass the most flux lines between pancake coils, when the plane of the probe's loop is perpendicular to their planes and to their radii.
QuoteAre these the new coils or old ones ?
No, i am still waiting which coil (weave1 or weave2) is best, or did you interpret this sentence of my:
Ok, is one more symmetric (better) then the other? as "one" (as in weave1) is the better one with your "Yes" answer?
QuoteFor the naked coils without balun - yes. In this case it can be seen that the amplitude of the H-fields decreases with frequency until 5.3MHz as it should with inductive reactance. After that, the coil's SRF takes over and peaks at 8.7MHz.
In the sweep without the balun and with caps (Smudge style) the trend is reversed. The H-filed's amplitude increases almost monotonically up to 8.7MHz as if it was determined by capacitive reactance, but it stays at low amplitude and before 8.7MHz it does not reach the -10dB level that the naked coils reached (I assume the H-probe has not moved and the amplitudes are comparable).
There is a little disturbance around 600kHz (I'll ignore for now). The LC resonance is not visible at 4MHz but the coil's SRF is still prominent at 8.7MHz. This is just wrong. The monotonic increase and low amplitude, too.
The sweep with the balun and with the caps clearly shows the LC resonance at 4MHz and coil's SRF at 8.7MHz at high amplitudes, which I expected, but the only difference from the previous case is the presence of the balun. The balun should not make this difference !
I am missing the sweep with naked coils powered by the balun to understand why that balun made such a difference.
I just found out that for the "3) pancake coils powerd by the PA, and the parallel cap (Smudge's style)." test i seem to have a loose soldering connection, so that might be the problem, will redo all tests....
QuoteThe H-field generated by the pancake coils (between them) should be perpendicular to the magnetic field from the magnets. That's the whole idea of this pancake coil arrangement.
This pancake H-field should also be perpendicular to the H-field that the toroidal coil would make, if it was powered.
I am very concerned if it is not so.
well, there is a detectable H field perpendicular to the pancake coils, but the loop picks up a stronger amplitude when parallel or semi parallel.
QuoteI do not understand that. Do you mean, their radii of curvature need to be the same ?
No, see picture, the curve of the loop should match the curve of the pancake coils like thid (( and not like this )(
QuoteOops. I do not understand again what front-rear means in this context, as well as left-right.
see picture
QuoteSo the only solution is to use a smaller sensor or spread them apart.
yes a smaller sensor would keep the pancakes at their present distance.
QuoteDid you mean to write "front and rear coil" ? The plural suggests four coils in total.
yes
QuoteTake a look at Smudge's FEMM simulation. The angles that the flux makes with respect to the planes of the pancake coils, are very important.
Watch out, as far as I remember, his sim was rotated counterintuitively and used current strips. Ask him for clarification if you cannot visualize the pancake wires on his sim.
This below sim picture?
Quote from: Itsu on 2020.07.21, 10:48:06
or did you interpret this sentence of my: Ok, is one more symmetric (better) then the other? as "one" (as in weave1) is the better one with your "Yes" answer?
Yes
Quote from: Itsu on 2020.07.21, 10:48:06
I just found out that for the "3) pancake coils powerd by the PA, and the parallel cap (Smudge's style)." test i seem to have a loose soldering connection, so that might be the problem, will redo all tests....
So if you have the caps out, maybe you could sweep them individually with TG+SA to see what thair self-resonance frequencies are.
Quote from: Itsu on 2020.07.21, 10:48:06
well, there is a detectable H field perpendicular to the pancake coils,
There is a perpendicular field between them ?! I expected the flux between the pancake coils to be parallel to their planes. This also means, that for the best pickup, the plane of the H-probe's loop should be perpendicular
* to this flux.
Quote from: Itsu on 2020.07.21, 10:48:06
No, see picture, the curve of the loop should match the curve of the pancake coils like thid (( and not like this )(
I understand now, but that is not supposed to pick up the flux between the pancake coils the best, because in such orientation, the plane of the H-probe's loop is parallel to the pancake's flux, so almost no flux enters the loop.
Quote from: Itsu on 2020.07.21, 10:48:06
This below sim picture?
Yes
*
...and perpendicular to the planes of the pancake coils.
QuoteThere is a perpendicular field between them ?! I expected the flux between the pancake coils to be parallel to their planes. This also means that the plane of the H-probe's loop, should be perpendicular to this flux for best pickup.
What i mean is that when the H-probe loop is held perpendicular to the pancake coil windings, there is a signal being picked up (H-field), but not as strong when the H-probe loop is parallel to these windings.
Quote from: Itsu on 2020.07.21, 11:05:00
What i mean is that when the H-probe loop is held perpendicular to the pancake coil windings,
Ooff, so that's OK.
The H-probe's loop, which is perpendicular to the pancake coil windings and their radii, is most sensitive to the magnetic flux which is parallel to the pancake coil windings and their radii.
In other words: The H-probe's loop encompasses the most flux lines between pancake coils, when the plane of the probe's loop is
perpendicular to their planes
AND to their radii. I have depicted this optimal position and orientation of the probing loop as a red line on Smudges's diagram attached below.
Any deviation from these optimal angles can be attributed to the fringing of the field when the probing loop is not placed precisely between the two pancake coils. The fringing also happens more when the pancakes are farther apart.
P.S.
I have also added little blue circles to Smudge's sim, to illustrate the crossections of the pancake's winding wires ...and crossections of cylindrical permanent magnets (without their flux)
QuoteYes
ok, will be building weave1 coils then.
QuoteSo if you have the caps out, maybe you could sweep them individually with TG+SA to see what thair self-resonance frequencies are.
Below 3 screenshots of sweeps (0 to 600Mhz) of the input caps:
1) 100pF trimmer cap sweep 0 to 600Mhz using TG and SA
2) 10nF cap sweep 0 to 600Mhz using TG and SA
3) 100pF plus 10nF Smudge style sweep 0 to 600Mhz using TG and SA
Itsu
Quote from: Itsu on 2020.07.21, 15:45:47
1) 100pF trimmer cap sweep 0 to 600Mhz using TG and SA
Is the dip approximately at 500kHz real or an artifact of the SA's display ?
As far as i can see its real, as i saw it on severall frequency ranges.
Here a short video of the H-probe orientation: https://www.youtube.com/watch?v=CJOMrmeu7cQ
Itsu
Quote from: Itsu on 2020.07.21, 16:28:54
As far as i can see its real, as i saw it on severall frequency ranges.
You know that a 100pF capacitor should exhibit a decreasing S21 attenuation as the frequency increases, which is an upward slope on your display. See my S21 sweep of a 100pF cap. So what inductance is causing the S21 attenuation to increase (downward slope) from 0dB
* do -25dB between 0Hz and 500kHz ?
If this was happening at higher frequencies, I would attribute the increasing attenuation to the inductance of the interconnects and cap's leads, but at such low frequency it does not make sense. Especially the cap's low attenuation
* at 0Hz (9kHz).
I treat your TG as Port1 and SA's input as Port2 so the transmittance/attenuation from Port1 to Port2 is S21 in RF parlance (yes, it is backwards).Quote from: Itsu on 2020.07.21, 16:28:54
Here a short video of the H-probe orientation: https://www.youtube.com/watch?v=CJOMrmeu7cQ
OK, that helps me a lot with your orientational terminology.
....but I saw no H-field probing between the coils :(, ...which is where the direction of the H-field matters most.
Perhaps this video will help:
https://youtu.be/2xy3Hm1_ZqI
*
Is the cap's attenuation at 0Hz (9kHz) equal to 0dB or -10dB ? I can't see it on such a wideband sweep. On my sweep, the attenuation of the 100pF cap is -63dB at 9kHz
Nice video, so there are commercial H-field probes.
Concerning the trimmer cap performance, i redid the sweep, but now from 0 to 1Mhz and the dip seems to be around 60Khz, so probably is indeed due to some SA artifact around the beginning of the frequency range.
See screenshot 1 and video
For the H-probe measurement inbetween the coils, there is hardly any room to manoeuvre, see also the video.
Seen your 100pF cap measurement, so here one also from 0 to 50Mhz, see screenshot 2.
Guess my Rigol looses some accuracy when spanning large ranges.
video: https://www.youtube.com/watch?v=2PQu_UjG92Q
Itsu
Quote from: Itsu on 2020.07.21, 19:58:11
Nice video, so there are commercial H-field probes.
Yes, but none that I've seen are made properly, i.e. with Litz center conductor and Litz cloth for the shield of the loop. Because if this, their amplitude vs. frequency relationship is not flat. At just several MHz, the skin effect adds to the resistance and the solid shield gets eddy currents induced in it, which opposes the external magnetic field.
Quote from: Itsu on 2020.07.21, 19:58:11
Concerning the trimmer cap performance, i redid the sweep, but now from 0 to 1Mhz and the dip seems to be around 60Khz, so probably is indeed due to some SA artifact around the beginning of the frequency range.
I hope it is only an artifact.
Does another 100pF cap (a fixed one - not a trimmer) exhibit this artifact, too ?
QuoteDoes another 100pF cap (a fixed one - not a trimmer) exhibit this artifact, too ?
Not the 10nF cap also tested earlier, but will check a fixed 100pF cap.
I redid some PA tests:
1) naked pancake coils powerd by the PA.
2) pancake coils powerd by the PA, through a symmetrizing 1:1 balun and the parallel cap (Smudge's style).
3) pancake coils powerd by the PA, and the parallel cap (Smudge's style).
See screenshots:
Itsu
Quote from: Itsu on 2020.07.21, 19:58:11
Seen your 100pF cap measurement, so here one also from 0 to 50Mhz, see screenshot 2.
Guess my Rigol looses some accuracy when spanning large ranges.
video: https://www.youtube.com/watch?v=2PQu_UjG92Q
Ok, so now everything is clear.
The dip is an artifact of calibration with a test fixture which is inappropriate at these frequencies.
If you do a test fixture like the one I did (or a better one without adapters), then this kHz dip will go away. See the attached photo and zoom into it, to see important details.
Ok, i will try tomorrow.
The fixed 100pF cap shows the same 58Khz dip, so will see tomorrow if it goes away.
Itsu
Quote from: Itsu on 2020.07.21, 20:30:01
I redid some PA tests:
I hope that the H-probe was between the pancake coils (as much as it can be) and did not move between the tests. If it moved - ignore most what I wrote below.
Quote from: Itsu on 2020.07.21, 20:30:01
1) naked pancake coils powerd by the PA.
The two peaks that you have marked are to be expected. One is due to SRF of the coils and the other one is due to their mutual inductance and is described in that e-book. I am attaching an excerpt from it below.
Note, that you are measuring the amplitude of the H-field generated by these pancake coils, and it is a measure that is unencumbered by any parasitic and imperfections of the coils and driving circuit. In other words, this H-field is the field that the protons in the water will be subjected to.
Anyway, look at the amplitude of the H-field at the target frequency of 4MHz. To me it looks like it is -13dB. Notice that this is
the highest amplitude of all of these 3 tests !!!
Quote from: Itsu on 2020.07.21, 20:30:01
2) pancake coils powerd by the PA, through a symmetrizing 1:1 balun and the parallel cap (Smudge's style).
Here you added the balun and parallel caps and the appearance of peak 1 @4MHz reflects it (the other peaks hove shifted slightly by inconsequentially), however the amplitude at that peak is only -34.47dB, that is ~21dB lower than with the naked coils ! 21dB less power means less than 0.001% of the H-field amplitude you had with the naked coils !
Quote from: Itsu on 2020.07.21, 20:30:01
3) pancake coils powerd by the PA, and the parallel cap (Smudge's style).
Here you added the parallel caps only and the appearance of peak 1 @4MHz reflects it (the other peaks hove shifted slightly by inconsequentially),
Notice that this peak at 4MHz is much higher than you had in sweep #2 with the balun, but still lower than with the naked coils.
The amplitude at that peak is only -18.62dB, that is still ~5dB lower than with the naked coils ! 5dB less power means less than 50% of the H-field amplitude you had with the naked coils.
So as these test show now, you get the best H-field amplitude by driving the pancake coils directly without the caps and balun. I wonder why... overvoltaged caps maybe?
BTW: It would be easier to do the calculations if you had the vertical units on the SA set to dBmV.
Quote from: verpies on 2020.07.21, 20:33:42
Ok, so now everything is clear.
The dip is an artifact of calibration with a test fixture which is inappropriate at these frequencies.
If you do a test fixture like the one I did (or a better one without adapters), then this kHz dip will go away. See the attached photo and zoom into it, to see important details.
I made a similar fixture like you used using SMA connectors, see picture, but the result is identical as before, see screenshot.
I have ordered some sma adapters, so will redo this test using other shorter coax to see if that makes some difference.
Itsu
QuoteSo as these test show now, you get the best H-field amplitude by driving the pancake coils directly without the caps and balun. I wonder why... overvoltaged caps maybe?
BTW: It would be easier to do the calculations if you had the vertical units on the SA set to dBmV.
So if i do not use the PA, then we would eliminate the overvoltage on the caps as the cause.
Will try that tonight, both in dBm as dBmV.
Quote from: Itsu on 2020.07.22, 15:02:38
I made a similar fixture like you used using SMA connectors, see picture, but the result is identical as before, see screenshot.
Did you calibrate/normalize by shorting the new fixture ?
Quote from: Itsu on 2020.07.22, 15:09:38
So if i do not use the PA, then we would eliminate the overvoltage on the caps as the cause.
I do not know with certainty that the caps are overvoltaged.
Without the PA the H-field amplitude will be much lower, won't it ?
QuoteDid you calibrate/normalize by shorting the new fixture ?
Yes, as usual
QuoteI do not know with certainty that the caps are overvoltaged.
Without the PA the H-field amplitude will be much lower, won't it ?
Yes, much lower, but hopefully still enough.
Redone the H-probe test in 2 situations without PA in dBmV
1) naked pancake coils powerd by the PA.
2)
3) pancake coils powerd by the PA, and the parallel cap (Smudge's style).
Itsu
I doubled the distance between the both pancake coils so i can insert the H-probe inbetween to take some measurements
This video shows the results: https://www.youtube.com/watch?v=ulA76NW0fT4
Itsu
Quote from: Itsu on 2020.07.22, 15:02:38
I made a similar fixture like you used using SMA connectors, see picture, but the result is identical as before, see screenshot.
WTF is going on ?!
The fixture seems adequate but the 52dBm higher signal power at 0Hz compared to 60kHz is indicative of an inductor - not a capacitor !
-52.16dBmW of power represents -26.08dBmV of voltage ...or 0.2% of the voltage amplitude appearing at 0Hz.
I know that in the case of this SA, 0Hz really is 9kHz, but still...a 99.8% loss of amplitude between 9kHz and 60kHz is huge and not a property of a capacitor !
The frequency range 9kHz-60kHz is within the specs of this SA, so the readings should be accurate.
Maybe the SA is making some weird error or interpolation when the Start frequency is set to 0Hz.
I do not mind if you always set it to 9kHz - the lowest frequency the SA is specified for.
Quote from: Itsu on 2020.07.22, 15:09:38
So if i do not use the PA, then we would eliminate the overvoltage on the caps as the cause.
Not if the caps were permanently damaged by the overvoltage.
Quote from: verpies on 2020.07.23, 11:20:36
WTF is going on ?!
The fixture seems adequate but the the 52dBm higher signal power at 0Hz compared to 60kHz is indicative of an inductor - not a capacitor !
-52.16dBmW of power represents -26.08dBmV of voltage ...or 0.2% of the voltage amplitude appearing at 0Hz.
I know that in the case of this SA, 0Hz really is 9kHz, but still...a 99.8% loss of amplitude between 9kHz and 60kHz is huge and not a property of a capacitor !
The frequency range 9kHz-60kHz is within the specs of this SA, so the readings should be accurate.
Maybe the SA is making some weird error or interpolation when the Start frequency is set to 0Hz.
I do not mind if you always set it to 9kHz - the lowest frequency the SA is specified for.
Let me try again with some shorter (RG174) coax and sma connectors only instead of those bulky coax ones with severall adapters.
I can try to start from 9Khz instead of 0 too.
By the way, tried the weave1 coil setup, but i miserably failed.
I get more appreciation for your coils as these are a pain to build.
I do have litz without any cloth around it and it wants to untwist all the time.
It also has more diameter (1.2mm) as specified (0.93mm), so i came out on 10 turns instead of 14, but i was unable to properly fixate them.
Itsu
Quote from: Itsu on 2020.07.22, 16:02:33
Redone the H-probe test in 3 situations without PA both in dBm as dBmV
First two sweeps are duplicated.
Only the last one is in dBmV (see the screenshots). The file: "balun plus caps no PA dBmV.png" contains the sweep in dBm despite its filename.
Anyway, it is not necessary to post both dBm and dBmV because the difference between these two units is just a factor of 2 (because the power varies as a square of the voltage).
However it is useful to normalize them to 0 dBmW and 0 dBmV and set the reference levels to these levels, for easy comparison. This way 0 dBmW and 0 dBmV simply means "no attenuation" (transmission coefficient = 1)
Quote from: Itsu on 2020.07.23, 11:36:53
I do have litz without any cloth around it and it wants to untwist all the time.
Oh yes, that could make the winding harder.
The silk braid around the strands keeps them from unraveling and the very fine silk fibers seem to aid in holding and polymerizing the Cyanoacrylate on contact. The braid also acts as a separator between turns and decreases the Proximity Effect.
The insulin syringe /w integrated needle (from a pharmacy) is the only applicator that controls the low-viscosity Cyonoacrylate dispension precisely enough. ( BTW: I prefer to dull the needle ).
Quote from: Itsu on 2020.07.23, 11:36:53
It also has more diameter (1.2mm) as specified (0.93mm),
The vendor should have stated that outright. The diameter of my wire is 1mm exactly (including the double silk braid).
Quote from: Itsu on 2020.07.22, 19:38:38
I doubled the distance between the both pancake coils so i can insert the H-probe inbetween to take some measurements
This video shows the results: https://www.youtube.com/watch?v=ulA76NW0fT4
The amplitude vs. orientation & position seems to be correct.
Between my coils, the amplitude in the optimum orientation and position (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35719) is ~20x higher than in others.
P.S.
Where do you switch the Rigol FG amplitude display between V
rms and V
pp ?
QuoteWTF is going on ?!
The fixture seems adequate but the 52dBm higher signal power at 0Hz compared to 60kHz is indicative of an inductor - not a capacitor !
-52.16dBmW of power represents -26.08dBmV of voltage ...or 0.2% of the voltage amplitude appearing at 0Hz.
I know that in the case of this SA, 0Hz really is 9kHz, but still...a 99.8% loss of amplitude between 9kHz and 60kHz is huge and not a property of a capacitor !
The frequency range 9kHz-60kHz is within the specs of this SA, so the readings should be accurate.
Maybe the SA is making some weird error or interpolation when the Start frequency is set to 0Hz.
I do not mind if you always set it to 9kHz - the lowest frequency the SA is specified for.
I used the shorter RG174 and sma connectors from my nanoVNA-F with my fixture, but the problem stays the same.
Even when starting at 9, 10, 50Khz or even 1Mhz (stopping at 50Mhz), we still see that dip, see screenshot.
Perhaps Partzman can confirm as he has the same SA i think.
QuoteFirst two sweeps are duplicated.
Only the last one is in dBmV (see the screenshots). The file: "balun plus caps no PA dBmV.png" contains the sweep in dBm despite its filename.
Anyway, it is not necessary to post both dBm and dBmV because the difference between these two units is just a factor of 2 (because the power varies as a square of the voltage).
However it is useful to normalize them to 0 dBmW and 0 dBmV and set the reference levels to these levels, for easy comparison. This way 0 dBmW and 0 dBmV simply means "no attenuation" (transmission coefficient = 1)
Ok, cleared up those screenshots, only 2 left in dBmV, must have been sleeping :-[
QuoteThe amplitude vs. orientation & position seems to be correct.
Between my coils, the amplitude in the optimum orientation and position is ~20x higher than in others.
P.S.
Where do you switch the Rigol FG amplitude display between Vrms and Vpp ?
So my H-probe orientation is OK? Seems your drawing shows it horizontal, not vertical as i have it.
I am not aware i switch from Vrms and Vpp on the FG, its always on Vpp i think.
Itsu
Quote from: Itsu on 2020.07.23, 15:51:17
Even when starting at 9, 10, 50Khz or even 1Mhz (stopping at 50Mhz), we still see that dip, see screenshot.
Perhaps Partzman can confirm as he has the same SA i think.
Yes, Partzman's input would shed some light on this 9-60kHz anomaly.
Quote from: Itsu on 2020.07.23, 15:51:17
So my H-probe orientation is OK? Seems your drawing shows it horizontal, not vertical as i have it.
That drawing depicts the crossection of the apparatus, so the only visible crossections of the pancake coils are, what you would call, at 12 o'clock and at 6 o'clock.
You video shows you probing at 3 o'clock (or 9 o'clock) position. These crossections of the pancake coils are not visible on Smudge's/my diagram.
Quote from: Itsu on 2020.07.23, 15:51:17
I am not aware i switch from Vrms and Vpp on the FG, its always on Vpp i think.
So it must be fixed in this particular firmware version. In mine it is fixed at V
rms
QuoteThat drawing depicts the crossection of the apparatus, so the only visible crossections of the pancake coils are, what you would call, at 12 o'clock and at 6 o'clock.
You video shows you probing at 3 o'clock (or 9 o'clock) position. These crossections of the pancake coils are not visible on Smudge's/my diagram.
yes, i am measuring in the 3 o'clock position and there is a 6x (3mA v 18mA) amplitude difference in the optimum position, but there is extended distance between the pancakes still.
QuoteSo it must be fixed in this particular firmware version. In mine it is fixed at Vrms
Ok, could be, i am on 00.01.14.
Quote from: Itsu on 2020.07.23, 15:51:17
Perhaps Partzman can confirm as he has the same SA i think.
So my H-probe orientation is OK? Seems your drawing shows it horizontal, not vertical as i have it.
Itsu
Yes, I have the same SA. I see the dip you refer to so did you use the TG or sweep from your FG?
Regarding your H probe, IMO the coil should be parallel to the flat coils not at 90 degrees.
Regards,
Pm
Hi PM,
thanks, i used the TG to sweep.
Concerning the H-probe, so not as i used it in my video?
itsu
Quote from: partzman on 2020.07.23, 19:16:32
IMO the coil should be parallel to the flat coils not at 90 degrees.
Why?
Do you agree, that when the plane of the sensor coil is perpendicular to any magnetic flux, then the sensor coil is most sensitive to the changes in that flux ?
Do you think that the flux generated by the pancake coils is perpendicular to the planes of these pancake coils ?
Finished attempt 2 of a weave 1 pancake coil.
Not very happy still with it, as the problem is to fixate it.
Anyway, its specifications:
OD: 119mm
ID: 87mm
Winding: 1 layer, 10 turns of 1.6mm Litz wire (120 strand), 3.5m long, antiradial weave 1.
Inductance: 17.6µH (@100kHz)
DC resistance: ~0.1Ω
itsu
Quote from: Itsu on 2020.07.23, 20:25:12
Finished attempt 2 of a weave 1 pancake coil.
Not very happy still with it, as the problem is to fixate it.
Don't feel bad. This wire is really thick and the lack of a silk braid makes it difficult to control.
Also notice, that I wound my coil in a 3mm deep
rectangular groove (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35569) in a plexiglass sheet, so I had its help in holding the shape of the coil and squeezing the wires together. I also used double-sticky tape at the bottom of the groove, but my tape was ~1mm thick and pliable. I also helped myself with the insulin syringe with cyanoacrylate in tight spaces. ...and the custom-made
* wire pushing tool (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35664) made the winding almost enjoyable.
*
at home out of a 6mm Teflon rod using a Dremel tool and a cutting wheel. 5min of work.
Quote from: verpies on 2020.07.23, 20:20:08
Why?
Do you agree, that when the plane of the sensor coil is perpendicular to any magnetic flux, then the sensor coil is most sensitive to the changes in that flux ?
Do you think that the flux generated by the pancake coils is perpendicular to the planes of these pancake coils ?
I agree that the position Itsu used for the H sensor in the video is most sensitive to the flux. However, it is my opinion without having built the device, that the H field is perpendicular to the plane of the pancake coil. I am probably wrong on this but it would be interesting to position the H sensor coil parallel to the pancake coils and see what the measurements show.
Regards,
Pm
Quote from: partzman on 2020.07.24, 00:14:06
I agree that the position Itsu used for the H sensor in the video is most sensitive to the flux.
In that particular video - yes.
It is my contention, that this would be the most sensitive orientation in any video.
Quote from: partzman on 2020.07.24, 00:14:06
However, it is my opinion without having built the device, that the H field is perpendicular to the plane of the pancake coil.
Do you realize, that if you are right about the direction of the flux between the pancake coils, then
this simulation (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35289)
* of magnetic flux is wrong and this entire NMR project does not make sense ?
*
The simulation was made in FEMM. It shows the flux between the cross-sections of pancake coils. The outer blue circle is simply the boundary for the FEMM simulation. The pancake coils are modeled as current strips with the current going into or out of the screen, which is a reasonable representation for a single layer of closely spaced wires. Of course FEMM is only a 2D simulation so it is showing the results for current strips that are infinitely long.Quote from: partzman on 2020.07.24, 00:14:06
I am probably wrong on this but it would be interesting to position the H sensor coil parallel to the pancake coils and see what the measurements show.
In the video below, he has positioned the plane of the sensor coil parallel to the planes of the pancake coils:
https://youtu.be/CJOMrmeu7cQ
Unfortunately, in that video, he has positioned the sensor coil only outside of the pancakes, where the flux direction and density are very different than between the pancakes.
Just playing around with my new nanoVNA-F device and sweeping (10Khz to 50Mhz) the 100pF trimmer cap.
This is the gain chart:
Looks similar is yours without any dip at the start.
Itsu
Yes, that looks correct. At 10kHz the cap has -75dB of voltage amplitude attenuation compared to a short. That's a very low transmission coefficient.
I can see that you did not configure the display to connect the dots with lines, yet ;)
Get yourself acquainted with the S11 and S21 scattering parameters and remember that they are RATIOS of voltage amplitudes (between the incident and transmitted or reflected signals). Ratios don't have units !
They are often confused with impedance and admittance of the DUT because they can be calculated from S11 and S21 measurements.
https://www.analogictips.com/what-are-the-application-and-measurements-s-parameters/
It is important to remember that Impedance and Admittance and VSWR are not measured by the VNA directly. They are only calculated from the S11 and S21 measurements. Most VNAs will do that calculation for you on-the-fly automatically when you request it.
However, these calculations are subject to errors, so if you ask the VNA to "measure" the impedance or admittance of the DUT, the VNA will have to calculate them SOMEWHAT IMPRECISELY from the raw S11 and S21 measurements.
If you're wondering what factors effect the precision of these calculations, then read the attached article and watch these videos:
https://youtu.be/1UbEz73FGCU
https://youtu.be/0JglLhrisOs
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35778)
The plot and schematics above are from a good article, which I am attaching to this message as a PDF file.
Quote from: Itsu on 2020.07.23, 19:35:28
Hi PM,
thanks, i used the TG to sweep.
itsu
Itsu,
It might be possible that the ~50kHz dip when using the TG for the sweep generator might be due to the minimum specified frequency for the TG being 100kHz. IOW, starting the sweep at a frequency <100kHz may create who knows what kind of artifacts in hardware or even firmware.
Regards,
Pm
Quote from: partzman on 2020.07.24, 13:58:34
It might be possible that the ~50kHz dip when using the TG for the sweep generator might be due to the minimum specified frequency for the TG being 100kHz. IOW, starting the sweep at a frequency <100kHz may create who knows what kind of artifacts in hardware or even firmware.
I thought that when the SA's spec states 9kHz - 1.5GHz on the front panel then it refers to all its components.
So that needs to kept in mind about this SA. To mitigate this anomaly, in the future start all your sweeps that involve the TG from 100kHz. For this NMR project, nothing interesting is happening below it anyway.
Quote from: verpies on 2020.07.24, 11:45:18
Yes, that looks correct. At 10kHz the cap has -75dB of voltage amplitude attenuation compared to a short. That's a very low transmission coefficient.
I can see that you did not configure the display to connect the dots with lines, yet.
Get yourself acquainted with the S11 and S21 parameters and remember that they are RATIOS of voltage amplitudes (between the incident and transmitted or reflected signals)
They are often confused with impedance and admittance of the DUT because they can be calculated from these S11 and S21 measurements.
https://www.analogictips.com/what-are-the-application-and-measurements-s-parameters/
Also, in case you INTENTIONALLY want to calculate the impedance or admittance of the DUT from the S11 and S21 measurements (most VNAs do it for you automatically when you request it) and you're wondering whether to use the S11 (one port) or S21 (two port measurement) and series or parallel connection to the DUT to accomplish that, then watch these videos:
https://youtu.be/1UbEz73FGCU
https://youtu.be/0JglLhrisOs
The plot and schematics above are from a good article, which I am attaching to this message as a PDF file.
P.S.
Remember that the firmware for your VNA is updated often.
its a completely different world i am seeing, so all the info is welcome, thanks.
Did some changes to the graphs, like adding lines between points etc., much better.
Itsu
Quote from: partzman on 2020.07.24, 13:58:34
Itsu,
It might be possible that the ~50kHz dip when using the TG for the sweep generator might be due to the minimum specified frequency for the TG being 100kHz. IOW, starting the sweep at a frequency <100kHz may create who knows what kind of artifacts in hardware or even firmware.
Regards,
Pm
PM, yes, looks like it, below screenshot when started from 100Khz O0
Where did you learned that (100Khz)?
Itsu
Here a nanoVNA measurement of my new weave 1 coil (series measurement) with severall graphs.
Need to dive into the info from verpies to get a good feel on what i see.
EDIT, did a normalisation at the coil...
Itsu
Quote from: Itsu on 2020.07.24, 15:33:23
PM, yes, looks like it, below screenshot when started from 100Khz O0
Where did you learned that (100Khz)?
Itsu
From the DSA800 User Guide. I think it is downloadable from the Rigol site but I can send you a copy if you can't find it. It was specified way down in section 5-2 in the manual.
regards,
Pm
Thanks, found it:
Itsu
Quote from: Itsu on 2020.07.24, 16:15:42
Here a nanoVNA measurement of my new weave 1 coil (series measurement) with severall graphs.
The Smith chart is spiraling because the real part of the impedance is increasing with frequency due to skin effect, radiation resistance, proximity effect. This is normal.
I am puzzled why your S21 Gain (which is just the magnitude of the S21) shows values above 0dB. Positive dB values mean amplification. Passive components always attenuate (in the ideal case the conduct the signal without losses) so the S21 measurement should always yield negative dB values for passive components. If you were measuring an amplifier, then the S21 magnitudes would be positive (in dB).
Quote from: Itsu on 2020.07.24, 16:15:42
...did a normalisation at the coil...
In VNA world they always call it "calibration".
There are several partial calibrations and two full ones: SOLT and TOSM.
I see what you mean by the amplification, must have something to do with calibration.
I now did it NOT at the nanoVNA device itself, but at the SMA fixture.
Will redo it a few times.....
Quote from: Itsu on 2020.07.24, 17:24:52
I now did it NOT at the nanoVNA device itself, but at the SMA fixture.
You should always calibrate for the
Measurement Planes (https://youtu.be/xLGSblc-Fec).
Errors can also be caused by passive components with LOOONNG settling times (e.g. ringing LC networks). The remedy is "Measurement delay" or long sweep times.
FWIW, here are some test results of measurements between two parallel flat pancake coils that are connected in a bucking mode that is, the flux on the inside of each coil is going in the same direction as depicted in the drawing below.
Each coil measures 65.2uH and the aid inductance is 170.6uH and the buck inductance is 89.5uH so the calculated k = .31 . These values are confirmed by Lbuck = ((1-k)*2)*(L1*L2)^1/2 = 89.9uH and
Laid = (1-((1-k)/2))*4*(L1*L2)^1/2 = 170.82uH.
The first pix and scope shot shows the position of the sense coil parallel to the flat coil with the resulting current measurement of 40.34ma rms at resonance.
The second pix and scope shot show the position of the sense coil 90 degrees to the flat coil with a current measurement of 1.235ma rms and is slightly off resonance.
My conclusion, which of course is subject to correction, is that the H field is on axis with the axis of the flat coils or at a right angle to the coil plane while the coil flux is bucking and in the same direction across the face
of each coil.
I am in no position to judge whether or not this will affect the operation of the device as intended.
Regards,
Pm
Hi Pm,
You can't have flux and H field at right angles to each other. Your flux is correct. Flux is the B field (density) and the H field follows the same lines. If your measurement is telling you otherwise then the measurement is suspect.
Smudge
Quote from: verpies on 2020.07.24, 17:41:01
You should always calibrate for the Measurement Planes (https://youtu.be/xLGSblc-Fec).
Errors can also be caused by passive components with LOOONNG settling times (e.g. ringing LC networks). The remedy is "Measurement delay" or long sweep times.
Ok, yes i tried that, but overdid it i think.
Now i used the calibration at the end of the S11 / S21 coax's, so close to the dut.
Below the result which is more accurate i think.
Quote from: partzman on 2020.07.24, 18:42:31
FWIW, here are some test results of measurements between two parallel flat pancake coils that are connected in a bucking mode
Are these coils connected electrically in parallel or serial "bucking mode" ?
Quote from: partzman on 2020.07.24, 18:42:31
that is, the flux on the inside of each coil is going in the same direction as depicted in the drawing below.
But the flux produced by one coil is called the H-flux. The amount of flux lines penetrating a given area of a perpendicular 2D plane is the magnetic flux density - often just called the H-field or B-field
* after interaction with a ferromagnetic.
The direction of the H-field is not parallel to the plane that serves to measure its density - it is perpendicular.
The vectors denoting the flux and flux density are parallel by definition so there is something wrong with your drawing.
Please, draw the direction of the electric current flowing in these coils and how they are connected (I cannot read that from the photo).
Also, please mark the printed side of the PCBs, so I can see from the edge-on view how they are oriented to each other. I assume that these two PCB coils are identical and single-sided. (pls confirm).
*
In equations, the letter "B" is often used to denote magnetic flux density, regardless of the origin of that flux (i.e. coil, magnet).Quote from: partzman on 2020.07.24, 18:42:31
Each coil measures 65.2uH and the aid inductance is 170.6uH and the buck inductance is 89.5uH so the calculated k = .31 . These values are confirmed by Lbuck = ((1-k)*2)*(L1*L2)^1/2 = 89.9uH and
Laid = (1-((1-k)/2))*4*(L1*L2)^1/2 = 170.82uH.
That makes sense but I still do not know how these coils are connected electrically.
Quote from: partzman on 2020.07.24, 18:42:31
The first pix and scope shot shows the position of the sense coil parallel to the flat coil with the resulting current measurement of 40.34ma rms at resonance.
The second pix and scope shot show the position of the sense coil 90 degrees to the flat coil with a current measurement of 1.235ma rms and is slightly off resonance.
Please decrease the frequency an order of magnitude, so the measurement is not deceived by capacitive coupling.
Quote from: partzman on 2020.07.24, 18:42:31
My conclusion, which of course is subject to correction, is that the H field is on axis with the axis of the flat coils or at a right angle to the coil plane while the coil flux is bucking
It's an interesting challenge, then !
Quote from: partzman on 2020.07.24, 18:42:31
[H-field] ...and in the same direction across the face of each coil.
Measured when the coils are together or separate ?
Also, in this context, does "across the face" mean parallel to the plane of the coil ?
Quote from: Smudge on 2020.07.24, 19:07:29
Hi Pm,
You can't have flux and H field at right angles to each other. Your flux is correct. Flux is the B field (density) and the H field follows the same lines. If your measurement is telling you otherwise then the measurement is suspect.
Smudge
Hi Smudge,
OK I understand but my measurements are what is shown. I will perhaps wind some new coils and run more tests to determine exactly what field is producing the parallel sense coil currents.
Regards,
Pm
Quote from: Itsu on 2020.07.24, 19:20:10
Below the result which is more accurate i think.
Yes it is much better now.
To make it easier to read, please keep the reference line at 0dB and at:
- the bottom of the plot for real values of S21 and Impedance and Admittance in linear units.
- the middle of the plot for all imaginary values
- the top of the plot for S21 real values and magnitudes in dB units (your "S21 Gain" plot already has the reference line positioned correctly).
Quote from: verpies on 2020.07.24, 20:09:08
Are these coils connected electrically in parallel or serial "bucking mode" ?
They are electrically connected in a serial bucking mode.
Quote
But the flux produced by one coil is called the H-flux. The amount of flux lines penetrating a given area of a perpendicular 2D plane is the magnetic flux density - often just called the H-field.
The direction of the H-field is not parallel to the plane that serves to measure its density - it is perpendicular.
The vectors denoting the flux and flux density are parallel by definition so there is something wrong with your drawing.
My depiction of the coil 'flux' may be incorrect but the coils are in a bucking mode and there is a measurement of some field creating current in the parallel sense coil.
Quote
Please, draw the direction of the electric current flowing in these coils and how they are connected (I cannot read that from the photo).
Also, please mark the printed side of the PCBs, so I can see from the edge-on view how they are oriented to each other. I assume that these two PCBs are identical and single-sided. (pls confirm).
The pcbs are double sided and the coil traces from one side to the other are exactly on top of each other so there would be maximum coupling and capacitance between each side in the original application. I used only the inside set of coils for this test with the outside coils left open. In rechecking my physical coil positions, I see that I complicated the situation by having the inside coils wound in opposite directions. This should not make any difference in the overall outcome as I checked for electrical buck and aid connections and used the buck connections for the test. However, I will rerun the test with the inside coils running the same direction as they face each other and then the electrical connections will be easy to see and analyze.
Quote
That makes sense but I still do not know how these coils are connected electrically.
Please decrease the frequency an order of magnitude, so the measurement is not deceived by capacitive coupling.
It's an interesting challenge, then !
Yes, I will do this in the test rerun.
Quote
Measured when the coils are together or separate ?
All coil measurements were taken with the coils positioned as shown with a spacing of 21.9mm.
Quote
Also, in this context, does "across the face" mean parallel to the plane of the coil ?
Yes.
Regards,
Pm
Quote from: partzman on 2020.07.24, 20:59:06
My depiction of the coil 'flux' may be incorrect but the coils are in a bucking mode and there is a measurement of some field creating current in the parallel sense coil.
I see only four possibilities that can explain this measurement:
- The coils and their interconnects have large unbalanced radial currents.
- Your probe is also sensitive to E-fields or to H-fields outside of the sensing loop.
- You made a mistake with coils' orientation/polarity.
- Smudge, Biot, Savart and I are wrong.
OK, this is a retest with new pcbs that each have both coils on each side connected in parallel with the winding direction identical. The pcb coil pairs are connected in a buck mode as can be seen in the photos below. The outside winding points are connected together and each inside winding point is used for the input and common respectively. The series connected bucking flat coils are operating at their respective SRF with the probe and loop inserted. The overall sensitivity is very high.
The first photo is of the parallel single turn sense winding as before and the first scope pix is the current measured.
With no change in the bucking flat coils, the second photo is an ~80T single layer coil positioned as shown or midway between the upper coil traces which somewhat represents the NMR application in discussion. The second and third scope pix show the open circuit voltage and the current respectively of this coil.
I have done lower frequency measurements as well and will follow up with those later on.
Regards,
Pm
Edit: I don't mean to derail or otherwise impede progress here so you may ignore any of my last posts if so desired.
Quote from: partzman on 2020.07.24, 23:51:26
Edit: I don't mean to derail or otherwise impede progress here so you may ignore any of my last posts if so desired.
No, no. Your input is great. A sound principle cannot be derailed...and if it can - it is not sound.
It helps us identify any unwanted effects. For example radial currents or currents flowing through the intrawinding capacitance.
Your 80T coil resembles a fragment of a toroidal coil. Although it does not curve and does not contain its flux entirely inside, it gives us insight what the toroidal coil is subjected to.
Whatever you probe feels, will be felt by the other component of the system, too, e.g. protons.
What is the difference between the single sensing loop (in the i-probe) placed parallel vs. perpendicular to the planes of the coils ?
Quote from: partzman on 2020.07.24, 23:51:26
The first photo is of the parallel single turn sense winding as before and the first scope pix is the current measured.
In that photo, the probing loop is placed at the 12 o'clock position (in coil's polar coordinates). Is there a difference in the sensed amplitude, when you place it at the 3 o'clock position away from the interconnects ?
Please move the pink wire to the 9 o'clock position for this measurement.
Quote from: Itsu on 2020.07.24, 19:20:10
Now i used the calibration at the end of the S11 / S21 coax's, so close to the dut.
Look how similar your S21 plot is to mine:
I put all 3 traces of S21 on one plot, in order to have only one screenshot to post.
Yes, thats amazing O0
Itsu
These are plots of my existing 2 pancake coils (none litz) only, in bucking mode (series measurement).
I see a resonance point at 12.3Mhz (red marker 1) in the S21 gain (Magnitude) plot.
I also see in the S21 real/imaginary plot the red marker 1 at the zero crossing meaning Xc and Xl are canceling (resonance).
But why do i see in the S11 Z plot the green 2 and blue 3 markers NOT at resonance?
Should in parallel resonance Impedance not be max. at resonance? Why 2 peaks?
Itsu
Quote from: Itsu on 2020.07.25, 09:43:18
I see a resonance point at 12.3Mhz (red marker 1) in the S21 gain (Magnitude) plot.
I also see in the S21 real/imaginary plot the red marker 1 at the zero crossing meaning Xc and Xl are canceling (resonance).
Yes
Quote from: Itsu on 2020.07.25, 09:43:18
But why do i see in the S11 Z plot the green 2 and blue 3 markers NOT at resonance?
Should in parallel resonance Impedance not be max. at resonance? Why 2 peaks?
Because of
this (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35778). You have 3.7kΩ/div there ! That's ~100x over the limit of usability!
Remember that S11 │Z│ is a calculated parameter - it is not measured by the VNA directly (it is calculated from the raw S11 measurement
* in this case).
That calculation has a 10% error at only 220Ω. Extrapolate that error to 2.2kΩ ...and above :o
Also, radiation resistance is a factor at the dip between markers.
*
Take a look at the raw │S11│ measurement in [dB] ( this is the "magnitude of S11" but your VNA proboably calls it "S11 Gain (dB)" ) and you will see what the VNA sees at Port 1. The S11 │Z│ is calculated from that...
The details of this calculation are in this PDF (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35828).
QuoteBecause of this. You have 3.7kΩ/div there ! That's ~100x over the limit of usability!
But those settings (3.7kΩ/div) cannot be changed, they are adjusted according to the presented min / max. data.
QuoteRemember that S11 │Z│ is a calculated parameter - it is not measured by the VNA directly (it is calculated from the raw S11 measurement* in this case).
That calculation has a 10% error at only 220Ω. Extrapolate that to 2.2kΩ ...and above :o
Also, radiation resistance is a factor at the dip between markers.
* Take a look at the raw │S11│ measurement in [dB] (your VNA proboably calls it "S11 Gain (dB)") and you will see what the VNA sees at Port 1. The S11 │Z│ is calculated from that...
The details of this calculation are in this PDF.
Below screenshot is the S11/S21 (no separate S11) gain (LogMag) plot (left bottom).
S11 is indeed kind of "none telling"
By the way, that PDF is hard to read as its kind of mingled.
Guess i have to start reading from this (highlighted) Configuration 3 part:
Itsu
Quote from: Itsu on 2020.07.25, 11:31:55
But those settings (3.7kΩ/div) cannot be changed, they are adjusted according to the presented min / max. data.
It is not a matter of setting the Ω/div scaling on the screen. It is a matter of having such high impedances in the plot that require such scaling. These impedances are the result of the S11 -> S11│Z│ conversion.
Quote from: Itsu on 2020.07.25, 11:31:55
By the way, that PDF is hard to read as it's kind of mangled.
Maybe this forum's code is mangling attachments again.
Remind me what your email or FTP server is, so I can send it to you directly. This file is small enough for an email attachment.
Quote from: Itsu on 2020.07.25, 11:31:55
Below screenshot is the S11/S21 (no separate S11) gain (LogMag) plot (left bottom).
S11 is indeed kind of "none telling"
That dip at the lowest frequency (10kHz ?) causes the autoscaling to obscure details at higher frequencies.
Try to get the │S11│measurement displayed with a separate scaling or in a separate plot. There are details to be seen in it.
OK, i put up S11 (right top) and adjusted the scaling to be more narrow (max 1.1 min 0.9)
EDIT, added a second plot with the S21 /S11 LogMag (dB) scale adjusted (+ - 0.5dB).
Below is a Linear plot of the │S21│ transmission/attenuation with phase information superimposed. Both are displayed Linearly.
It can be seen, that from the start (9kHz) to the 1st marker, the transmission of the signal THRU the coil decreases. This is typical of an inductor that blocks high frequencies.
However, from the 1st marker to the 2nd marker, the transmission THRU the coil increases - like through a capacitor :o
This is evidenced by the phase of the signal transmitted THRU the coil between the Start (9kHz) and the 1st marker, where the received voltage (which is proportional to Rx current) lags behind the applied voltage (like in an inductor) - this manifests itself as being below the 0º reference line (it never quite reaches the textbook -90º, though).
From the 1st marker to the 2nd marker, the received voltage (proportional to Rx current) leads the applied voltage (like in a capacitor).
It is important to remember that the current transmitted THRU our coil (modeled as a parallel LC circuit) is not the same as the reactive current circulating IN it.
The question you should be asking yourself is WHAT PATH the current takes between the 1st and the 2nd marker, and what magnetic field is generated by this current* ?
* BTW: This is the reason why I asked Partzman to decrease his test frequency.
Quote from: verpies on 2020.07.25, 14:12:52
Below is a plot of the │S21│ transmission/attenuation with phase information superimposed.
It can be seen, that from the start (9kHz) to the 1st marker, the transmission of the signal THRU the coil decreases. This is typical of an inductor that blocks high frequencies.
However, from the 1st marker to the 2nd marker, the transmission THRU the coil increases - like through a capacitor :o
This is evidenced by the phase of the signal transmitted THRU the coil between the Start (9kHz) and the 1st marker, where the received voltage (which is proportional to Rx current) lags behind the applied voltage (like in an inductor) - this manifests itself as being below the 0º reference line (it never reaches the textbook -90º, though).
From the 1st marker to the 2nd marker, the received voltage (proportional to Rx current) leads the applied voltage (like in an capacitor).
The question you should be asking yourself is WHAT PATH the current takes between the 1st and the 2nd marker, and what magnetic field is generated by this current*.
* BTW: This is the reason why I asked Partzman to decrease his test frequency.
Well, if this a naked coil, than that path should be the intrawiring capacitance (parallel capacitance).
If any magnetic field is generated by this current i doubt it as thats normally done by inductance.
Quote from: Itsu on 2020.07.25, 14:27:23
Well, if this a naked coil, than that path should be the intrawiring capacitance (parallel capacitance).
Yes but the current flowing thru the coil (modeled as a parallel LC circuit) is not the same as the current circulating in that coil (especially at resonance).
Quote from: Itsu on 2020.07.25, 14:27:23
If any magnetic field is generated by this current i doubt it as that's normally done by inductance.
Actually, ANY current generates a magnetic field.
The current flowing through a capacitor, too ...albeit in a different direction.
The next question you should be asking is what is the ratio of the current flowing in the coil's wires vs. current flowing in the intrawinding capacitance ? Before SRF, at SRF, after SRF ...
well, if current flowing in a capacitor generates a magnetic field but in a different direction, then before
SRF we have 100% current flowing in the coil's wires, at SRF 50% in each because of these different directions and
after SRF 100% current flowing in the intrawinding capacitance.
This test is done with a large rectangular sense coil as seen in the pix below. It matches the mean shape of the pcb coil but is slightly off center due to the bobbin spacers.
The first scope pix shows the resultant sense current at resonance and the second scope pix shows sense current at a lower frequency or off resonance.
As Verpies suggested, it must be a capacitive coupling to the sense winding and current probe that is creating the current in the loop. Otherwise, there should be little to no current in the sense loop with the bucking coil arrangement as shown but it would be interesting to know the exact interaction going on here.
It might be worth saying that in my experiments with transmission lines and 90 degree placed coils, the toroid winding in the NMR device will appear as a single turn to the flat coils under the right conditions.
Regards,
Pm
Quote from: Itsu on 2020.07.23, 20:25:12
Finished attempt 2 of a weave 1 pancake coil.
Not very happy still with it, as the problem is to fixate it.
Anyway, its specifications:
OD: 119mm
ID: 87mm
Winding: 1 layer, 10 turns of 1.6mm Litz wire (120 strand), 3.5m long, antiradial weave 1.
Inductance: 17.6µH (@100kHz)
DC resistance: ~0.1Ω
itsu
Finished my 2th weave1 pancake coil.
its specifications:
OD: 119mm
ID: 87mm
Winding: 1 layer, 10 turns of 1.6mm Litz wire (120 strand), 3.5m long, antiradial weave 1.
Inductance: 17.8µH (@100kHz)
DC resistance: ~0.1Ω
When putting them in series (no matter how aiding or bucking), i have 34.7uH.
Itsu
Quote from: partzman on 2020.07.25, 16:32:57
It might be worth saying that in my experiments with transmission lines and 90 degree placed coils, the toroid winding in the NMR device will appear as a single turn to the flat coils under the right conditions.
You're right about this, but we have a secret weapon to kill this effect.
Namely, winding the toroidal coil using
the same scheme (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35639) as the new pancake coils.
Quote from: Itsu on 2020.07.25, 16:33:30
When putting them in series (no matter how aiding or bucking), i have 34.7uH.
My coils do not behave like that.
My combined inductance readings, when measuring them with an LCR meter @ 100kHz, are:
- 40µH when connected in series, far apart and their planes are at right angles.
- 20µH when connected in series, 10mm apart, coaxial, planes parallel and magnetically Opposing.
- 60µH when connected in series, 10mm apart, coaxial, planes parallel and magnetically Aiding.
Ok, doing it that way i have:
My combined inductance readings, when measuring them with an LCR meter @ 100kHz, are:
- 34.7µH when connected in series, far apart and their planes are at right angles.
- 14.4µH when connected in series, 10mm apart, coaxial, planes parallel and magnetically Opposing.
- 55µH when connected in series, 10mm apart, coaxial, planes parallel and magnetically Aiding.
Itsu
When putting together the toroidal coil with the new pancake coils i measure 18.7uH for the opposing pancake coils.
Toroidal coil measured earlier: 242 turns of 1mm wire, measuring 31uH @ 100Khz and 0.3 Ohm.
The below diagram shows the new dimensions.
Itsu
Quote from: Itsu on 2020.07.25, 19:10:45
Toroidal coil measured earlier: 242 turns of 1mm wire, measuring 31uH @ 100Khz and 0.3 Ohm.
How many layers does the toroidal coil have?
How is it wound?
1 layer, spaced closely together (inside touching, outside some space)
But i see a problem with the pancake coils inductance, the selfresonance is already below 4Mhz, so if we need 4.2Mhz as NMR frequency thats not going to work.
Quote from: Itsu on 2020.07.25, 19:34:27
1 layer, spaced closely together (inside touching, outside some space)
What weave ?
Quote from: Itsu on 2020.07.25, 19:34:27
But i see a problem with the pancake coils inductance, the selfresonance is already below 4Mhz, so if we need 4.2Mhz as NMR frequency that's not going to work.
I'll get back to that.
Quote from: Itsu on 2020.07.25, 15:03:53
well, if current flowing in a capacitor generates a magnetic field but in a different direction, then before
SRF we have 100% current flowing in the coil's wires, at SRF 50% in each because of these different directions and
after SRF 100% current flowing in the intrawinding capacitance.
I agree that at resonance, the currents in both branches of an ideal parallel LC circuit are equal and that means 50% current ratio.
However, keep in mind that the current "transmitted" THROUGH this coil
is not the same as the reactive current circulating IN that coil
*.
This is because of the phenomenon known as "Current Magnification" in parallel LCR circuits
**.
The ratio of the current circulating IN the coil vs. the current "transmitted" THROUGH the coil is simply known as the Quality Factor of the coil ...or just Q.
*
You once asked me: "Where does the energy of this circulating current come from? " Answer: From the past, when the power supply/generator was first switched on. It was during that initial moment, when the circulating current has been built up.**
...and "Voltage Magnification" in series LCR circuits.
Quote from: verpies on 2020.07.26, 01:48:19
What wave ?
I'll get back to that.
See picture:
Quote from: Itsu on 2020.07.26, 09:12:16
See picture:
So with this weave, the following Partzman's comment applies:
Quote from: partzman on 2020.07.25, 16:32:57
...the toroid winding in the NMR device will appear as a single turn to the flat coils under the right conditions.
That "under the right conditions" sounds like that it is a good thing that the toroid winding in the NMR device will appear as a single turn to the flat coils, but i guess its not.
So we need a weave 1 like scheme for the toriodal coil too to kill that effect.
I have received the water tube .STL file from Conrad, so looking for a 3D printer to print some tubes.
Itsu
Some pictures from the setup using the new weave 1 pancake coils.
Quote from: Itsu on 2020.07.26, 11:56:45
That "under the right conditions" sounds like that it is a good thing that the toroid winding in the NMR device will appear as a single turn to the flat coils, but i guess its not.
It isn't. This resultant single turn forms a loop that will pick up the signal from the pancake coils - not from the innards of the toroidal coil.
I suggest a simple experiment with a circular loop of wire
* of the same diameter as the mean diameter of the water tube (100mm) and placed between the pancake coils in place of the water tube and toroidal coil.
Scope how much voltage that loop picks up when the pancake cols are driven as they are supposed to be. That induced voltage signal will compete with the NMR signal :(
*
If you are very ambitious, use a piece of coaxial cable to make this loop. This way you can experiment how much difference the shield of the coax makes, when it is used as the E-field shield - just like in that video (https://youtu.be/2xy3Hm1_ZqI?t=377) ...but bigger.Quote from: Itsu on 2020.07.26, 11:56:45
So we need a weave 1 like scheme for the toriodal coil too to kill that effect.
Yes, the weave1 scheme can be flattened to a single layer solenoidal coil or single layer toroidal coil. The wire crossings should be at the ID and OD as not to force the increase of the distance between pancakes and magnets.
In other applications where the symmetry and thickness of the winding is not critical (e.g. in transformers, chokes...) a two layer winding coming back on itself would be an option.
Quote from: Itsu on 2020.07.26, 12:22:38
Some pictures from the setup using the new weave 1 pancake coils.
I think Smudge is eagerly waiting for a video of NMR hunting with this setup.
I really want him to see the crosstalk and the tuned LC circuits ringing "louder" and longer than the protons ...at the same frequency >:-)
Because of this, I don't think any NMR signal can be discerned without further coil optimization and E-field shielding, but seeing these problems with his own eyes might elicit some useful feedback from him.
Well, I did not intend to post any more on this subject but after sleeping on it overnight, I realized what was really going. So I re-assembled the test set and cleaned it up a bit as seen in the pix below. The difference now is that there are two sense coils, one is a 58T solenoid and the other is the 1T rectangular both positioned as shown.
The schematic is next showing the fields as I propose them to exist. There are two simultaneous H fields present according to the experimental results IMO.
The last pix is the scope shot of the resulting waveforms. CH1(yel) is the input from the 50 ohm source Rigol SG, CH2(blu) is the ocv across the solenoid winding and CH4(grn) is the current in the 1T rectangular winding.
The H1 field is produced by the conventional current flow in L1 and L2 and with these coils in a buck configuration, an open circuit voltage of 13.11v rms is produced in the solenoid winding. L1 and L2 are 33T each.
The measured L1 buck L2 = 83.6uH and the measured L1 to L2 inter-winding capacitance = 7.0-7.5pfd assembled. The resonant frequency is 6.388MHz so the required capacitance to series resonate with 83.6uH would
be C = 1/w^2*L = 7.4pfd which is close to the measured value.
So, H2 is therefore produced by the series resonance of L1buckL2 and the inter-winding capacitance Ciw with the H2 field flowing between L1 and L2 as shown in the schematic.
If this is correct, there may be possibilities for some interesting field interactions.
Also, in the present proposed NMR configuration, the H2 field will exist when L1 and L2 are bucking and tuned for resonance IMO.
Regards,
Pm
Quote from: partzman on 2020.07.26, 16:03:57
The schematic is next showing the fields as I propose them to exist. There are two simultaneous H fields present according to the experimental results IMO.
Yes, a current flowing through capacitance will generate a second H-field.
There is capacitance between the turns of one coil (intrawinding capacitance
*) as well as capacitance between the coils themselves (interwinding capacitance). All of them generate their own H-fields.
Could you amend your drawing to depict the entire electric circuit, which is made by the current flowing through that interwinding capacitance ?
*
If there weren't, a single coil would not exhibit a self-resonance frequency.
Quote from: partzman on 2020.07.26, 16:03:57
The schematic is next showing the fields as I propose them to exist. There are two simultaneous H fields present according to the experimental results IMO.
If you denoted the H2 field as flowing into or out of the screen then that would fit with the field from displacement current through the capacitor, and would be detected by the solenoid coil. The single turn shorted coil looks to be slightly tilted from vertical, and that would then see some through flux from the H1 field. Does tilting that coil either side of vertical cause the induced current to go through zero?
Smudge
Quote from: verpies on 2020.07.26, 16:23:09
Yes, a current flowing through capacitance will generate a second H-field.
There is capacitance between the turns of one coil (intrawinding capacitance*) as well as capacitance between the coils themselves (interwinding capacitance). All of them generate their own H-fields.
Could you amend your drawing to depict the entire electric circuit, which is made by the current flowing through that interwinding capacitance ?
* If there weren't, a single coil would not exhibit a self-resonance frequency.
You mean like this?
Pm
Quote from: partzman on 2020.07.26, 17:02:47
You mean like this?
No, no. Complete the Ciw1 electric path all the way to signal generator.
These interwinding capacitances you've added, are correct, btw.
Quote from: Smudge on 2020.07.26, 16:41:36
If you denoted the H2 field as flowing into or out of the screen then that would fit with the field from displacement current through the capacitor, and would be detected by the solenoid coil.
Yes, the magnetic flux should be perpendicular to the current loop's plane. That's why I asked Partzman to draw that current loop including Ciw1 and SigGen.
What do you think of this current's path and the direction of the magnetic flux generated by it, after the toroidal coil is surrounded by a grounded Litz cloth ?
Quote from: Smudge on 2020.07.26, 16:41:36
If you denoted the H2 field as flowing into or out of the screen then that would fit with the field from displacement current through the capacitor, and would be detected by the solenoid coil.
I'm not sure I understand you here but I will try to figure it out with perhaps more testing.
Quote
The single turn shorted coil looks to be slightly tilted from vertical, and that would then see some through flux from the H1 field. Does tilting that coil either side of vertical cause the induced current to go through zero?
Smudge
No, the current reduces slightly due to a slight change in frequency due to the very small change in parasitic capacitance.
Pm
OK, let's take one issue at a time to make sure I understand correctly.
Is my simplified depiction of H1 in my previous drawings correct for the cross section of a set of bucking coils spaced as shown?
Pm
Quote from: partzman on 2020.07.26, 18:00:28
Is my simplified depiction of H1 in my previous drawings correct for the cross section of a set of bucking coils spaced as shown?
IMO the direction of the H1 on the last drawing is correct.
Quote from: verpies on 2020.07.26, 18:02:54
IMO the direction of the H1 on the last drawing is correct.
OK, then I think I have the H field sources reversed and incorrect. As Smudge pointed out, it is the H field from the inter-winding capacitance Ciw that creates that induces the solenoid coil and that field should be coming out of or going into the page and with that I agree. IOW, each flat coil is a capacitor plate and the H field would be in the same plane as the flat coils.
Then, the H field generated by the flat coils induces the single turn sense coil and that field is like I had depicted in the drawing. So, the cross H2 field vector arrows are wrong along with the logic.
However, as I remember some of the 1T loop tests, something doesn't yet add up but I don't want to take anymore time away from the NMR project so I'll do some more testing and when I think I have conclusive evidence one way or the other, I'll post the results.
Pm
This is the schematic of the bucking coils cross section that I believe is now correct. I will not post any more on this unless prompted.
Regards,
Pm
Quote from: partzman on 2020.07.26, 20:06:39
This is the schematic of the bucking coils cross section that I believe is now correct. I will not post any more on this unless prompted.
Yes, the direction of H1 & H2 appears correct now.
What about H3 field generated by the Cit ?
I just realized that I never posted any plots of my naked bucking coils connected in series (coaxial & parallel). These plots are attached below with different spacings between the coils indicated in the filename. The │S21│ plots have linear vertical scales, starting from 1 at the top. The phase is in degrees (depicted linearly on the vertical axis).
Please remember that just because the coil strongly attenuates the signal being transmitted THRU it ( manifested as low │S21│ ), does not mean that the reactive current circulating IN it* is small.
In fact quite the opposite can be true because of "Current Magnification"* determined by the Q of the coil.
* as in a parallel LC network.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35901)
All (animated).gif
Quote from: verpies on 2020.07.26, 14:39:40
It isn't. This resultant single turn forms a loop that will pick up the signal from the pancake coils - not from the innards of the toroidal coil.
I suggest a simple experiment with a circular loop of wire* of the same diameter as the mean diameter of the water tube (100mm) and placed between the pancake coils in place of the water tube and toroidal coil.
Scope how much voltage that loop picks up when the pancake cols are driven as they are supposed to be. That induced voltage signal will compete with the NMR signal :(
* If you are very ambitious, use a piece of coaxial cable to make this loop. This way you can experiment how much difference the shield of the coax makes, when it is used as the E-field shield - just like in that video (https://youtu.be/2xy3Hm1_ZqI?t=377) ...but bigger.
Yes, the weave1 scheme can be flattened to a single layer solenoidal coil or single layer toroidal coil. The wire crossings should be at the ID and OD as not to force the increase of the distance between pancakes and magnets.
In other applications where the symmetry and thickness of the winding is not critical (e.g. in transformers, chokes...) a two layer winding coming back on itself would be an option.
I used a single loop and the special prepared coax loop (see above video) for some measurements, but it seems that there is hardly any difference in signal pickup between those 2 loops (both 10Vpp at resonance).
I must confess that i had to make 3 of those coax (RG58) loops as the first 2 seem to work very erratically (signal / no signal, depending how to hold the loop), so it can be that there is something wrong, but as far as i can see (and measure) this last one should be OK.
Video here: https://www.youtube.com/watch?v=7lsNtqDSnFo
2 problems noted:
# self resonance frequency very (to) low 3.3Mhz for NMR
# if coax loop is OK, then very much crosstalk is present
Itsu
Quote from: Itsu on 2020.07.27, 08:37:52
I used a single loop and the special prepared coax loop (see above video) for some measurements, but it seems that there is hardly any difference in signal pickup between those 2 loops (both 10Vpp at resonance).
The lack of difference means that the capacitive coupling does not play a great role IN THAT FIELD ORIENTATION.
Quote from: Itsu on 2020.07.27, 08:37:52
I must confess that i had to make 3 of those coax (RG58) loops as the first 2 seem to work very erratically (signal / no signal, depending how to hold the loop), so it can be that there is something wrong, but as far as i can see (and measure) this last one should be OK.
Yes, the erratic performance could be due to the construction of the loop (the aluminum shield does not solder well and is prone to bad contacts and shorts), however there is another explanation which should be investigated further.
Namely, does the axial* movement of the sensing loop cause the received signal to go through zero amplitude?
On the video, I saw you making only the radial move with the sensing loop. I love that glass cube, btw !
It is important that the sensing loop is very planar (flat) - its shadow should be a straight line. The planes of the 3 coils must be parallel all the time. A 2x thinner wire would allow for more selective axial positioning, too (e.g. the same wire that the toroidal coil is made with).
*
"axial" as in moving along the axis of the pancake coils and the sensing loop, ...in other words: initially close to one pancake, next exactly between pancakes and finally close to the other pancake. Quote from: Itsu on 2020.07.27, 08:37:52
2 problems noted:
# self resonance frequency very (to) low 3.3Mhz for NMR
Yes, and that means that achieving a high amplitude H1 field at a higher frequency will require a lot of driving power. Paradoxically, the high circulating current I
C at resonance is accompanied by low current draw from the generator I
G. This is caused by high "Current Magnification" or the ratio I
C / I
G also known as Q. Remember, that it is the high I
C which is responsible for generating the high amplitude of the desired H1 field - not the I
G. This helpful effect does not occur as much outside of resonance.
Also, above the self-resonance frequency, more current flows in the intrawinding capacitance (inter-turn capacitance or "Cit" in Partzman's parlance) than in the turns of the coil.
Furthermore, the interwinding current increases through the interwinding capacitance ("Ciw" in Partzman's parlance).
Currents flowing through the Cit and Ciw generate magnetic fields H
2 and H
3, respectively, which are oriented in a different direction than the desired field H
1. Smudge please comment on this.
Quote from: Itsu on 2020.07.27, 08:37:52
# if coax loop is OK, then very much crosstalk is present
It was the entire idea of this experiment to demonstrate this.
This means the the toroidal coil should be wound with a special scheme (analogous to weave1 or weave2), that will cancel its circumferential induction.
P.S.
The Cit for one of my coils is 175pF (measured at 100kHz with an LCR meter).
The Ciw for a pair of my coils is 19.7pF at 2mm, 8.4pF at 9mm, 5.8pF @ 16mm, 4.3pF at 23mm (all measured at 100kHz when coaxial and parallel)
My detailed coil dimensions and other data are here (https://www.overunityresearch.com/index.php?topic=3924.msg83054#msg83054).
Quote from: verpies on 2020.07.26, 23:11:05
Yes, the direction of H1 & H2 appears correct now.
What about H3 field generated by the Cit ?
Here is my depiction of the Cit H Field. This is maybe not easy to visualize because with the flat wound coil, Cit is a spiral and the H Field completion is at each end with the lead outs. Also, the S flow which is not shown is entering the capacitance area from each side due to EXH .
Regards,
Pm
QuoteYes, the erratic performance could be due to the construction of the loop (the aluminum shield does not solder well and is prone to bad contacts and shorts), however there is another explanation which should be investigated further.
Namely, does the axial* movement of the sensing loop cause the received signal to go through zero amplitude?
On the video, I saw you making only the radial move with the sensing loop. I love that glass cube, btw !
It is important that the sensing loop is very planar (flat) - its shadow should be a straight line. The planes of the 3 coils must be parallel all the time. A 2x thinner wire would allow for more selective axial positioning, too (e.g. the same wire that the toroidal coil is made with).
* "axial" as in moving along the axis of the pancake coils and the sensing loop, ...in other words: initially close to one pancake, next exactly between pancakes and finally close to the other pancake.
I used a 1mm magnet wire as loop now and made some axial movements which showed NO "received signal to go through zero amplitude".
There is a dip in the middle from 10Vpp with loop against either (rear / front) coil and 8Vpp in the middle.
Another observation is the sweep from 1Khz to 10Mhz with this loop, which shows a high amplitude signal BEFORE resonance when the is loop against the front pancake coil and a high amplitude signal AFTER resonance when the loop is against the rear pancake coil, see screenshots.
Quote# self resonance frequency very (to) low 3.3Mhz for NMR
This means that i have to reduce the number of turns on the pancake coils to get to NMR resonance (~4.2Mhz) which causes that it does not overlap on the toroidal coil completly.
Itsu
Quote from: Itsu on 2020.07.27, 14:00:25
This means that i have to reduce the number of turns on the pancake coils to get to NMR resonance (~4.2Mhz) which causes that it does not overlap on the toroidal coil completly.
...or make pancake coils with a double parallel windings.
...or increase the spacings between the turns to reduce Cit.
...or both.
Notice that my coils have a higher SRF. This is due to their slightly different diameter or larger spacing between turns provided by the silk braid,...or both.
Quote from: Itsu on 2020.07.27, 14:00:25
I used a 1mm magnet wire as loop now and made some axial movements which showed NO "received signal to go through zero amplitude".
There is a dip in the middle from 10Vpp with loop against either (rear / front) coil and 8Vpp in the middle.
I expected that dip. Perhaps it is only -2V deep because the sensing loop is not ideally parallel to the pancake coils....or the capacitive coupling makes up for the remaining 8V of amplitude, after all.
Quote from: Itsu on 2020.07.27, 14:00:25
Another observation is the sweep from 1Khz to 10Mhz with this loop, which shows a high amplitude signal BEFORE resonance when the is loop against the front pancake coil and a high amplitude signal AFTER resonance when the loop is against the rear pancake coil, see screenshots.
This would suggest a capacitive coupling to pancake coils which are not driven symmetrically with respect to ground. Does the loop made out of coaxial wire also manifest this variation with the same movement ?
What if you connect only one end
* of that 1mm wire sensing loop to the tip of the scope probe and short the probe's tip to ground with a 10kΩ resistor (that can be done on the probe itself) ?
This will treat the entire sensing loop as a plate of a capacitor.
How much voltage do you see when getting closer and further away from the pancake coil which is driven asymmetrically wrt ground ?
*
hold the other end in the air with tape to keep its circular shape but without shorting the loop
Here is my take on the H2 and H3 fields, see images below. Note that I show the Ciw displacement current going across the center of the assembly, and really it depends on where the Ciw capacitance resides, going from the "hot" end of one coil to the "cold" end of the other and that could be a current that is not horizontal, and that would alter the H3 field pattern.
Smudge
@Smudge
I see.
How would you model the CIT and the associated displacement current iIT in this type (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35637) of pancake coil ?
...and the HIT* field generated by such iIT ?
* the HIT moniker describes the same field as H3 in previous messages.
Quote from: verpies on 2020.07.27, 02:00:11
I just realized that I never posted any plots of my naked bucking coils connected in series (coaxial & parallel). These plots are attached below with different spacings between the coils indicated in the filename. The │S21│ plots have linear vertical scales, starting from 1 at the top. The phase is in degrees (depicted linearly on the vertical axis).
Please remember that just because the coil strongly attenuates the signal being transmitted THRU it ( manifested as low │S21│ ), does not mean that the reactive current circulating IN it* is small.
In fact quite the opposite can be true because of "Current Magnification"* determined by the Q of the coil.
* as in a parallel LC network.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35901)
All (animated).gif
Nice animated gif from those plots, shows nicely what the between distance does.
Below a plot of my new pancake coils at 16mm apart with similar settings as you:
Blue S21 LinMag, Purple S21 phase (22.5°/div).
Taken from the nanoVNA device as the PC app (nanoVNA saver) does not allow Linear settings.
EDIT: just found out that the PC app does show |S21| as linear, see right bottom. S21 Phase is right top.
Itsu
QuoteThis would suggest a capacitive coupling to pancake coils which are not driven symmetrically with respect to ground. Does the loop made out of coaxial wire also manifest this variation with the same movement ?
Yes it does but less.
QuoteWhat if you connect only one end* of that 1mm wire sensing loop to the tip of the scope probe and short the probe's tip to ground with a 10kΩ resistor (that can be done on the probe itself) ?
This will treat the entire sensing loop as a plate of a capacitor.
How much voltage do you see when getting closer and further away from the pancake coil which is driven asymmetrically wrt ground ?
* hold the other end in the air with tape to keep its circular shape but without shorting the loop
The front pancake coil (driven with red FG lead) gives 17Vpp on the 10K terminated loop when touching, the rear pancake coil (connected to the black FG lead) gives 22Vpp when touching.
Itsu
Quote from: Itsu on 2020.07.27, 16:23:40
The front pancake coil (driven with red FG lead) gives 17Vpp on the 10K terminated loop when touching, the rear pancake coil (connected to the black FG lead) gives 22Vpp when touching.
How much difference to that test does the connecting of the scope's ground to FG's ground (at coil's terminal) make ?
So without the 10K but with scope and FG grounds tight together its: front 20Vpp, rear 27Vpp.
Quote from: Smudge on 2020.07.27, 14:40:56
Here is my take on the H2 and H3 fields, see images below. Note that I show the Ciw displacement current going across the center of the assembly, and really it depends on where the Ciw capacitance resides, going from the "hot" end of one coil to the "cold" end of the other and that could be a current that is not horizontal, and that would alter the H3 field pattern.
Smudge
Smudge,
I'm trying to understand what you're explaining so I will focus mainly on the H Field creating the induction in the solenoid coil. Do I understand correctly that your H3 Field creates the induction in the offset solenoid coil?
I originally thought that the H3 Field you show going across the center of the assembly was the cause of the solenoid induction but it didn't fit the experimental results IIRC. I now view the Ciw as two circular plates parallel to one another which should exhibit a single direction H Field filling the cross section of those plates.
If I understand your view, I should be able to position two solenoid coils, one on top of the other in the cross sectional area and should see opposite polarities on their output, correct?
Regards,
Pm
Quote from: Itsu on 2020.07.27, 18:57:18
So without the 10K but with scope and FG grounds tied together its: front 20Vpp, rear 27Vpp.
No,no, with the scope and FG grounds tied together AND with the 10kΩ resistor in place (from tip to ground).
Also, since you i-probe is does not require a galvanic connection to the sensing loop, what happens when you short this loop and put the i-probe on it (thus making a giant isolated H-probe) ?
Does the amplitude of the signal sensed this way go through zero as you move the loop axially ?
QuoteNo,no, with the scope and FG grounds tied together AND with the 10kΩ resistor in place (from tip to ground).
ok, i thought so, but as nothing changed then, i went for the removal of the 10K.
So nothing changes when connecting scope and FG grounds together and 10K between scope tip and ground (front 17Vpp, rear 23Vpp)
QuoteAlso, since you i-probe is does not require a galvanic connection to the sensing loop, what happens when you short this loop and put the i-probe on it (thus making a giant isolated H-probe) ?
Does the amplitude of the signal sensed this way go through zero as you move the loop axially ?
Using the H-probe is again kind of cumbersome as it does not fit inbetween the pancake coils, but with some fiddling and increasing the distance, it looks like
the front pancake (red lead FG) gives more signal then the rear pancake (black FG lead) without any dip or peak in the middle (linear).
Itsu
Quote from: Itsu on 2020.07.27, 20:24:30
,...it looks like the front pancake (red lead FG) gives more signal then the rear pancake (black FG lead) without any dip or peak in the middle (linear).
WTF?! The H-sensing loop is galvanically isolated from the scope's probe completely. Also, it should not couple preferentially to one of the pancakes and the sensed H-field's amplitude should cross zero as it is moved axially.
Smudge, just consider the flux lines that penetrate the sensing loop's plane.
Quote from: Itsu on 2020.07.27, 20:24:30
Using the H-probe is again kind of cumbersome as it does not fit in between the pancake coils
Oh, sorry - I forgot about this silly/serious problem.
Perhaps you can twist an extension in that sensing loop wire like on the attached diagram (I drew the twist with different colors but it can be made out of the same wire).
Quote from: Itsu on 2020.07.27, 15:45:25
Nice animated gif from those plots, shows nicely what the between distance does.
If you liked it, here is another one.
It depicts the S21 │Z│ and Phase of one naked coil on a linear scale, as the terminating resistance (in the middle break) varies form 0Ω to 200Ω.
Notice, that the impedance magnitudes at Peak#1 & Peak#3
decrease as the coil's terminating resistance at the middle break
increases. C.C
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35941)
When you consider the │Z│ ratio of (Peak#1 - Dip#2) / (Peak#3 - Dip#2) you can see that at 0Ω termination, the ratio of these peaks is 4.8 and at 200Ω termination the ratio of these peaks is 10.7.
This means that the │Z│ Peak#3 at 30MHz is most likely due to reflection since it decreases ~2x faster
* than Peak#1 as the coil's termination at the middle break is matched better and better.
*
relative to Dip#2
Quote from: verpies on 2020.07.27, 21:22:15
WTF?! The H-sensing loop is galvanically isolated from the scope's probe completely. Also, it should not couple preferentially to one of the pancakes and the sensed H-field's amplitude should cross zero as it is moved axially.
Smudge, just consider the flux lines that penetrate the sensing loop's plane.
Oh, sorry - I forgot about this silly/serious problem.
Perhaps you can twist an extension in that sensing loop wire like on the attached diagram (I drew the twist with different colors but it can be made out of the same wire).
That picture is exactly as i tried it first and it gave the result i mentioned.
Lateron i widened the distance between the pancakes and tried the i-Probe on the main loop with similar results.
Let me focus on the first method again for more accurate measurements.
Itsu
Quote from: verpies on 2020.07.28, 01:36:04
If you liked it, here is another one.
It depicts the S21 │Z│ and Phase of one naked coil on a linear scale, as the terminating resistance (in the middle break) varies form 0Ω to 200Ω.
Notice, that the impedance magnitudes at Peak#1 & Peak#3 decrease as the coil's terminating resistance at the middle break increases. C.C
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35941)
You can see that the │Z│ Peak#3 at 30MHz is due to reflection since it decreases significantly as the coil's termination at the middle break is matched better and better, while the │Z│ Peak#1 stays relatively constant at 4.7MHz, which makes it look like it is related more to the parallel LC self-resonance of the coil (i.e. L*Cit resonance), ...rather than to the transmission-line like reflections.
Nice, so it goes from a high Q reactive environment to a low Q resistive one.
Quote from: partzman on 2020.07.27, 19:31:38
I'm trying to understand what you're explaining so I will focus mainly on the H Field creating the induction in the solenoid coil. Do I understand correctly that your H3 Field creates the induction in the offset solenoid coil?
Yes, but note that the images only depict half of the pancake coil. To represent the full pancake coil there should be a mirror image below it. In the NMR experiment the two images would be well separated, in your pancake coil the images would be almost touching.
QuoteI originally thought that the H3 Field you show going across the center of the assembly was the cause of the solenoid induction but it didn't fit the experimental results IIRC. I now view the Ciw as two circular plates parallel to one another which should exhibit a single direction H Field filling the cross section of those plates.
There is something wrong in that analogy. If the pancake coils were driven by separate isolated current sources and there was a potential difference between those sources then your analogy would apply. But that is not the case, the coils are connected and driven from a single source. To establish the Ciw electric field you must examine where the potential difference arises (Verpies showed this in a post but I can't find it now). Clearly there is a potential difference between each coil's start (let's call this the hot end) and finish (let's call this the cold end) connections, and also from each coil's hot end to the other coil's cold end. Where these occur in space depend upon the winding directions, if each coil is CW or CCW then the series bucking connection puts the hot ends together (say on the outer radius) and the two cold ends together (on the inner radius). That minimises Ciw.
QuoteIf I understand your view, I should be able to position two solenoid coils, one on top of the other in the cross sectional area and should see opposite polarities on their output, correct?
Yes, but note the answer to your first question, the image shows only (say the top) half of the pancake coil. Thus a small solenoid that is moved from top to bottom of the pancake coil could exhibit one reversal of polarity as it traverses the top half then another as it moves across the center and another as it traverses the bottom half. But for your pancake coil the middle one might be seen as just a dip in magnitude and you would get just one reversal from top to bottom
Smudge
Quote from: verpies on 2020.07.27, 14:54:49
@Smudge
I see.
How would you model the CIT and the associated displacement current iIT in this type (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35637) of pancake coil ?
...and the HIT* field generated by such iIT ?
* the HIT moniker describes the same field as H3 in previous messages.
Then my E field or displacement current arrows would have alternate directions and their H fields would cancel at distances from the surface greater than the turn spacings. Clearly this would almost eliminate that H
IT field at the toroidal coil.
Smudge
Quote from: Itsu on 2020.07.28, 08:06:48
Nice, so it goes from a high Q reactive environment to a low Q resistive one.
Yes, but │Z│ Peak#3 is affected ~2x more than │Z│ Peak#1.
When you consider the │Z│ ratio of (Peak#1 - Dip#2) / (Peak#3 - Dip#2) you can see that at 0Ω termination, the ratio of these peaks is 4.8 and at 200Ω termination the ratio of these peaks is 10.7.
This means that the │Z│ Peak#3 at 30MHz is most likely due to reflection since it decreases ~2x faster
* than Peak#1 as the coil's termination
** is matched better and better (from 0Ω to 200Ω).
*
relative to Dip#2**
at the middle break
Quote from: Itsu on 2020.07.28, 07:57:31
Let me focus on the first method again for more accurate measurements.
Remember that warning which is printed on your i-probe.
Put a tape or heat shrink tubing on that part of the sensing loop which touches the i-probe. We want to maintain electric isolation and keep the capacitive coupling between i-probe and the sensing loop, to the minimum. The thin enamel on the 1mm magnet wire is not much of a barrier.
Quote from: Smudge on 2020.07.28, 08:53:54
Clearly there is a potential difference between each coil's start (let's call this the hot end) and finish (let's call this the cold end) connections,...
Fortunately, Itsu's and my pancake coils, do not have the "hot end" and "cold end" anymore.
Quote from: Smudge on 2020.07.28, 08:53:54
...,and also from each coil's hot end to the other coil's cold end.
However we still have this problem. Especially when the "front pancake" and "back pancake" are not driven symmetrically with respect to ground.
This is less of the problem then the previous one, though, because C
IW is an order of magnitude smaller that C
IT. Anyway, electric field between pancakes and the associated displacement current can by shunted to ground by a suitable conductive shield, but this shield cannot support eddy currents or it will attenuate the desirable high-frequency H1 field. ...and this means Litz cloth (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35877).
Nonetheless, I still don't see why an
isolated loop (shown edge-on in red below) placed exactly between the bucking pancakes and parallel to them, would have any current induced in it.
Quote from: verpies on 2020.07.28, 11:46:42
Remember that warning which is printed on your i-probe.
Put a tape or heat shrink tubing on that part of the sensing loop which touches the i-probe. We want to maintain electric isolation and keep the capacitive coupling between i-probe and the sensing loop, to the minimum. The thin enamel on the 1mm magnet wire is not much of a barrier.
Using the below diagram as how i measure.
I have a 11cm wide pvc pipe in the middle where the 1mm single loop wire is put around, so we only can have
the axial movement between the front and rear pancake coil.
There is a minimum (going through zero??) point, but its not in the middle between the 2 pancakes.
Its closer to the front pancake coil no matter how the FG red lead is connected (there is a strength difference though).
So we have an asymmetric field being picked up.
Video here: https://www.youtube.com/watch?v=iy4itwUW-iI
Itsu
Quote from: Smudge on 2020.07.28, 08:53:54
If the pancake coils were driven by separate isolated current sources and there was a potential difference between those sources then your analogy would apply. But that is not the case,
This sentence of yours gave me an idea.
Would you agree to alter your design and connect the pancake coils in parallel ? Still bucking, of course.
This way, both of them would be at the same electric potential and the displacement current flowing through C
IW would become non-issue.
The phase coherence would have to be maintained by keeping the interconnects equal length after the parallel T-splitter, but that is easy.
Quote from: verpies on 2020.07.28, 15:47:35
This sentence of yours gave me an idea.
Would you agree to alter your design and connect the pancake coils in parallel ? Still bucking, of course.
This way, both of them would be at the same electric potential and the displacement current flowing through CIW would become non-issue.
The phase coherence would have to be maintained by keeping the interconnects equal length after after the parallel T-splitter, but that is easy.
Of course I agree, and with the attention being paid to making both coils as identical as possible the two currents should be virtually identical. Will the inductance change be of benefit?
Smudge
Quote from: Itsu on 2020.07.28, 15:25:39
So we have an asymmetric field being picked up.
Video here: https://www.youtube.com/watch?v=iy4itwUW-iI
We sure do!
The asymmetry should not be so great when the generator is isolated and the sensing loop is isolated from the i-probe.
That is almost an order of magnitude greater induced current next to the back pancake compared to the front pancake.
P.S.
Warning - nit picking follows: The small loop around the i-probe should have the smallest area possible. The large sensing loop should be as flat as possible and parallel to the pancakes.
..but I doubt these three imperfections could cause such a large asymmetry.
Quote from: Smudge on 2020.07.28, 16:12:56
Will the inductance change be of benefit?
Yes, Itsu would love to have a higher self-resonance frequency.
...but I am afraid the parallel conected C
IT will compensate for the smaller inductance.
Also, smaller inductance means lower impedance at higher frequencies and that is good.
Quote from: Smudge on 2020.07.28, 08:53:54
Yes, but note that the images only depict half of the pancake coil. To represent the full pancake coil there should be a mirror image below it. In the NMR experiment the two images would be well separated, in your pancake coil the images would be almost touching.
There is something wrong in that analogy. If the pancake coils were driven by separate isolated current sources and there was a potential difference between those sources then your analogy would apply. But that is not the case, the coils are connected and driven from a single source. To establish the Ciw electric field you must examine where the potential difference arises (Verpies showed this in a post but I can't find it now). Clearly there is a potential difference between each coil's start (let's call this the hot end) and finish (let's call this the cold end) connections, and also from each coil's hot end to the other coil's cold end. Where these occur in space depend upon the winding directions, if each coil is CW or CCW then the series bucking connection puts the hot ends together (say on the outer radius) and the two cold ends together (on the inner radius). That minimises Ciw.
Yes, but note the answer to your first question, the image shows only (say the top) half of the pancake coil. Thus a small solenoid that is moved from top to bottom of the pancake coil could exhibit one reversal of polarity as it traverses the top half then another as it moves across the center and another as it traverses the bottom half. But for your pancake coil the middle one might be seen as just a dip in magnitude and you would get just one reversal from top to bottom
Smudge
Smudge,
I respectively have to disagree with your above analysis based on the experimental results below. I had worked up my proposal in schematic form but it doesn't exactly fit the test results when the flat coils are driven from the inner leads but I will post it anyway for reference. It does seem to fit however when the coils are driven from the outside leads so more analysis is needed here. Perhaps the phasing of the H1 Field is affecting the inner drive results.
Anyway, the first pix is of the test setup which consists of two identical wound solenoid coils placed on top of one another and centered as close as possible in the upper pcb foil traces. The buck inductance measured 57.3uH and the Ciw measured 35pf at low signal levels with this assembly.
The first scope pix is with the inside leads of the bucking coils driven by the SG. As can be seen, the open circuit voltages are in-phase but the inner solenoid coil CH2(blu) is considerably lower in amplitude than the outer solenoid coil CH3(pnk).
The second scope pix is with the outside leads of the bucking coils driven by the SG.
Here the voltages oc are near equal in amplitude and phase which is what I expected to see for the inner drive as well. Both drive conditions above were tuned for resonance and in off-resonance tuning, the phase and amplitude of the solenoid oc voltages would change.
Regards,
Pm
Edit: I have replaced both inner and outer drive scope pix as the previous outer drive data did not have grounds connected on the solenoid coils. Now we see the situation is reversed as the inner solenoid CH2(blu) is higher in amplitude that the outer CH3(pnk). The length of the pcb traces get shorter from the outside to the inside so basically the inner solenoid should not have as much influence from the directional H2 Field as the outer solenoid IMO.
Quote from: verpies on 2020.07.28, 16:15:24
We sure do!
The asymmetry should not be so great when the generator is isolated and the sensing loop is isolated from the i-probe.
That is almost an order of magnitude greater induced current next to the back pancake compared to the front pancake.
P.S.
Warning - nit picking follows: The small loop around the i-probe should have the smallest area possible. The large sensing loop should be as flat as possible and parallel to the pancakes.
..but I doubt these three imperfections could cause such a large asymmetry.
I perfected the big and small loops so the small loop is similar in size as the H-probe loop used earlier and the big one is as flat as can be.
But no noticable difference was seen in the asymmetry shown earlier.
Itsu
OK, upon more examination, I questioned why there was no differential when the ground was missing on the previous test so I ran another with a 1K load resistor placed across the outside solenoid coil and measured the differential voltage with two probes. There was no differential voltage at any frequency as far as I could tell!
So, this means there is no H2 Field at least as I had proposed so I will end this posting of these results on this thread.
Sorry for all the distraction. :-[
Regards,
Pm
I have made preliminary tests of my pancake coils connected in parallel. Their magnetic fields are opposing.
When they are 2mm apart (coaxial and parallel) their combined inductance is 2.6µH, at 8mm: 4.9µH, at 16mm: 6.3µH, at 23mm: 7.2µH and when they are far apart and perpendicular their combined inductance is 10µH. These inductances were measured with an LCR meter at 100kHz.
Below is their S21 │Z│ and Phase plot from 9kHz to 50MHz at different separation distances:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35966)
...and a zoomed-in plot from 3MHz to 12MHz at different separation distances:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35968)
Notice, that in this configuration the coils exhibit very high Q.
The Q increases as the separation distance increases.
The self-resonance frequency decreases as the separation distance increases.
The phase of the signal transmitted THRU the coils stays at the textbook +90º (inductive) right up to the self-resonance peak, where it makes an abrupt transition to -88º (capacitive). This is a very clean phase response because these coils act very inductively almost up to their self-resonance frequency.
P.S.
My interconnect to the coils after the Tee are not equal length (one is 4cm longer). This is because I did not have short SMA pigtails of equal length today.
Quote from: partzman on 2020.07.28, 18:22:40
Anyway, the first pix is of the test setup which consists of two identical wound solenoid coils
I'd like to notice that these coils have an odd number of layers and are susceptible to longitudinal induction. They will form a 1 turn loop with the test leads.
Quote from: Itsu on 2020.07.28, 20:31:47
I perfected the big and small loops so the small loop is similar in size as the H-probe loop used earlier and the big one is as flat as can be.
But no noticeable difference was seen in the asymmetry shown earlier.
This is not good!
I am starting to suspect, there is something wrong with one of your coils, i.e. superficially tinned ends of the Litz wires, or a short somewhere .
...or some kind of disturbance from your workbench - try rotating the entire apparatus or moving it to a different place.
Quote from: verpies on 2020.07.29, 01:44:41
This is not good!
I am starting to suspect, there is something wrong with one of your coils, i.e. superficially tinned ends of the Litz wires, or a short somewhere .
...or some kind of disturbance from your workbench - try rotating the entire apparatus or moving it to a different place.
I remeasured the both pancake coils and both are solid around 17.8uH @ 100Khz.
I retinned the litz connections and moved the setup 90° and to a different place.
I reversed the i-probe, but the offset to center of the dip (going through zero?) stays.
I do notice that when in dip position, the series connection between the pancake coils is very sensitive to touch and squeezing.
The harder i squeeze the litz, the greater the signal amplitude becomes.
Will look into paralleling the coils later today.
Itsu
Quote from: Itsu on 2020.07.29, 09:16:59
I do notice that when in dip position, the series connection between the pancake coils is very sensitive to touch and squeezing.
The harder i squeeze the litz, the greater the signal amplitude becomes.
Even if you squeeze it with a plastic tool ?
No, then it stays in its dip
So if it is very sensitive only to your hands then it means that this junction is very "hot", i.e. it has a high voltage amplitude with respect to ground.
Does it remain "hot" if you solder this junction instead of using the connector with screws?
Does a 1:1 symmetryzing balun decrease this sensitivity ?
A quick test using my old pancake coils with the similar setup as above, shows a much higher resonance (10Mhz), but also shows NOT this asymmetric behaviour in that there is a linear tapering off of the amplitude going from FG red lead connected coil (max) to FG black lead connected coil (min).
No dip in the center nor close to either pancake.
Will do your suggested tests later today, have to run now.
Itsu
Quote from: Smudge on 2020.07.28, 16:12:56
Of course I agree, and with the attention being paid to making both coils as identical as possible the two currents should be virtually identical.
@Itsu
Keeping the quote above in mind, it would be a good idea to compare these two coils with your VNA.
This can be done with these two measurements from 100kHz-50MHz and 100kHz-10MHz (in total, 4 measurements per coil):
- │S21│ also known as "S21 Magnitude", but I think your VNA calls it "S21 Gain", with the Reference Value = 1U and set to be displayed at the top of the plot.
- S21 Phase, with the Reference Value = 0º and set to be displayed in the middle of the plot.
The │S21│and Phase can be on the same plot.
These measurements should be done using a good RF test fixture soldered to the coil, similar to this one:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35739)
Of course, do a TOSM or SOLT calibration first with the RF test fixture attached and the coil disconnected. Later solder the fixture to the coil and keep it away from any conductive stuff, when measuring it.
Quote from: verpies on 2020.07.29, 10:45:46
So if it is very sensitive only to your hands then it means that this junction is very "hot", i.e. it has a high voltage amplitude with respect to ground.
Does it remain "hot" if you solder this junction instead of using the connector with screws?
Does a 1:1 symmetryzing balun decrease this sensitivity ?
It remains "hot" when soldering the series connection and no change there when using a 1:1 balun.
Will do your suggested VNA tests later, what does reference value "1U" mean?
Itsu
Quote from: Itsu on 2020.07.29, 16:04:38
What does reference value "1U" mean?
I assume you already know what a "reference value" is in SA and VNA.
The "1U" means "1 unit" This is because the │S21│ is a ratio of 2 voltages
* and a ratio is dimensionless.
For passive devices the │S21│ ratio will vary between 1 and 0. ( 1U meaning full THRU transmission and 0U - full THRU attenuation, this way 500mU would mean 50% attenuation of the voltage amplitude ). These are linear measures.
You can also set the display to dB. For passive devices the │S21│ratio will vary between 0 dB and -∞ dB. ( 0dB meaning full THRU transmission and -∞dB meaning full THRU attenuation, this way -6dB would mean 50% attenuation of the voltage amplitude ). These are logarithmic measures.
*
the voltage amplitude of the incident signal (at Port1) and the received signed (at Port2). Your VNA might number its ports A & B instead of 1 & 2.
Itsu,
When it is convenient for you, I'm curious if you get any output from your toroid coil when positioned properly in between the driven bucking flat coils when using just two scope probes for measurement? IOW, no grounds connected to the toroid coil so the two scope probes would then be taking a differential measurement of any output voltage across the toroidal coil.
I see no output in my setup when there is no ground reference for the solenoid coil no matter what the position of the coil in relation to the flat coils!
Regards,
Pm
Edit: The only output is an equal voltage on both coil leads but this is from the E Field of the flat coils and is not differential between the coil leads.
Quote from: verpies on 2020.07.29, 14:48:03
@Itsu
Keeping the quote above in mind, it would be a good idea to compare these two coils with your VNA.
This can be done with these two measurements from 100kHz-50MHz and 100kHz-10MHz (in total, 4 measurements per coil):
- │S21│ also known as "S21 Magnitude", but I think your VNA calls it "S21 Gain", with the Reference Value = 1U and set to be displayed at the top of the plot.
- S21 Phase, with the Reference Value = 0º and set to be displayed in the middle of the plot.
The │S21│and Phase can be on the same plot.
These measurements should be done using a good RF test fixture soldered to the coil, similar to this one:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35739)
Of course, do a TOSM or SOLT calibration first with the RF test fixture attached and the coil disconnected. Later solder the fixture to the coil and keep it away from any conductive stuff, when measuring it.
Made some measurements and included some in 1 screenshot, so i have 4 screenshots total.
100Khz to 50Mhz from each coil (front and rear) including 3 measurements in each screenshot (actually 4 with smith chart).
100Khz to 10Mhz from each coil (front and rear) including 3 measurements in each screenshot (actually 4 with smith chart).
the 3 measurements in each screenshot are:
S21 (linear gain as i understand)
S21 Gain (Log gain)
S21 phase
1st screenshot front coil 100Khz - 50Mhz
2th screenshot rear coil 100Khz - 50Mhz
3th screenshot front coil 100Khz - 10Mhz
4th screenshot rear coil 100Khz - 10Mhz
Phase plot was adjusted to + / - 90°
Itsu
Quote from: partzman on 2020.07.29, 18:52:39
Itsu,
When it is convenient for you, I'm curious if you get any output from your toroid coil when positioned properly in between the driven bucking flat coils when using just two scope probes for measurement? IOW, no grounds connected to the toroid coil so the two scope probes would then be taking a differential measurement of any output voltage across the toroidal coil.
I see no output in my setup when there is no ground reference for the solenoid coil no matter what the position of the coil in relation to the flat coils!
Regards,
Pm
Edit: The only output is an equal voltage on both coil leads but this is from the E Field of the flat coils and is not differential between the coil leads.
PM,
i did some differential probe measurements on the toroidal coil earlier in this thread when still using the normal (not litz) pancake coils, see:
https://www.overunityresearch.com/index.php?topic=3924.msg82752#msg82752
So there was output.
I will try again with the new litz wire pancakes, but it seems there is some problem with them as can be seen by the VNA outputs above.
Itsu
Quote from: Itsu on 2020.07.29, 19:37:40
PM,
i did some differential probe measurements on the toroidal coil earlier in this thread when still using the normal (not litz) pancake coils, see:
https://www.overunityresearch.com/index.php?topic=3924.msg82752#msg82752
So there was output.
I will try again with the new litz wire pancakes, but it seems there is some problem with them as can be seen by the VNA outputs above.
Itsu
Itsu,
OK, thanks for pointing out your previous test. O0 I must be doing something wrong here!
Regards,
Pm
PM,
here my new 2 litz wire pancakes (front and rear) with in between the water tube with toroidal coil.
Input on the left from FG (resonance) in the pancakes (series bucking), ouput toroidal coil on the right using differential probes.
Output is 100pF trimmer cap and 600K resistor parallel to coil tuned for max.
Screenshot shows the red differential output.
By the way, the (red) data in the bottom box (Value, Mean, Min etc) can that be removed so to keep only Value?
Itsu
Quote from: Itsu on 2020.07.29, 20:37:49
PM,
here my new 2 litz wire pancakes (front and rear) with in between the water tube with toroidal coil.
Input on the left from FG (resonance) in the pancakes (series bucking), ouput toroidal coil on the right using differential probes.
Output is 100pF trimmer cap and 600K resistor parallel to coil tuned for max.
Screenshot shows the red differential output.
By the way, the (red) data in the bottom box (Value, Mean, Min etc) can that be removed so to keep only Value?
Itsu
Itsu,
OK, thanks for posting that result.
On the Math measurement at the bottom of your window, you have "Statistics" turned on. To change this, press the 'Measure' button on the front panel and then the 'More' button on the screen, and "Statistics" will be the top choice. Simply turn it off and will then only show the main parameter of your measurement.
I notice you have the peak-to-peak measurement selected but you can choose the rms function if you wish. Also, a trick you can use if you have say like in this case peak-to-peak selected and you want to read rms but don't wish to change the Math, again press "measure" on the front panel and then "add measurement" on the screen. Even though you may already have 4 measurements used, you can still select and one of the channels and select "snapshot" and you will have all the parameters showing on the screen for that channel. I know I'm probably preaching to the choir but just in case you were not aware of this!
Regards,
Pm
Quote from: Itsu on 2020.07.29, 19:31:24
Made some measurements and included some in 1 screenshot, so i have 4 screenshots total.
Yup, they are not identical. Not even close :'(
Most likely it is caused by a short, or varying C
IT, or imprecise dimensions.
How to fix it? Unfortunately you have to rewind but this time put 1mm nylon fishing line between the turns, as a spacer.
This will widen the coil, stabilize the C
IT as well as
decrease it, so the coil's self-resonance frequency will go up. The nylon fishing line can be removed later to decrease dielectric losses and increase Q (most glues don't stick to nylon)
A coil former with a rectangular groove tailored to fit the Litz wire and the spacer (fishing line) helps immensely. The groove helps to squeeze the turns and maintain the precise inter-turn spacing as well as the I.D. and O.D of the pancake. This is how I was able to obtain two coils that are so similar.
I put a double sticky tape on the bottom of my groove and used small drops of low-viscosity cyanoacrlylate from an insulin syringe to secure the wires near their crossings. The syringe and
Push Tool (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35664) makes this process almost enjoyable.
Quote from: partzman on 2020.07.29, 22:00:08
Itsu,
OK, thanks for posting that result.
On the Math measurement at the bottom of your window, you have "Statistics" turned on. To change this, press the 'Measure' button on the front panel and then the 'More' button on the screen, and "Statistics" will be the top choice. Simply turn it off and will then only show the main parameter of your measurement.
I notice you have the peak-to-peak measurement selected but you can choose the rms function if you wish. Also, a trick you can use if you have say like in this case peak-to-peak selected and you want to read rms but don't wish to change the Math, again press "measure" on the front panel and then "add measurement" on the screen. Even though you may already have 4 measurements used, you can still select and one of the channels and select "snapshot" and you will have all the parameters showing on the screen for that channel. I know I'm probably preaching to the choir but just in case you were not aware of this!
Regards,
Pm
Thanks PM,
that statistics off worked, but it keeps the box though which is now empty for the most part.
But it looks less distractive. O0
Itsu
Quote from: verpies on 2020.07.30, 02:31:24
Yup, they are not identical. Not even close :'(
Most likely it is caused by a short, or varying CIT, or imprecise dimensions.
How to fix it? Unfortunately you have to rewind but this time put 1mm nylon fishing line between the turns, as a spacer.
This will widen the coil, stabilize the CIT as well as decrease it, so the coil's self-resonance frequency will go up. The nylon fishing line can be removed later to decrease dielectric losses and increase Q (most glues don't stick to nylon)
A coil former with a rectangular groove tailored to fit the Litz wire and the spacer (fishing line) helps immensely. The groove helps to squeeze the turns and maintain the precise inter-turn spacing as well as the I.D. and O.D of the pancake. This is how I was able to obtain two coils that are so similar.
I put a double sticky tape on the bottom of my groove and used small drops of low-viscosity cyanoacrlylate from an insulin syringe to secure the wires near their crossings. The syringe and Push Tool (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35664) makes this process almost enjoyable.
Yes, i noticed the differences in the plots which is funny (not) as the LCR meter shows almost identical inductances.
I also noticed that the super glue i used caused the litz wire (without cloth) to thicken (and harden), so it looks like it somehow reacts with the litz wire coating or so, perhaps indeed causing a short in the end.
Ok, back to the drawing board and get out my fishing gear.
Itsu
Quote from: Itsu on 2020.07.29, 20:37:49
Input on the left from FG (resonance) in the pancakes (series bucking), ouput toroidal coil on the right using differential probes.
For posterity: Ideally the amplitude from the toroidal coil should be zero, when not in NMR.
Quote from: Itsu on 2020.07.29, 20:37:49
Output is 100pF trimmer cap and 600K resistor parallel to coil tuned for max.
How does that compare to the amplitude from the 1-turn sensing coil on that 110mm gray sewer pipe ?
I cannot compare them directly because this one has a different output circuit and different probing.
For comparison, I would prefer both to be without the cap at the output - only the 600K resistor.
Quote from: verpies on 2020.07.30, 10:10:18
For posterity: Ideally the amplitude from the toroidal coil should be zero, when not in NMR.
How does that compare to the amplitude from the 1-turn sensing coil on that 110mm gray sewer pipe ?
I cannot compare them directly because this one has a different output circuit and different probing.
For comparison, I would prefer both to be without the cap at the output - only the 600K resistor.
To compare you mean not using the i-probe, but both a voltage probe and both the coil/loop terminated with the 600K resistor.
Quote from: Itsu on 2020.07.30, 11:04:32
To compare you mean not using the i-probe, but both a voltage probe and both the coil/loop terminated with the 600K resistor.
I think for Partzman the voltage amplitude measurement would be more meaningful (with slight loading by e.g the 600K resistor)
I like the i-probe measurements better, but I have already seen a lot of them.
I have measured the H-field amplitude by a 10mm loop* in 3 orientations between coaxial and parallel pancake coils (weave1, spaced 23mm) connected in parallel and brute-force fed from a a FG+PA without any matching circuits.
1) Radial: 29mA
2) Circumferential: 108µA
3) Axial
a) Near front coil: 2mA
b) Between coils: 5µA
c) Near rear coil: 2mA
This is the textbook field orientation predicted in Smudge's simulation.
All measurements were made at the 50mm radius and (i.e. exactly between the windings of the front and rear coils) and at the midpoint between pancake planes.
P.S.
When the coils were connected in series, these measurements were all over the place.
* Home-made AC H-field sensor constructed from a 3mm coaxial cable, which has Litzed copper center conductor and Litzed copper braid.
QuoteP.S.
When the coils were connected in series, these measurements were all over the place.
So what does that mean? Is the bucking series configuration to sensitive to get it right/stable?
If so, why was this then the first option?
Itsu
Quote from: Itsu on 2020.07.30, 18:40:07
So what does that mean? Is the bucking series configuration to sensitive to get it right/stable?
It creates different electric potentials at the front & rear pancakes. These potentials distort measurements and create crosstalk.
Quote from: Itsu on 2020.07.30, 18:40:07
If so, why was this then the first option?
Because series connection does not have a problem keeping the current equal in both pancakes. With parallel connection extra care must be taken to keep the coils' params the same and even the cable length after the Tee matters.
Also, it is impossible to foresee every engineering difficulty. The bench is the ultimate proof-reader.
Quote from: verpies on 2020.07.30, 11:36:31
I think for Partzman the voltage amplitude measurement would be more meaningful (with slight loading by e.g the 600K resistor)
I like the i-probe measurements better, but I have already seen a lot of them.
Not sure PM is really involved in this specific setup, but below screenshots show the new pancake coils at 16mm apart in series bucking with:
1st the toriodal coil loaded with 690K
2th the single loop loaded with 690K.
Not much difference between them, so PM his statement that the toroidal coil acts like a single turn comes thru.
Itsu
Quote from: Itsu on 2020.07.30, 19:01:16
Not much difference between them, so PM his statement that the toroidal coil acts like a single turn comes thru.
Ok, so it is verified.
I also have a problem with this circumferential induction when I do not use the special winding to prevent it, but I can find an axial position
* between the pancakes where it crosses zero amplitude.
If my pancakes are unbalanced, e.g. by different cables in the parallel connection, then this "zero" becomes illusive for me , too.
*
axial position of the big 1-turn sensing loop or a toroidal coil without the anti-circumferential induction winding. Always coaxial and parallel to the planes of the pancakes.
I put up my old (non litz) pancake coils with the toroidal coil inbetween loaded with 690K.
These old pancake coils are connected parallel and the voltage in resonance (7Mhz) measured across the 690K was:
Parallel 1 20Vpp
Parallel 2 104Vpp
So in this parallel configuration with the old coils we have strong crosstalk.
Quote from: Itsu on 2020.07.30, 19:40:21
So in this parallel configuration with the old coils we have strong crosstalk.
Of course, these coils are not symmetrical center-to-periphery and the current flowing through the C
IT creates
* a magnetic field H
IT, which distorts the desirable H-field. The magnetic field must be very symmetrical in order to cancel at the middle (between the pancake coils).
In other words, these old coils have a "hot end" and the following quote from Smudge no longer applies:
Quote from: Smudge on 2020.07.28, 09:05:11
Then my E field or displacement current arrows would have alternate directions and their H fields would cancel at distances from the surface greater than the turn spacings.
Clearly this would almost eliminate that HIT field at the toroidal coil.
*
It also creates an E-field wrt ground, which is picked up by scope probes.
Right, so it will be pointless to use this setup to hunt for NMR as the crosstalk would completely "cover" the possible NMR reaction.
The problem is that your new proposed litz wire coil in scheme weave 1 with the 1mm fishing line inbetween would be almost impossible to make, at least i have problems envisioning how.
Would it be an alternative to use thicker litz wire (2mm) in weave 1 scheme and losing 2 turns (8 instead of 10) and still cover the toroidal coil and be in the 4.2Mhz resonance range?
Quote from: Itsu on 2020.07.30, 20:29:25
Right, so it will be pointless to use this setup to hunt for NMR as the crosstalk would completely "cover" the possible NMR reaction.
I think so, but I in my opinion, Smudge would like to see you try anyway because I can bet that his expectation is that with this "filling factor" the effect will be much stronger than usual and it will overwhelm the crosstalk. Right Smudge ?
Quote from: Itsu on 2020.07.30, 20:29:25
Would it be an alternative to use thicker litz wire (2mm) in weave 1 scheme and losing 2 turns (8 instead of 10) and still cover the toroidal coil and be in the 4.2Mhz resonance range?
The problem with that is that with a larger wire diameter, the surface area that makes up C
IT increases, and that more than compensates for the decreased inductance of fewer turns.
IMO better than a thicker wire, would be a 1mm wire with an integrated spacer (e.g. the silk braid).
Quote from: Itsu on 2020.07.30, 20:29:25
The problem is that your new proposed litz wire coil in scheme weave 1 with the 1mm fishing line inbetween would be almost impossible to make, at least i have problems envisioning how.
See the diagram below. The spacers are denoted in green color. Notice, that they
do not participate in the wire crossings. If you make them out of non-stick plastic (e.g. Nylon or Teflon), then they can be easily removed, leaving air gaps and a high Q coil.
QuoteI think so, but I in my opinion, Smudge would like to see you try anyway because I can bet that his expectation is that with this "filling factor" the effect will be much stronger than usual and it will overwhelm the crosstalk. Right Smudge ?
Ok, so what would be the preferred setup, parallel 1 (20Vpp 6.4uH) or parallel 2 (104V 16uH)? Guess Par. 1.
QuoteThe problem with that is that with a larger wire diameter, the surface area that makes up CIT increases, and that more than compensates for the decreased inductance of fewer turns.
IMO better than a thicker wire, would be a 1mm wire with an integrated spacer (the silk braid).
See the diagram below. The spacers are denoted in green color. Notice, that they do not participate in the wire crossings. If you make them out of non-stick plastic (e.g. Nylon or Teflon), then they can be easily removed, leaving air gaps and a high Q coil.
Would love to see that diagram.
Itsu
Quote from: verpies on 2020.07.31, 01:03:54
I think so, but I in my opinion, Smudge would like to see you try anyway because I can bet that his expectation is that with this "filling factor" the effect will be much stronger than usual and it will overwhelm the crosstalk. Right Smudge ?
I was hoping that the NMR frequency could be found, perhaps as just a small dip or peak (associated with the presence of the water, i.e. only there when the water is there) superimposed on to the crosstalk. Then you can concentrate on just that frequency, no need for wide band sweeps and all the problems with that. At the fixed frequency you can do all sorts of things to null out the crosstalk without having to dismantle, modify and rebuild. Without the water present play around with some external nulling feedback of some sort until you get near zero output. Then add water.
Smudge
Quote from: Itsu on 2020.07.31, 08:10:32
Ok, so what would be the preferred setup, parallel 1 (20Vpp 6.4uH) or parallel 2 (104V 16uH)? Guess Par. 1.
Yes, the configuration with the smaller inductance is the one with the opposing magnetic fields.
Notice, that with this configuration, as the coils get farther apart, the inductance increases.
Quote from: Itsu on 2020.07.31, 08:10:32
Would love to see that diagram.
I re-attached it to the previous message.
Quote from: Smudge on 2020.07.31, 09:48:04
I was hoping that the NMR frequency could be found, perhaps as just a small dip or peak (associated with the presence of the water, i.e. only there when the water is there) superimposed on to the crosstalk. Then you can concentrate on just that frequency, no need for wide band sweeps and all the problems with that. At the fixed frequency you can do all sorts of things to null out the crosstalk without having to dismantle, modify and rebuild. Without the water present play around with some external nulling feedback of some sort until you get near zero output. Then add water.
Smudge
Smudge,
thanks, i see what you mean, i will use the old pancakes in parallel mode for now to hunt for a small dip or peak superimposed on to the crosstalk.
Itsu
Quote from: verpies on 2020.07.31, 10:30:19
Yes, the configuration with the smaller inductance is the one with the opposing magnetic fields.
Notice, that with this configuration, as the coils get farther apart, the inductance increases.
I re-attached it to the previous message.
OK, seen it, so this would be the finer cut and paste work. O0
@Smudge
Let's consider two 50mm average radius pancake coils (connected in parallel) and a 10mm 1-turn sensor loop placed between these coils, as depicted on the diagram below in orange color?
When the plane of the small orange loop is perpendicular to the blue line and at a distance of 50mm from the red axis, this loop is the most sensitive to the changes of the flux generated by the pancakes in the radial direction.
What phase of voltage do you expect between the input to the pancakes and the voltage induced across the ends of this small orange loop ?
Both old pancake coils in parallel bucking (6.4uH), ringmagnets in place, double distilled water in tube.
Fed via 1:1 balun and 10nF:320pF caps (had to add 220pF parallel to the 100pF var. cap to get into 4Mhz resonance range).
1W input (on the power meter) from the FG fed PA.
Sweeping both manual as via FG between 3.5 and 4.5Mhz.
Output toriodal coil tuned with 100pF parallel trimmer cap for resonance loaded with 687K (differential probing).
Output voltage in the 70 to 120Vpp range (without water more).
No dips, peaks or other anomilies seen.................
Itsu
Quote from: Itsu on 2020.08.01, 17:10:02
No dips, peaks or other anomilies seen.................
Did you account for the NMR being very narrow in frequency ?
Quote from: Itsu on 2020.08.01, 17:10:02
Sweeping both manual as via FG between 3.5 and 4.5Mhz.
The width of any anomaly would be less than one pixel on the 1MHz wide sweep.
Yes, thats what i understand also from Smudge.
So starting with a 1Mhz wide sweep, then gradually narrowing it down.
Quote from: verpies on 2020.08.01, 15:41:26
@Smudge
Let's consider two 50mm average radius pancake coils (connected in parallel) and a 10mm 1-turn sensor loop placed between these coils, as depicted on the diagram below in orange color?
When the plane of the small orange loop is perpendicular to the blue line and at a distance of 50mm from the red axis, this loop is the most sensitive to the changes of the flux generated by the pancakes in the radial direction.
What phase of voltage do you expect between the input to the pancakes and the voltage induced across the ends of this small orange loop ?
From the pure magnetic point of view (no displacement current through the air near the pick up coil in a direction to cause induction) I would expect a 90 degree phase between the induced open circuit voltage in the test coil and the current in the pancake coils (the propagation delay at 4Mhz is about 5 thousandths of a degree per mm, so it's effect would be below the phase measurement accuracy). I would expect a small phase between the voltage across the pancake coils and the induced voltage in the test coil because the resistance of the pancake coils creates a less than 90 degree phase shift between applied voltage and current.
Smudge
Quote from: Smudge on 2020.08.02, 08:20:50
I would expect a small phase between the voltage across the pancake coils and the induced voltage in the test coil because the resistance of the pancake coils creates a less than 90 degree phase shift between applied voltage and current.
What about the voltage phase vs. frequency relationship of the sensor loop and pancake coils connected in parallel ?
The pancake coils are coaxial and parallel and the distance between them is 16mm and their Interwinding Capacitaence (C
IW) is 5.8pF at that distance.
The parameters of one pancake coil are:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns of 1mm Litz wire (68 strand), 3.2m long, antiradial
weave1 (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35585).
When the middle break is opened:
Intrawinding/Interturn Capacitance (C
IT): 174.75pF (@100kHz)
ESR: 218Ω (@100kHz)
When the middle break is closed:
Inductance: 19.65µH (@100kHz)
DC resistance: 0.22Ω
My sensor coil is 10mm in diameter, 60nH, it is shielded against capacitive coupling and is loaded with 50Ω + 0j.
P.S.
In my system, there is a 5cm mismatch in the transmission line length after the Tee that feeds the pancake coils in parallel.
Quote from: verpies on 2020.08.02, 12:57:57
What about the voltage phase vs. frequency relationship of the sensor loop and pancake coils connected in parallel ?
The pancake coils are coaxial and parallel and the distance between them is 16mm and their Interwinding Capacitaence (CIW) is 5.8pF at that distance.
The parameters of one pancake coil are:
OD: 112mm
ID: 91mm
Winding: 1 layer, 10 turns of 1mm Litz wire (68 strand), 3.2m long, antiradial weave1 (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35585).
When the middle break is opened:
Intrawinding/Interturn Capacitance (CIT): 174.75pF (@100kHz)
ESR: 218Ω (@100kHz)
When the middle break is closed:
Inductance: 19.65µH (@100kHz)
DC resistance: 0.22Ω
My sensor coil is 10mm in diameter, 60nH, it is shielded against capacitive coupling and is loaded with 50Ω + 0j.
P.S.
In my system, there is a 5cm mismatch in the transmission line length after the Tee that feeds the pancake coils in parallel.
IMO the displacement current through the intrawinding capacitance is in the wrong direction (I.e.radial) to create flux in your test coil. And the 5.8pF interwinding capacitance should not come into play with the pancake coils in parallel as the coils are at the same potential.
Smudge
@Smudge
I agree.
Thank you for noticing the directions of the displacement current but that's not a direct answer to my question.
Quote from: verpies on 2020.08.02, 19:49:06
@Smudge
I agree.
Thank you for noticing the directions of the displacement current but that's not a direct answer to my question.
Your question was "What about the voltage phase vs. frequency relationship of the sensor loop and pancake coils connected in parallel ?"
Do you want me to predict the relationship or to comment on a measured relationship?
Smudge
Quote from: Smudge on 2020.08.03, 07:46:45
Do you want me to predict the relationship or to comment on a measured relationship?
I would like you to predict the voltage phase relationship.
In you prediction please consider the reactance of the pancake coils as the VNA's AC voltage source is trying to push current into them, as well as its transmission line like effects, e.g reflections from the coil's midpoint.
Also note, that the sensing coil is loaded by the 50Ω+j0Ω internal impedance of the VNA's receiver ( the signal generator has the 50Ω+j0Ω internal impedance, too )
FYI: The sensing coil is constructed using
this method (https://youtu.be/2xy3Hm1_ZqI?t=380) out of a 2mm diameter Litz coaxial cable. Note, that there is a 11cm rigid coaxial cable extension (in 3mm O.D. plastic stiffener) in front of the sensing loop.
Quote from: verpies on 2020.08.03, 12:17:54
I would like you to predict the voltage phase relationship.
In you prediction please consider the reactance of the pancake coils as the VNA's AC voltage source is trying to push current into them, as well as its transmission line like effects, e.g reflections from the coil's midpoint.
Also note, that the sensing coil is loaded by the 50Ω+j0Ω internal impedance of the VNA's receiver ( the signal generator has the 50Ω+j0Ω internal impedance, too )
FYI: The sensing coil is constructed using this method (https://youtu.be/2xy3Hm1_ZqI?t=380) out of a 2mm diameter Litz coaxial cable. Note, that there is a 11cm rigid coaxial cable extension (in 3mm O.D. plastic stiffener) in front of the sensing loop.
OK. Predicting transmission line effects is very difficult as it is a complicated transmission line. Taking a simplistic view, from your data where you have 3.2m of wire it can be considered as a 1.6m long line terminated in a short circuit at the far end (the mid point). From your capacitance measurement the distributed capacitance is 109 pF/m and from your inductance measurement the distributed inductance is 12.28 uH/m. That yields a time delay of 36.6 nS/m along the line, i.e. 58.54 nS total. Its first resonant frequency where the line is one quarter wavelength is then 4.27 MHz, which just happens to be around the frequency we are interested in. That is not good news as it indicates difficulty in pumping current into the line, which would tend towards an open circuit at resonance. Your 218 Ohms CSR at 100 KHz is also worrying, as it is more likely due to the HF resistance of the wire than to dielectric losses, and that resistance could be greater at 4MHz. It is many years since I got involved with the use of Litz wire, but I do remember that above a certain frequency Litz wire is of no use, it is better to use a single conductor, possibly with a low conductivity coating such as silver. When I look back at the introduction of ferrite-rod antenna for portable radios in place of the old wire frame antenna, the long-wave coils were multi-layer wave-woven with Litz wire occupying a small length of the rod, while the medium-wave coils used a single-layer single-strand wire wound along the length of the rod. And I think our 4MHz falls into the medium-wave category. Do you have any measurements that support the use of Litz wire at our frequency? I don't think I can add much more, other than below that quarter wave resonance the input will appear inductive and above it will appear capacitive.
Smudge
Quote from: Smudge on 2020.08.05, 09:53:48
Do you have any measurements that support the use of Litz wire at our frequency? I don't think I can add much more, other than below that quarter wave resonance the input will appear inductive and above it will appear capacitive.
Of course, I have many measurements for you.
For example, the first plot (3rd and 4th, too) is of two unconnected but identical 10mm H-field sensing loops placed 8mm apart (coaxial & parallel). See Fig.1
Notice that together with the VNA's internal impedance of 50Ω they form a high-pass filter. Also note, that their voltage phase shift is close to 90º.
The second plot is of two pancake coils connected in parallel and placed 16mm apart (coaxial & parallel). See Fig.2
From it, you can deduce how much total current flows into these coils (note that this current also includes any displacement currents!)
I don't know how proficient you are in reading VNA plots but the vertical scale for the green │S21│measurement of the pancake coils is 100mU per division (linear scale) and the 1 (or 1000mU) is at the very top of the plot and 0 is at the very bottom of the plot. The labels of the vertical axis do not show this because the │S21│ trace does not have focus. The blue S21 Phase plot has the focus (this is indicated by the blue S21 rectangle in the upper-left corner of the plot area).
It is important to keep in mind that the │S21│measurement is a dimensionless
ratio of two voltage amplitudes: the incident signal and the signal received after being transmitted through the coils. 1 means that the full voltage amplitude is transmitted through and 0 means that nothing is transmitted through. The 100mU means that 10% is transmitted through, 200mU means 20% ...etc. Of course, anything that is not transmitted through is either reflected back to the source or dissipated as heat/ultrasonics/far-field EM radiation.
@Smudge
Considering all of the above, what S21 Phase plot do you expect when the VNA's transmitter is connected directly to the two paralleled pancake coils separated by 16mm and the VNA's receiver is connected directly to the 10mm sensor loop (orange) perpendicular and centered along the blue radial line in the diagram below ?
Also, do you expect the orange sensor loop's distance from the red line (when sliding along the blue line) to influence the plot of S21 Phase vs. f ?
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014)
Quote from: verpies on 2020.08.05, 13:09:43
Of course, I have many measurements for you.
For example, the first one is of two identical 10mm H-field sensing loops placed 8mm apart (coaxial & parallel).
Notice that together with the VNA's internal impedance of 50Ω they form a high-pass filter. Also note, that their voltage phase shift is close to 90º.
OK, you would expect a linear rise with frequency since the induced voltage in the output loop is proportional to the frequency. As for the phase plot it is near 90 degrees near zero frequency as it should be, and propagation delay will cause that to become less than 90 degrees at higher frequencies. That 33 degree approx reduction from 90 degrees at 50MHz represents a line length of about 0.5m. What lengths of cables connected the two loops to the transmitter and receiver? If that is about 0.5m then all is as it should be.
QuoteThe second measurement is of two pancake coils connected in parallel and placed 16mm apart (coaxial & parallel).
From it, you can deduce how much total current flows into these coils (note that this current also includes any displacement currents!)
The coils are connected in parallel so this must be transmission to another coil, presumably the 10mm H-field sensor. That resonance at 5.695MHz could be the transmission line effect being a quarter wavelength shorted line, but then you would expect the next dip at 3/4 wavelength to be at 3 times this frequency whereas it is around 5 times. And at around 3 times frequency it is a peak, not a dip. But you do get this 1-3-5 relationship showing there is a transmission line affect.
QuoteI don't know how proficient you are in reading VNA plots but the vertical scale for the green │S21│ measurement of the pancake coils is 100mU per division (linear scale) and the 1 (or 1000mU) is at the very top of the plot and 0 is at the very bottom of the plot. The labels of the vertical axis do not show this because the │S21│ trace does not have focus. The blue S21 Phase plot has the focus (this is indicated by the blue S21 rectangle in the upper-left corner of the plot area).
It is important to keep in mind that the │S21│measurement is a dimensionless ratio of two voltage amplitudes: the incident signal and the signal received after being transmitted through the coils. 1 means that the full voltage amplitude is transmitted through and 0 means that nothing is transmitted through. The 100mU means that 10% is transmitted through, 200mU means 20% ...etc. Of course, anything that is not transmitted through is either reflected back to the source or dissipated as heat/ultrasonics/far-field EM radiation.
It is many many years since I had dealings with network analyzers, we used Hewlett Packard ones back in the 1960's. I don't know what to suggest to get that first dip away from 4-ish MHz other than winding (or printing) fewer turns using wide strip, maybe even just a single turn of very wide strip.
Smudge
Quote from: verpies on 2020.08.05, 13:31:49
@Smudge
Considering all of the above, what S21 Phase plot do you expect when the VNA's transmitter is connected directly to the two paralleled pancake coils separated by 16mm and the VNA's receiver is connected directly to the 10mm sensor loop (orange) perpendicular and centered along the blue radial line in the diagram below ?
Also, do you expect the orange sensor loop's distance from the red line (when sliding along the blue line) to influence this S21 Phase plot ?
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014)
Haven't you just sent me that measurement or did I presume wrongly? I would much rather analyze a measurement than predict an outcome. I would not expect sliding along the blue line to significantly influence the phase plot.
Smudge
Quote from: Smudge on 2020.08.05, 15:17:45
Haven't you just sent me that measurement or did I presume wrongly?
No, no. I posted this diagram before but I never posted the signal received by this orange 10mm sensing loop.
This diagram serves mainly to define the directional/orientational nomenclature.
Quote from: Smudge on 2020.08.05, 15:08:45
OK, you would expect a linear rise with frequency since the induced voltage in the output loop is proportional to the frequency.
That is only one side of the coin.
The other 10mm loop (the transmitting one) forms a filter with the VNA's internal impedance.
We cannot expect this transmitting loop to generate a constant amplitude flux variations, because it is driven by a constant voltage source and the current amplitude flowing through it is not constant with frequency.
Quote from: Smudge on 2020.08.05, 15:08:45
What lengths of cables connected the two loops to the transmitter and receiver?
2cm and 6cm. Yes, they are not equal length, because I burned out my SMA cables in an earlier high power run. But since then I have acquired new ones. Do you want me to repeat this measurement with two 15cm cables after the Tee ?
Quote from: Smudge on 2020.08.05, 15:08:45
The coils are connected in parallel so this must be transmission to another coil, presumably the 10mm H-field sensor.
O, o! Back up....
There was no H-field sensor used in that measurement. It was just a S21 measurment of the naked pancake coils connected in parallel.
Quote from: Smudge on 2020.08.05, 15:17:45
I would much rather analyze a measurement than predict an outcome.
I will post a 4 dimensional measurment, i.e. an │S21│ and S21 Phase vs. frequency plots of the signal received by the 10mm H-field probe as its distance from the red line varies in micrometer steps (robotically positioned).
First, let's just straighten out the confusion about the previous measurement of the naked pancake coils, which did
not involve the H-field probe.
Quote from: verpies on 2020.08.05, 15:50:20
That is only one side of the coin.
The other 10mm loop (the transmitting one) forms a filter with the VNA's internal impedance.
We cannot expect this transmitting loop to generate a constant amplitude flux variations, because it is driven by a constant voltage source and the current amplitude flowing through it is not constant with frequency.
OK, a constant voltage source should create a constant rate of change of current, which for sine waves is a current proportional to the reciprocal of the frequency. That should cancel out the induced voltage being proportional to frequency, so we should see a flat response but we don't. So is that fall-off with frequency due to increased resistance, radiation or a combination of both? I would not expect the resistance change (skin effect) to show such a linear response.
Quote2cm and 6cm. Yes, they are not equal length, because I burned out my SMA cables in an earlier high power run. But since then I have acquired new ones. Do you want me to repeat this measurement with two 15cm cables after the Tee ?
Well comparison with the existing result would tell you whether cable length is significant.
QuoteO, o! Back up....
There was no H-field sensor used in that measurement. It was just a S21 measurment of the naked pancake coils connected in parallel.
OK. tell me if I am wrong in now assuming you have 50 ohm cables connecting transmitter to receiver but somewhere in the middle where they join you have the parallel coils connection shunted across that 50 ohm line. I think I need to know the cable dimensions, i.e. the physical set-up, in order to comment usefully
Smudge
Edit, I have just realised that I am talking rubbish in my first reply here. There is not a fall-off, there is a linear rise. So if the current remains constant with frequency then the coil resistance must dominate and be constant, but again I wouldn't expect that. I'll look again at the slope of that line and see what I come up with.
Quote from: Smudge on 2020.08.05, 16:19:58
OK. tell me if I am wrong in now assuming you have 50 ohm cables connecting transmitter to receiver but somewhere in the middle where they join you have the parallel coils connection shunted across that 50 ohm line. I think I need to know the cable dimensions, i.e. the physical set-up, in order to comment usefully
The diagram below should make it clear what I was describing in
this post (https://www.overunityresearch.com/index.php?topic=3924.msg83769#msg83769). I am not listing the lengths of the transmission lines before the calibration plane because their influence is calibrated out.
Also, the pigtail transmission lines AFTER the Tee, were 2cm and 6cm in the old setup, but in the new setup, they can be 15cm and 15cm ...or... 5cm and 5cm
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36063)
What would be a preferred input matching circuit when using parallel opposing pancake coils?
Not the one shown below which is for series opposing.
Itsu
Why not? To these caps it does not matter whether these coils are connected in series or in parallel. These caps will absorb the coils' internal CIT anyway.
Just don't expect a low SWR because these coils are not designed to radiate their energy into the far field (as radio waves). Nor are they designed to dissipate this energy as heat.
They are designed to maintain a high reactive current (which stores the energy in local near-fields that never leave the coils) and to reflect almost all of this energy, back into these caps. IOW: They are designed to slosh the energy between inductance and capacitance.
The circulating current will be much much higher than the input current, because of the the current magnification ratio iCIRCULATING / iINPUT ...also known as Q.
Quote from: Smudge on 2020.08.05, 16:19:58
Edit, I have just realised that I am talking rubbish in my first reply here. There is not a fall-off, there is a linear rise. So if the current remains constant with frequency then the coil resistance must dominate and be constant, but again I wouldn't expect that. I'll look again at the slope of that line and see what I come up with.
That tiny H-field loop has very low inductance, so it will look like a short circuit against the 50 Ohm output impedance of the transmitter, hence the current will remain constant as the voltage is kept constant. Then induction into the receiving loop explains the linear rising output voltage, except I would have expected the transmitter loop inductance to show up at the higher frequency 50MHz end.
Smudge
Because, as i understood, the input caps act as a impedance step up from the low (50 Ohm) input impedance to the high series impedance, while the parallel circuit will be low impedance already.
An S11 scan at the input shows the low (1 Ohm) input impedance with a shallow (2.5 Ohm) peak at reasonance (4.3Mhz).
But i forgot the purpose of these coils as to not radiate so i am sure you are right in that it behaves differently as i expect.
Itsu
Quote from: Itsu on 2020.08.06, 07:59:12
Because, as i understood, the input caps act as a impedance step up from the low (50 Ohm) input impedance to the high series impedance, while the parallel circuit will be low impedance already.
A parallel LC circuit at resonance acts as an open circuit. Even one pancake coil acts as a parallel LC circuit. Two pancakes connected in parallel - even more so.
If you look at
the 2nd plot (https://www.overunityresearch.com/index.php?topic=3924.msg83769#msg83769), then you can see this response at 5.7MHz ( and the higher dip is from shorted-transmission-line behavior ).
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36076)
Quote from: Itsu on 2020.08.06, 07:59:12
An S11 scan at the input shows the low (1 Ohm) input impedance with a shallow (2.5 Ohm) peak at reasonance (4.3Mhz).
You should print out
this diagram (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=35778) and permanently glue it to the back of your VNA.
Quote from: Smudge on 2020.08.06, 07:53:43
That tiny H-field loop has very low inductance,
Yes, around 20nH.
The transmitter's internal impedance and receiver's impedance in parallel are 25Ω.
Don't this impedance and inductance form a high pass RL filter ?
Quote from: Smudge on 2020.08.06, 07:53:43
... except I would have expected the transmitter loop inductance to show up at the higher frequency 50MHz end.
At higher frequencies it looks like this:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36086)
The phase wraps-around from -180º to +180º.The second plot below was done with one H-Field probe physically flipped 180º.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36088)
This phase in the second plot also wraps-around from -180º to +180º.
Quote from: verpies on 2020.08.06, 10:32:33
Yes, around 20nH.
The transmitter's internal impedance and receiver's impedance in parallel are 25Ω.
Don't this impedance and inductance form a high pass RL filter ?
That is one way of looking at it, but the two 50 Ohm impedances are not in parallel, there is the transmission path in between. The 10nH coil shunting the Tx 50 Ohms forms a high pass filter, then the other 10nH coil driving into the Rx 50 Ohms forms a low pass filter. I think the two in series accounts for the linear phase progression over the very wide frequency range.
QuoteAt higher frequencies it looks like this:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36086)
The phase wraps-around from -180º to +180º.
The second plot below was done with one H-Field probe physically flipped 180º.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36088)
This phase in the second plot also wraps-around from -180º to +180º.
Those phase flips are purely math artefacts (as anyone who has played with trig functions in Excel will tell you) and of no significance, the phase progression is continuous.
Smudge
Quote from: Smudge on 2020.08.06, 13:40:36
Those phase flips are purely math artefacts
Of course, that's what wrapping-around is.
So what phase relationships do you expect in
this experiment (https://www.overunityresearch.com/index.php?topic=3924.msg83770#msg83770) ?
Quote from: verpies on 2020.08.06, 16:16:30
Of course, that's what wrapping-around is.
So what phase relationships do you expect in this experiment (https://www.overunityresearch.com/index.php?topic=3924.msg83770#msg83770) ?
Sorry for the delay in replying but I have been busy dealing with family matters. Going back to your data posted in reply 510 where you observed a resonance at 5.695MHz. I misunderstood the set up so my reply 512 where I thought this would be a problem was wrong. Your reply 518 showed the set up and (if we ignore the small phase delays along the coax cables) you had the pancake coils connected in parallel having an inductance that you quoted earlier as 4.9uH at the 16mm separation, and the signal was fed from transmitter to receiver through this inductance. The coax connections added lumped capacitance across each coil, and you also had earlier measurements of intra-turn capacitance of 174.75pF for a single pancake coil, which is far greater than the coax capacitance. Thus you had the signal being fed through a parallel LC circuit that calculates to have a resonance of 3.85MHz based on those values, whereas the actual resonance was that 5.695MHz. That created the dip in transmission. That is a good result since in the actual NMR experiment we want to use resonance to get maximum current into the coils, and we can easily lower the self resonance by adding capacitance. The bad news is that we cannot use the input matching circuit I originally proposed, which puts a high value capacitance in series with the one producing the resonance.
Now answering your specific questions, your method of connecting the coils via the S21 fixture and the Tee is simply connecting the parallel LC circuit across the 50 Ohm signal source. This is of course a complete mismatch at resonance and does not make use of the high Q that can be obtained with better matching. So I would expect the signal induced into the H-field sensor to follow a low Q resonance peaking at 5.9MHz. For the phase plot I would expect this be +90 degrees at the lowest frequency, falling as the frequency rises then rising again to pass through +90 degrees at that self resonant frequency, then rising towards +180 degrees.
To maximize the current into the coils we must use a high impedance source, and that requires a step-up transformer between the 50 ohm signal source and the coils. I am sorry that this wasn't made clear at the start, but I did not envisage the complex interleaved winding arrangements that would create such huge self capacitance.
Smudge
@Smudge
In this setup (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36069), do you expect the sensor orange loop's distance from the red line (when sliding along the blue line) to influence the plot of S21 Phase vs. f ?
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014)
Quote from: Smudge on 2020.08.11, 10:11:58
Your method of connecting the coils via the S21 fixture and the Tee is simply connecting the parallel LC circuit across the 50 Ohm signal source. This is of course a complete mismatch at resonance and does not make use of the high Q that can be obtained with better matching.
This is a measurement technique, not an optimal driving technique.
Quote from: Smudge on 2020.08.11, 10:11:58
...I did not envisage the complex interleaved winding arrangements that would create such huge self capacitance.
With this winding arrangement, the turns of the coil are spaced the same distance, as with the Archimedean spiral winding. The turns also have almost the same area of surface contact (a little less at crossings).
Seeing how much problems this inter-turn capacitance creates, now I am advocating putting an air gap between the turns (generated by removing a consistent spacer used during the winding process).
Quote from: Smudge on 2020.08.11, 10:11:58
The bad news is that we cannot use the input matching circuit I originally proposed, which puts a high value capacitance in series with the one producing the resonance.
I don't understand.
From the point of view of these inductances (L
1 and L
2), the external capacitance C
EXT and the inter-turn capacitances C
IT1 and C
IT2, are all in parallel.
Quote from: Smudge on 2020.08.11, 10:11:58
So I would expect the signal induced into the H-field sensor to follow a low Q resonance peaking at 5.9MHz. For the phase plot I would expect this be +90 degrees at the lowest frequency, falling as the frequency rises then rising again to pass through +90 degrees at that self resonant frequency, then rising towards +180 degrees.
Did you account for the frequency characteristics of the H-field probe, being superimposed on this measurement ? These characteristics were posted in post #510 (https://www.overunityresearch.com/index.php?topic=3924.msg83769#msg83769) and discussed in subsequent posts.
Take a look at the measurement of the signal received by this H-field probe when it is placed exactly between (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014) the pancake coils at the radius (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014) of 50.75mm, when using
these connections (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36069).
I kept quiet the last weeks to give verpies and Smudge a chance to clear up some things and to avoid the heat here on the bench as it is unnatural hot here in the Netherlands at this time of the year having a heat wave the last 10 days with temperatures daily over 33°C.
What i did was testing the old pancake coils i have in parallel opposing mode sweeping the frequency in little steps both by FG as manual.
But in these 2 weeks or so i did not see anything out of the ordinary.
So either the NMR is not there or hidden underneath the crosstalk signals.
I will be waiting for my new litz wire to arrive so i can build 2 new pancake coils as proposed by verpies here:
https://www.overunityresearch.com/index.php?topic=3924.msg83667#msg83667
which will be a challenge on its own.
Regards Itsu
Quote from: Itsu on 2020.08.16, 12:11:41
I kept quiet the last weeks to give verpies and Smudge a chance to clear up some things
I have more measurement data for Smudge when he replies.
Quote from: Itsu on 2020.08.16, 12:11:41
...and to avoid the heat here on the bench as it is unnatural hot here in the Netherlands at this time of the year having a heat wave the last 10 days with temperatures daily over 33°C.
That ambient temperature is sick!
Quote from: Itsu on 2020.08.16, 12:11:41
What i did was testing the old pancake coils i have in parallel opposing mode sweeping the frequency in little steps both by FG as manual.
But in these 2 weeks or so i did not see anything out of the ordinary.
So either the NMR is not there or hidden underneath the crosstalk signals.
Low electric and magnetic crosstalk
* and the "DC" magnetic field uniformity from the magnets is the key to success.
Quote from: Itsu on 2020.08.16, 12:11:41
I will be waiting for my new litz wire to arrive so i can build 2 new pancake coils as proposed by verpies...
which will be a challenge on its own.
Get yourself a good coil former, with an appropriate groove.
*
achieved by spatial topology and shielding or by temporal separation
Quote from: verpies on 2020.08.16, 14:42:11
I have more measurement data for Smudge when he replies.
I have not abandoned you. My wife was rushed to hospital last Thursday with a serious blood infection. Not covid related thank goodness but is cellulitis, infected from small ankle injury some two weeks previously. She is on intravenous strong antibiotic drip so have to take her to hospital each day for this treatment. Haven't had time to look into your latest data but will do so when these dratted hospital visits stop.
Smudge
Wishing the best for you Smudge! May your wife recover quickly
and completely.
OK I am back and wife is now quite well thank you. Getting back to the questions at hand, here is my interpretation of Verpies' measurement for the response of the two pancake coils in reply 510 using the connections shown with his S21 fixture and TEE. Note that the outer casings of the Tee and the coax cables connecting to each coil are not at ground potential, so they contribute leakage capacitance to ground shunted across the output. At low frequencies the phase delay along those coax cables is insignificant, but the cables do add capacitance across each coil that adds to the intraturn capacitance. The connection diagram simplifies to the one shown below. Thus one would expect a dip in the response at the LC resonant frequency as indeed occurs at about 5.7MHz in the measurement.
At somewhat higher frequencies the method of winding and feeding the pancake coils each using 3.2m of wire causes them to appear as a length of shorted transmission line of length 1.6m, and that feature has to be taken into consideration. That would tend towards an open circuit at the first resonant frequency where the line is a quarter wavelength, a short circuit at the next frequency where the line is a half wavelength and another open circuit when the line is three-quarters wavelength and so on. That half wavelength case is seen in the measurement as a peak at 18.7MHz, followed by the three-quarters wavelength dip at 29.5MHz. The broad quarter wavelength dip at around 9MHz is overshadowed by the 5.7MHz lumped constant dip but does cause that to be assymmetric. At even higher frequencies the coax cables lengths begin to take effect, but that is way outside our expected NMR freqencies.
Looking at the measurements using the H-field detector I have added the typical amplitude response for the LC resonance of the pancake coils at just above 5MHz. Clearly this peak does not appear, we merely get an inflection of the rising curve expected for the half wavelength transmission line peaking near 18MHz. That tells us that at the 5MHz resonance the coil current sloshing back and forth between the L and the C is there but not enough to show a peak. However note that here the Q of the LC circuit is low as the 50 Ohm input resistance is across it. The coil current (hence the detected H field) can be increased if an appropriate input system that allows higher Q is applied. The other detail at higher frequencies is really only of academic interest. I have added detail along the top of the chart to indicate roughly regions where the LC approach, the transmission line effect and the coax cable lengths overlap.
My conclusion is that the lumped constant LC approach is satisfactory at the expected NMR frequency, but my suggested input circuit is not suitable due to the high self capacitance of the special pancake coils. It should be relatively easy to create an input circuit that allows much higher Q and hence greater coil current. The transmission line effect in the special pancake coils, while seriously affecting the coil current at high frequencies that are not of interest, has a small effect at our expected NMR frequency that should be even smaller when the LC circuit obtains higher Q.
Smudge
@Smudge
I am still analyzing your latest reply but I noticed that it did not consider this question (https://www.overunityresearch.com/index.php?topic=3924.msg83846#msg83846).
P.S.
I hope that you (plural) did not catch the virus in the hospital. Are you in the U.S., where the only effective prophylactics have been politicized?
Quote from: verpies on 2020.08.21, 19:55:32
@Smudge
I am still analyzing your latest reply but I noticed that it did not consider this question (https://www.overunityresearch.com/index.php?topic=3924.msg83846#msg83846).
As the H field is radial within the pancake coils I would expect it to diminish with increasing radial distance from the axis. Thus I would expect maximum H field value with the sensor close to the minimum radius of the pancake coils, reducing in value as the sensor moves to the outer radius, and of course dropping drastically as it moves away from the coils conductors. That peak could be exaggerated by the quarter wavelength transmission line effect in the pancake coils where maximum current occurs at the shorted end.
QuoteP.S.
I hope that you (plural) did not catch the virus in the hospital. Are you in the U.S., where the only effective prophylactics have been politicized?
My wife got the infection in strong wind when the shed door blew against her lower leg causing her skin to be penetrated and a wooden splinter to enter. It developed as cellulitis. She was tested for Covid 19 on entry to the hospital and that came back negative. I have not developed any Covid 19 symptoms so far. I am in the UK where there is no evidence of effective prophylactics against Covid 19 being deliberately politicized, but as is normal in life the stringent protocols needed to ensure any new regimes are absolutely safe is holding back any introductions. Of course that is seen by some as being politically motivated by big pharma.
Smudge
Quote from: Smudge on 2020.08.22, 13:59:31
As the H field is radial within the pancake coils I would expect it to diminish with increasing radial distance from the axis. Thus I would expect maximum H field value with the sensor close to the minimum radius of the pancake coils, reducing in value as the sensor moves to the outer radius, and of course dropping drastically as it moves away from the coils conductors.
Does that meet your expectations?
(https://i.imgur.com/5gmW1RP.gif)
Click on the image to magnify it.
This animation packs a lot of information so viewing it frame by frame, might give you more time to analyze it as it changes.
The red numbers at the bottom indicate the radial distance of the H-field sensor loop in mm from the major axis (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36069) (we recently discussed the frequency response of this sensor loop).
Quote from: Smudge on 2020.08.22, 13:59:31
That peak could be exaggerated by the quarter wavelength transmission line effect in the pancake coils where maximum current occurs at the shorted end.
Yeah, about that...
Could the displacement currents and standing waves in that coiled "transmission line" explain the spatial H-field distribution vs. f, depicted in this animation ?
Quote from: verpies on 2020.08.22, 15:24:56
Does that meet your expectations?
There is a lot of information in that animation much of which is of little significance wrt our NMR experiment. Those wild variations in the S21 value at high frequencies well above the expected NMR frequency is only of academic interest. I should point out to others watching this thread that we are interested in maximising the H value and the S21 value does not give us H directly. I think we can ignore those wild peaks and troughs as they do not relate to our work. I have neither the software, the time or the inclination to fully analyze all that information. I have eyeballed the animation and crudely extracted the S21 value at a fixed frequency of 5MHz which is close to the expected NMR frequency. I get the data plotted below. The S21 value is a maximum when the H probe is at a radius of about 53 cm which conveniently places it nicely within the pancake coil region that would be occupied by the water sample.
QuoteYeah, about that...
Could the displacement currents and standing waves in that coiled "transmission line" explain the spatial H-field distribution vs. f, depicted in this animation ?
Probably, but things are much simpler in our frequency region of interest.
Smudge
Quote from: Smudge on 2020.08.23, 08:51:29
...we are interested in maximising the H value and the S21 value does not give us H directly.
Why not?
S21 is a ratio of two voltage amplitudes. These two amplitudes can represent anything.
In this case it is a ratio of the voltage amplitude applied to the coils and the voltage amplitude induced in the H-field sensor's loop
*.
Doesn't the amplitude of the voltage induced in the H-field sensor's loop give us the amplitude of the H-field directly ?
*
loaded by the 50Ω resistance of the receiver.
Quote from: verpies on 2020.08.23, 12:58:44
Doesn't the amplitude of the voltage induced in the H-field sensor's loop give us the amplitude of the H-field directly ?
IMHO, not necessarily. In my experiments with the rectangular pcb flat coils spaced apart at a fixed distance and run in a resonant bucking mode, circular, triangular and even oval shaped H field sense coils yielded no output when measured differentially. However, a square shaped sense coil that fit up to the inside edges of the flat coils yielded output. My conclusion was the E field was the source but this is my opinion at this point.
Regards,
Pm
Quote from: verpies on 2020.08.23, 12:58:44
Why not?
S21 is a ratio of two voltage amplitudes. These two amplitudes can represent anything.
In this case it is a ratio of the voltage amplitude applied to the coils and the voltage amplitude induced in the H-field sensor's loop*.
Doesn't the amplitude of the voltage induced in the H-field sensor's loop give us the amplitude of the H-field directly ?
* loaded by the 50Ω resistance of the receiver.
Not directly as the voltage is proportional to frequency, hence in the S21 plots against frequency the higher frequency values are over emphasised.
Edit. Forget that statement, the voltage on the pancake coils is also proportional to frequency so you are right.
Smudge
Quote from: partzman on 2020.08.23, 15:21:08
My conclusion was the E field was the source but this is my opinion at this point.
I agree that E-field plays a role with series connected pancake coils and unshielded sensing loop, but I do not think, that this is the case, with the pancake coils connected in parallel, because both coils are at the same electric potential. Also the H-field sensing loop is small and electrically shielded in my system.
Furthermore, the 90º rotation of the H-Field sensor's loop with respect to the pancake coils influences the voltage appearing across it greatly and 180º rotation shifts the phase of the voltage appearing across it by 180º, too. That is an additional evidence that the voltage appearing across it is the result of Faraday's induction.
Quote from: Smudge on 2020.08.23, 08:51:29
There is a lot of information in that animation much of which is of little significance wrt our NMR experiment. Those wild variations in the S21 value at high frequencies well above the expected NMR frequency is only of academic interest. I should point out to others watching this thread that we are interested in maximising the H value and the S21 value does not give us H directly.
I think there is nothing wrong with trying to understand what is going on because a deeper understanding can lead to further optimization methods.
For example, take a look at the the H-field's amplitude 58mm away from the major axis. It is 3.3x higher at 23MHz than at 4.3MHz. That is over 11x more power - nothing to sneeze at.
A better understanding of this phenomenon would allow us to generate an order of magnitude more powerful H-fields at the desired NMR frequency, instead of the undesired 23MHz ...or better.
Also, a better understanding would allow us to have a better control over the
spatial distribution of the H-field's amplitudes in the radial direction.
Quote from: Smudge on 2020.08.23, 08:51:29
I have neither the software, the time or the inclination to fully analyze all that information.
It took me an entire day to build a robotic H sensor positioner and make 182 of these H-field measurements for you.
Ask Itsu to provide you with the access to the video of this experiment, if you are curious how it was performed.
Quote from: verpies on 2020.08.23, 20:13:34
I think there is nothing wrong with trying to understand what is going on because a deeper understanding can lead to further optimization methods.
For example, take a look at the the H-field's amplitude 58mm away from the major axis. It is 3.3x higher at 23MHz than at 4.3MHz. That is over 11x more power - nothing to sneeze at.
A better understanding of this phenomenon would allow us to generate an order of magnitude more powerful H-fields at the desired NMR frequency, instead of the undesired 23MHz ...or better.
Also, a better understanding would allow us to have a better control over the spatial distribution of the H-field's amplitudes in the radial direction.
Yes, I see that 11x more power at a resonant frequency of the shorted transmission line inside the pancake coils, which is offering a high Q. To get that effect at our NMR frequency would require pancake coils wound with longer length fine wire of more turns, which with their special winding technique may not be practical. On the other hand I see the LC resonance as low Q which could be improved greatly with the right input circuit, and that could be a more practical way of achieving the same effect. However, being just an armchair observer, I am not in a position to dictate what is the best way forward, that must come from you guys doing the work.
QuoteIt took me an entire day to build a robotic H sensor positioner and make 182 of these H-field measurements for you.
I do appreciate all that you and Itsu are doing, and I am sorry if my response offended you. I am frustrated that I can't get hands-on experience to tell us what to do next and rebelled against being led down a more complicated route. Please accept my apology.
Smudge
Still stuck waiting for my litz wire.
I have one former ready for a 10 turns weave 1 type coil with 1mm nylon separation between turns.
So i hope to make some measurements soon.
Regards Itsu
FWIW, I ran an experiment with my two flat pcb coils in parallel but wired in bucking mode such that the voltage differential between opposite coil turns is zero. In this mode, I could measure appreciable voltage/current in an H field sense coil of any shape with differential probes. The problem is, the equivalent inductance is now ~1/2 each coil thus raising the resonance frequency.
A possible solution may be a current multiplier as described in the following link-
https://www.accelinstruments.com/Applications/WaveformAmp/Magnetic-Field-Generator.html
Here, the resonance could be controlled with Cs and Cp. Just a thot.
Regards,
Pm
The article does not mention whether the inter-turn capacitance CIT is the parallel capacitance CP depicted on the schematic diagram below:
(https://www.accelinstruments.com/wpimages/wp7481c428_06.png)
This is important because any current flowing through the CIT distorts the direction of the magnetic field generated by the turns of the coil. A current flowing in an external parallel capacitance, does not do that.
Quote from: verpies on 2020.09.02, 00:05:47
The article does not mention whether the inter-turn capacitance CIT is the parallel capacitance CP depicted on the schematic diagram below:
(https://www.accelinstruments.com/wpimages/wp7481c428_06.png)
This is important because any current flowing through the CIT distorts the direction of the magnetic field generated by the turns of the coil. A current flowing in an external parallel capacitance, does not do that.
I assumed Cp was an external and have both bench built and simulated the concept with success but at lower frequencies than this NMR requires.
Any large Cit could be improved as you suggested by gaps in between the turns on the flat coil.
What amazed me was the difference in my tests between the series connected bucking coils with the facing windings going the same direction (no differential H field detected) and the parallel connected bucking coils with the facing windings counterwound (excellent differential H field detection)!
Regards,
Pm
Quote from: partzman on 2020.09.02, 13:51:37
What amazed me was the difference in my tests between the series connected bucking coils with the facing windings going the same direction (no differential H field detected) and the parallel connected bucking coils with the facing windings counterwound (excellent differential H field detection)!
I just realized, that I might understand the phrase "differential H field " differently than you had meant it. There are two plausible interpretations of that phrase in this context:
1) An amplitude of the difference between two H fields' magnitudes occuring at two different places or two different spatial directions (or two different times).
2) An amplitude of a single H field measured by an unloaded sensor coil connected to an oscilloscope with two probes that have floating grounds at the probes' ends*, which sense voltage amplitudes that are later subtracted.
*
but these two grounds are connected together at the scope.
Quote from: verpies on 2020.09.07, 08:33:43
I just realized, that I might understand the phrase "differential H field " differently than you had meant it. There are two plausible interpretations of that phrase in this context:
1) An amplitude of the difference between two H fields' magnitudes occuring at two different places or two different spatial directions (or two different times).
2) An amplitude of a single H field measured by an unloaded sensor coil connected to an oscilloscope with two probes that have floating grounds at the probes' ends*, which sense voltage amplitudes that are later subtracted.
* but these two grounds are connected together at the scope.
Sorry, I should have been more distinct. Your option 2) above is what I was meaning. However, the sensor coils were loaded with 100k down to 1K resistors and exhibited power output within ~90% of the input power to the bucking coils.
Pm
Ordered Litz wire parcel officially went missing :(
New one is on its way
Itsu
:)
One day after the local PostOffice confirmed to me the parcel was missing (no status update since 12 Aug.) and it was confirmed missing by the Senders Post Organisation, it arrived this morning 11 Sept.
Unfortunately the sender already send a replacement, so unnecessary costs are being made.
Itsu
First prototype coil made as weave-1 with 1mm nylon spacing.
Id=81mm, Od=120mm, 1mm diameter litz wire 200 strands x 0.04m.
Further measurements will follow.
Itsu
Quote from: Itsu on 2020.09.12, 20:09:54
First prototype coil made as weave-1 with 1mm nylon spacing.
You are getting better at this - I like the quality.
Please make the input to this coil using RF techniques. No screwy terminals and long leads...
Quote from: Itsu on 2020.09.12, 20:09:54
Further measurements will follow.
Because this coil is expected to have a high impedance at 3-4MHz please use the Series Measurement (https://www.overunityresearch.com/index.php?topic=3924.msg83438#msg83438) topology with your VNA for the first measurement.
This will give you the impedance of the coil or S21 vs. frequency, but it will not tell you the H-field amplitude vs. frequency, because the RF current does not have to flow through the coil's windings (even when they are spaced apart).
To obtain the ratio of H-field amplitude / Input voltage amplitude vs. frequency, you'll need to use your H-field probe on the receiving port in S21 measurement, to sense the generated H-field.
With only one coil, orient the H-Field probe for maximum amplitude. However, with two coils, orient it to sense the magnetic flux in the radial direction (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014).
Quote from: verpies on 2020.09.12, 23:46:36
[clip]
However, with two coils, orient it to sense the magnetic flux in the radial direction (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014).
Is this correct? Wouldn't you want to measure the H field in the axial direction? Or are we saying there is a difference between the flux field of the coils and the H field?
Regards,
Pm
I made some initial measurements on the new coil.
Number of turns is 10
Inductance measured at 100Khz = 16uH
Resistance = 0.4 Ohm
nanoVNA shows the below output using the shown connection to the coil (series measurement).
Itsu
Quote from: partzman on 2020.09.13, 19:48:52
Is this correct? Wouldn't you want to measure the H field in the axial direction?
No, the flux from the permanent magnets is in the axial direction.
The flux from the coils must be perpendicular to that axial flux, it must be e.g. Radial. To be most sensitive in the radial direction, the sensing loop must be oriented as pictured in orange color on this diagram (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014).
The video of my robotic positioner had the sensing loop oriented like that.
Quote from: partzman on 2020.09.13, 19:48:52
Or are we saying there is a difference between the flux field of the coils and the H field?
We are not.
Quote from: Itsu on 2020.09.13, 20:09:24
I made some initial measurements on the new coil.
Number of turns is 10
Inductance measured at 100Khz = 16uH
Resistance = 0.4 Ohm
Does the length of the wire used to wind this coil make sense as the one responsible for that 28MHz peak at Marker2, when you treat it as a shorted transmission line of half that length ?
Would you be willing to cut the coil at the middle (like I did with mine) and solder a 220Ω pot in the break to terminate it with a matching resistance that will inhibit reflections ?
BTW: you could measure the C
IT directly this way, too.
P.S.
Does anyone remember the round trip time of that shorted coax in the Dally's contraption ?
QuoteDoes the length of the wire used to wind this coil make sense as the one responsible for that 28MHz peak at Marker2, when you treat it as a shorted transmission line of half that length ?
I started with a piece of 5m litz wire, and ended up cutting 88cm of both ends, so the coil itself is about 324cm.
Half of that is 162cm, so i do not see a reason there for the 28MHz (10m wavelength) peak.
QuoteWould you be willing to cut the coil at the middle (like I did with mine) and solder a 220Ω pot in the break to terminate it with a matching resistance that will inhibit reflections ?
BTW: you could measure the CIT directly this way, too.
I can try, but i am not sure it can be done now as i used some superglue at that start point.
I might destroy more then i like.
QuoteP.S.
Does anyone remember the round trip time of that shorted coax in the Dally's contraption ?
I don't, and i think it was never shown/mentioned.
Itsu
Quote from: verpies on 2020.09.13, 20:55:10
No, the flux from the permanent magnets is in the axial direction.
The flux from the coils must be perpendicular to that axial flux, it must be e.g. Radial. To be most sensitive in the radial direction, the sensing loop must be oriented as pictured in orange color on this diagram (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36014).
The video of my robotic positioner had the sensing loop oriented like that.
We are not.
OK, thanks. I really meant co-axial but that doesn't change your answer any. I don't want to take away from this thread but I will run tests to see why I detect a very strong co-axial H field in my setup as it could not be from the theoretical displacement current IMO as the voltage differential between adjacent turns is essentially zero.
Regards,
Pm
Before splitting the coil in 2, i first wanted to try some H-field measurements.
QuoteTo obtain the ratio of H-field amplitude / Input voltage amplitude vs. frequency, you'll need to use your H-field probe on the receiving port in S21 measurement, to sense the generated H-field.
With only one coil, orient the H-Field probe for maximum amplitude. However, with two coils, orient it to sense the magnetic flux in the radial direction.
See below video on how i tried it, not sure its the correct way, especially the calibration of the "thru" part and to what value's i should look.
Video here: https://www.youtube.com/watch?v=xJtPG3lDiEM
Itsu
Quote from: Itsu on 2020.09.14, 08:11:05
I started with a piece of 5m litz wire, and ended up cutting 88cm of both ends, so the coil itself is about 324cm.
Half of that is 162cm, so i do not see a reason there for the 28MHz (10m wavelength) peak.
According to basic Ham math for quarter-wavelength stubs:
f[MHz] = 75[m/MHz] * Vf / L[m]
At 0.8 velocity factor (Vf):
A shorted 162cm stub will appear as an open at 37MHz, 111MHz, 185MHz, ...and all of the odd harmonics.
A shorted 162cm stub will appear as a short at 74MHz, 148MHz , 222MHz, ...and all of the even harmonics.
However at 0.6 velocity factor (Vf):
A shorted 162cm stub will appear as an open at 28MHz, 83MHz, 139MHz, ...and all of the odd harmonics.
A shorted 162cm stub will appear as a short at 55MHz, 111MHz, 167MHz , ...and all of the even harmonics.
...all rounded up to the nearest MHz.
Quote from: Itsu on 2020.09.14, 20:10:10
See below video on how i tried it, not sure its the correct way, especially the calibration of the "thru" part and to what value's i should look.
Yes H-field "thru" calibration is difficult because you do not have a device that converts constant voltage amplitude (from the VNA) to constant amplitude H-field at various frequencies.
If your VNA had a constant current amplitude output then this would not be a problem.
Take a look at my attempt to do that some pages ago, where I used two identical 10mm shielded loops parallel to each other. In that attempt the constant voltage amplitude signal from the VNA's transmitter obviously was inducing less and less H-field amplitude in the transmitting coil as the frequency was increasing (because the coil's impedance was increasing with frequency), so it was behaving like a low pass filter. The receiver coil had the opposite characteristic. Your H-field probe is flat, because it is compensated internally up to 70MHz. You proved it by clamping the i-Probe on 15mm of Litz wire, which was shorting the VNA's transmitter....or was it the SA's TG ?
Anyway, fudging the "thru" by as a straight-through electric connection and assuming, that the H-field probe is flat, still lets us see the relative magnitudes of the H-Field and it is obvious that the field is the strongest on the middle of the winding (not to be confused with the center of the coil) and at 27.5MHz.
With one coil, the orientation of the H-field probe at the middle of the winding should be such as to maximize the amplitude sensed by it, ...but not so close that the capacitive coupling can take place.
With two coils and this probe between them, its orientation should be radial like in my video, period.
Let's investigate why the H-field's amplitude is so much higher at 27MHz than at 3MHz. Perhaps we can utilize the phenomenon responsible for it to our advantage.
When you cut the coil in the middle (take a look at the photo of my coil where I did it) please use the picopulser, tee, fast scope and 220Ω pot, to do a simple TDR experiment and find out what the characteristic impedance and velocity factor are, when this winding is treated as a transmission line.
Quote from: verpies on 2020.09.15, 01:09:38
According to basic Ham math for quarter-wavelength stubs:
f[MHz] = 75[m/MHz] * Vf / L[m]
At 0.8 velocity factor (Vf):
A shorted 162cm stub will appear as an open at 37MHz, 111MHz, 185MHz, ...and all of the odd harmonics.
A shorted 162cm stub will appear as a short at 74MHz, 148MHz , 222MHz, ...and all of the even harmonics.
However at 0.6 velocity factor (Vf):
A shorted 162cm stub will appear as an open at 28MHz, 83MHz, 139MHz, ...and all of the odd harmonics.
A shorted 162cm stub will appear as a short at 55MHz, 111MHz, 167MHz , ...and all of the even harmonics.
...all rounded up to the nearest MHz.
OK, found the formula:
http://www.66pacific.com/calculators/full-wave-loop-antenna-calculator.aspx
L = 75 * Vf / F or
F = 75 * Vf / L
But the key factor here is the velocity factor.
How do we know that for this SINGLE litz wire, i mean is the nylon inbetween the turns the dielectric causing the slowing effect?
https://www.picwire.com/resources/technical-articles/velocity-factor/ mentions:
Material c @ 1.0GHz VOP Delay (ns)
Nylon 4.5-3.6 47-53% 2.16-1.92
Itsu
Quote from: verpies on 2020.09.15, 01:10:23
Yes H-field "thru" calibration is difficult because you do not have a device that converts constant voltage amplitude (from the VNA) to constant amplitude H-field at various frequencies.
If your VNA had a constant current amplitude output then this would not be a problem.
Take a look at my attempt to do that some pages ago, where I used two identical 10mm shielded loops parallel to each other. In that attempt the constant voltage amplitude signal from the VNA's transmitter obviously was inducing less and less H-field amplitude in the transmitting coil as the frequency was increasing (because the coil's impedance was increasing with frequency), so it was behaving like a low pass filter. The receiver coil had the opposite characteristic. Your H-field probe is flat, because it is compensated internally up to 70MHz. You proved it by clamping the i-Probe on 15mm of Litz wire, which was shorting the VNA's transmitter....or was it the SA's TG ?
Anyway, fudging the "thru" by as a straight-through electric connection and assuming, that the H-field probe is flat, still lets us see the relative magnitudes of the H-Field and it is obvious that the field is the strongest on the middle of the winding (not to be confused with the center of the coil) and at 27.5MHz.
With one coil, the orientation of the H-field probe at the middle of the winding should be such as to maximize the amplitude sensed by it, ...but not so close that the capacitive coupling can take place.
With two coils and this probe between them, its orientation should be radial like in my video, period.
Let's investigate why the H-field's amplitude is so much higher at 27MHz than at 3MHz. Perhaps we can utilize the phenomenon responsible for it to our advantage.
When you cut the coil in the middle (take a look at the photo of my coil where I did it) please use the picopulser, tee, fast scope and 220Ω pot, to do a simple TDR experiment and find out what the characteristic impedance and velocity factor are, when this winding is treated as a transmission line.
Yes, it was the SA's TG:
https://www.overunityresearch.com/index.php?topic=3924.msg83003#msg83003
Ok about cutting the coil in the middle:
https://www.overunityresearch.com/index.php?topic=3924.msg83295#msg83295
I will try to put a potmeter there, because as mentioned in my earlier post, knowing the characteristic impedance and velocity factor seems important.
So i treat the both single litz wires as a "cable" terminated by the 220 Ohm pot and adjusting for minimum reflections (impedance), right?
Itsu
Quote from: Itsu on 2020.09.15, 08:18:11
How do we know that for this SINGLE litz wire, i mean is the nylon inbetween the turns the dielectric causing the slowing effect?
When you treat this coil as a
pair of Litz wires with the Nylon dielectric between them, then you get a coiled
Twin-Lead transmission line shorted at the end. The dielectric of the transmission line influences its velocity factor greatly.
If you still have the Nylon fishing line spacer between the wire turns, remove it to leave air gaps behind. Air is a better dielectric than Nylon. Cyanoacrylate glue bonds very well to acrylic/Plexi, but
not to Nylon.
Quote from: Itsu on 2020.09.15, 08:43:26
So i treat the both single litz wires as a "cable" terminated by the 220 Ohm pot and adjusting for minimum reflections (impedance), right?
Yes, also try to measure precisely the round trip delay of these reflections with a fast scope and the Picopulser using RF techniques, because this will give us the velocity factor. To maintain constant impedance, try to keep the leads to the coil spaced with the same gap as between the winding turns.
Quote from: Itsu on 2020.09.15, 08:43:26
Yes, it was the SA's TG:
https://www.overunityresearch.com/index.php?topic=3924.msg83003#msg83003
I guess that was when you did not have the VNA yet.
You might want to repeat this experiment by clamping the i-Probe on a piece
* of Litz wire, which shorts the VNA's transmitter.
*
as short as possible !
QuoteWhen you treat this coil as a pair of Litz wires with the Nylon dielectric between them, then you get a coiled Twin-Lead transmission line shorted at the end. The dielectric of the transmission line influences its velocity factor greatly.
If you still have the Nylon fishing line spacer between the wire turns, remove it to leave air gaps behind. Air is a better dielectric than nylon. Cyanoacrylate glue bonds very well to acrylic/Plexi, but not to Nylon.
OK, yes i still have the nylon between the litz wire.
I did remove the nylon which occupied the space where now the litz wire is and it did not came of easily.
QuoteYes, also try to measure precisely the round trip delay of these reflections with a fast scope and the Picopulser using RF techniques, because this will give us the velocity factor. To maintain constant impedance, try to keep the leads to the coil spaced with the same gap as between the winding turns.
Ok, will try
QuoteI guess that was when you did not have the VNA yet.
You might want to repeat this experiment by clamping the i-Probe on a piece* of Litz wire, which shorts the VNA's transmitter.
Yes before i had the nanoVNA, so i can try it with the nanoVNA.
Managed to remove the nylon spacers, but it weakens the whole pancake coil somewhat.
Had to restore some loosened litz with superglue.
Toke a nanoVNA scan, but somehow the saver program does not accept some values as i get this on connecting to the device:
2020-09-17 10:26:07,192 - NanoVNASaver.SweepWorker - WARNING - Got a non-float data value: -9.811650276 -0.571220398 (-9.811650276)
2020-09-17 10:26:07,393 - NanoVNASaver.SweepWorker - CRITICAL - Tried and failed to read data 1 20 times. Giving up.
So i only could take a screenshot of the nanoVNA device itself, see screenshot below.
Seems the resonance has shifted upwards to 9Mhz.
Anyway, will cut the middle and install a 220 Ohm pot to do some TDR measurements.
Regards Itsu
Quote from: Itsu on 2020.09.17, 08:46:36
2020-09-17 10:26:07,192 - NanoVNASaver.SweepWorker - WARNING - Got a non-float data value: -9.811650276 -0.571220398 (-9.811650276)
2020-09-17 10:26:07,393 - NanoVNASaver.SweepWorker - CRITICAL - Tried and failed to read data 1 20 times. Giving up.
Most likely that is an inferior USB cable.
A good USB cable should have 90Ω impedance and thick conductors. It is hard to identify these without instruments and a good test jig.
Also, what is this "DISK: 99.9%" message ?
Quote from: Itsu on 2020.09.17, 08:46:36
So i only could take a screenshot of the nanoVNA device itself, see screenshot below.
Seems the resonance has shifted upwards to 9Mhz.
Not only that but the linear S21 measurement shows the amplitude to be decreasing as the frequency increases almost monotonically. The high frequency peak at ~32MHz is still there but it is much lower than the peak at low frequencies, which is as it should be because the reactance of an ideal inductor increases with frequency.
How was the H-probe oriented and and how was it positioned with respect to the winding ?
QuoteThat most likely is a inferior USB cable.
A good USB cable should have 90Ω impedance and thick conductors. It is hard to identify these without instruments and a good test jig.
Also, what is this 99.9% disk message ?
So you mean the usb cable that connects between the nanoVNA and my PC? Its the one supplied by the nanoVNA, but i will try another one.
That 99.9% disk message seems to be the free space on an internal storage.
QuoteNot only that but the linear S21 measurement shows the amplitude to be decreasing as the frequency increases almost monotonically. The high frequency peak is much lower than the peak at low frequencies.
How was the H-probe oriented and and how was it positioned with respect to the winding ?
I was using the series measurement, so not the H-field probe.
Itsu
Quote from: Itsu on 2020.09.17, 09:48:07
So you mean the usb cable that connects between the nanoVNA and my PC?
Yes
Quote from: Itsu on 2020.09.17, 09:48:07
I was using the series measurement, so not the H-field probe.
Right, I should have seen it.
Have put off "validate data" in the saver programm and now it continues.
Below is the same layout data as with the nylon spacers output shown a few posts above (serial measurement):
Itsu
Below screenshot is with the H-Field probe.
Now i see a strong 5Mhz peak
Quote from: Itsu on 2020.09.17, 12:34:13
Below screenshot is with the H-Field probe.
Now i see a strong 5Mhz peak
Yeah!, I see it too. It bothers me because it means that for some reason the H-field has over 7x higher amplitude at 5MHz compared to 1MHz - that's 50x more power, where we are expecting less power !
According to MPTT, in order to obtain maximum power transfer to a reactive load, the reactance of the source should be of equal magnitude but opposite sign. ..but the VNA's transmitter source impedance is 50Ω+0j, isn't it ?
There wasn't anything unusual going on at 5MHz when you were doing the purely electric series measurement before.
What was the H-probes's orientation and position when you acquired that signal ?
Does the H-probe have a flat response according to the VNA?
You can check it by clamping the i-Probe on a piece* of Litz wire, which shorts the VNA's transmitter.
*
as short as possible !
Quote from: Itsu on 2020.09.17, 12:11:12
Have put off "validate data" in the saver program and now it continues.
So you turned off the error detection and are ignoring the data transmission errors.
Anyway, is the green graph the Real and the blue graph the Imaginary component on your S21 Real/Imaginary plot ?
QuoteYeah!, I see it too. It bothers me because it means that for some reason the H-field has over 7x higher amplitude at 5MHz compared to 1MHz - that's 50x more power, where we are expecting less power !
According to MPTT, in order to obtain maximum power transfer to a reactive load, the reactance of the source should be of equal magnitude but opposite sign. ..but the VNA's transmitter source impedance is 50Ω+0j, isn't it ?
I guess the problem is the calibration on "thru" with the H-Field probe.
I again calibrated it in thru in the middle, but could it be that the H-field is distributed differently without the nylon?
Yes, the VNA transmitter suppose to be 50 Ohm.
QuoteThere wasn't anything unusual going on at 5MHz when you were doing the purely electric series measurement before.
What was the H-probes's orientation and position when you acquired that signal ?
No, nothing unusual going on, no equipment other then the nano and PC was on.
H-probe was as in the last video at 2:23, see picture below.
QuoteDoes the H-probe have a flat response according to the VNA?
You can check it by clamping the i-Probe on a piece* of Litz wire, which shorts the VNA's transmitter.
I will check again later today the "serial measurement", including the flat response of the VNA.
QuoteSo you turned off the error detection and are ignoring the data transmission errors.
Yes, seems like it
QuoteAnyway, is the green graph the Real and the blue graph the Imaginary component on your S21 Real/Imaginary plot ?
Well, thats not clear from the Graph, as it does not specify, but looking at the nanoVNA itself, the blue is the real, the green the imaginary.
Itsu
QuoteDoes the H-probe have a flat response according to the VNA?
You can check it by clamping the i-Probe on a piece* of Litz wire, which shorts the VNA's transmitter.
Below graphs show the sweep from 10KHz to 100MHz, so fairly flat.
Quote from: Itsu on 2020.09.17, 15:38:02
I guess the problem is the calibration on "thru" with the H-Field probe.
Yes, I have the same problem.
I just do the H-probe "thru" calibration by clamping the i-Probe on a very-very short bundle of 4 thick Litz wires, that short the VNA's transmitter through a custom N fixture.
If anyone knows a better way to convert a constant voltage amplitude signal to a constant amplitude H-field - speak up.
Quote from: Itsu on 2020.09.17, 16:03:19
Below graphs show the sweep from 10KHz to 100MHz, so fairly flat.
The │S21│ vertical scale is too coarse (1.25U / div.) to see any variations but the logarithmic scale shows <2dB variation up to 50MHz, so not bad.
Quote from: verpies on 2020.09.17, 18:09:38
The │S21│ vertical scale is too coarse (1.25U / div.) to see any variations but the logarithmic scale shows <2dB variation up to 50MHz, so not bad.
I had to manually adjust the Data portion to get some resolution.
The funny thing is i had to use a comma as decimal point instead of the usual american period as that would "stick" to a max scale of 1 :D
Itsu
It's great to have that vertical scaling under control.
The i-Probe's response is still too flat to explain that H-field's amplitude peak at 5MHz.
Does the S21 Impedance in the purely electric Series Measurement have its complex component go to zero at 5MHz ?
Quote from: verpies on 2020.09.17, 18:51:41
It's great to have that vertical scaling under control.
The i-Probe's response is still too flat to explain that H-field's amplitude peak at 5MHz.
Does the S21 Impedance in the purely electric Series Measurement have its complex component go to zero at 5MHz ?
Hmmm, not sure what you mean by "the S21 Impedance", but there is nothing going to zero at 5Mhz in the impedance related graphs.
See below the possible S21 graps i can select (the bottom one is TDR), so only real and imaginary.
The possible 3 S11 inpedance graphs i have selected next to this S21 one are shown in the screenshot.
Its a freshly calibrated Series Measurement sweep, marker 1 (red) is at 5MHz.
Itsu
Quote from: Itsu on 2020.09.17, 19:22:16
Hmmm, not sure what you mean by "the S21 Impedance",
The "S11 R+jX (Ω)", is what I would call "S11 Impedance". You have that plot displayed, but we don't need it in the Series Measurement.
What we need is "S21 R+jX (Ω)", which is what I would call "S21 Impedance".
"S21 R+jX (Ω)" plot is different from the "S21 Real/Imaginary" plot because the latter depicts a
ratio of voltage amplitude at the receiver divided by voltage amplitude at the transmitter.
Notice, that the units of that ratio are V/V or just Unity ( often written as: U or mU or kU ), not (Ω) like for Impedance.
Internally, the VNA makes the same measurement for both, it just converts it to the real Ohms of the resistance (R) and imaginary Ohms of the reactance (jX), which are components of the complex impedance.
Right, thanks,
so it seems the Saver program, does not have the option to show this "S21 R+jX (Ω)" impedance plot.
The nanoVNA itself has something called "resistance" and "reactance" for both S11 as S21.
The below screenshot shows them for S21, is that useable?
Nothing at 5Mhz either though.
Itsu
Quote from: Itsu on 2020.09.18, 07:27:00
so it seems the Saver program, does not have the option to show this "S21 R+jX (Ω)" impedance plot.
Nooooo, it can show "S11 R+jX (Ω)" but not "S21 R+jX (Ω)" ?!!!! WTF ?!! If I were you, I would gripe about this lack of symmetry to the developers of this program.
Quote from: Itsu on 2020.09.18, 07:27:00
The nanoVNA itself has something called "resistance" and "reactance" for both S11 as S21.
As any VNA should.
Quote from: Itsu on 2020.09.18, 07:27:00
The below screenshot shows them for S21, is that usable?
Yes, it shows that the DUT acts capacitively below 10MHz and above 32MHz. I find the former hard to believe
*.
Between 10MHz and 32MHz it shows inductive behavior (positive reactance).
Also, the reactance increases monotonically from 0-32MHz - as it should.
When dealing with reactance [X] plots (phase plots, too), you should place the markers where it crosses the ZERO or the horizontal axis. On that plot it would be around 10MHz and 32MHz.
What values of X and R do you get when you zoom 0-10MHz and place the markers at 5MHz ?
*
This is most likely due to the Series Measurement having a huge error for impedances around zero. The Shunt-Thru (https://www.overunityresearch.com/index.php?topic=3924.msg83438#msg83438) measurement has a smaller error at low impedances. Only a splice of these two measurements would give you a complete and true plot of the DUT's impedance.
QuoteNooooo, it can show "S11 R+jX (Ω)" but not "S21 R+jX (Ω)" ?!!!! WTF ?!! If I were you, I would gripe about this lack of symmetry to the developers of this program.
There are 3 PC programms i can find dealing with the nanoVNA, but none is able to display that "S21 R+jX (Ω)" graph :(
QuoteWhat values of X and R do you get when you zoom 0-10MHz and place the markers at 5MHz ?
See screenshot:
Yeah, it is highly unlikely that the reactance is negative (capacitive) at 5MHz with these 1mm air gaps and the real resistance is so close to 50Ω.
If the inductance of this coil is still 16µH after the removal of the nylon spacers, then its inductive reactance at 5MHz should be above +500Ω (because XL=2πfL).
Below 50Ω, the Shunt-Trough Measurement is more accurate. Above 50Ω the Series Measurement is more accurate. At 5 MHz we are expecting more than 500Ω.
We have a huge measurement error somewhere and the amplitude of the H-field at 5MHz supports it...
I measured the coil still to be 16uH without the nylon spacers using my LCR meter.
New calibration and using still the "series measurement" shows the same above (10KHz-10Mhz) output, using the nanoVNA itself, so not the PC programm with data validate disabled.
Itsu
QuoteNooooo, it can show "S11 R+jX (Ω)" but not "S21 R+jX (Ω)" ?!!!! WTF ?!! If I were you, I would gripe about this lack of symmetry to the developers of this program.
When i posted the request for the S21 R+jX graph on the nanoVNA forum, this is the first response i got just now:
S21 is a measure of device transmission characteristics. R+jX is an impedance measurement and that is why it can be displayed as a S11 option but not S21.
I supposed if you used an external bridge then S21 could be used to measure R+jX, but no program currently available for the NanoVNA has an external bridge option.
Itsu
Quote from: Itsu on 2020.09.18, 16:17:03
S21 is a measure of device transmission characteristics.
Impedance also affects transmission when the DUT is in series. See:
https://youtu.be/0JglLhrisOs
Quote from: Itsu on 2020.09.18, 16:17:03
I supposed if you used an external bridge then S21 could be used to measure R+jX,
Yet the NanoVNA displays this plot (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36297).
Here (https://www.keysight.com/main/editorial.jspx?cc=CA&lc=eng&ckey=1536926&nid=-11143.0.00&id=1536926) is a proof from one of the big boys (Keysight) that Transmission Impedance (Z
T) is a thing and it can be calculated ONLY from the S21 scattering and Z
0 (system impedance. Usually 50Ω).
In other words: If you can measure S21 and know Z
0, then you can calculate the S21 Impedance.
Please remove if inappropriate
Here a comment from fellow I have great appreciation for [many years]and has some thoughts/experience ? [good] on topic here.
Turbo
Quote
That last pic is good but to drive like that is critical in terms of timing even the connecting leads have to be cut to match half and quarter wavelengths and its a key element in separating and extracting and decoupling the in and out coming signals from each other without overloading the pre amp.
Both should be close to exact opposites as possible on the Smith chart.
end quote
PS
as mentioned I will refrain from cross posting !
from Mirror topic here
https://overunity.com/18502/nuclear-magnetic-resonance-nmr/msg550686/#new
Quote from: Chet K on 2020.09.18, 17:27:52
That last pic is good but to drive like that is critical in terms of timing even the connecting leads have to be cut to match half and quarter wavelengths
Do you know which pic he is referring to ?
Anyway, he is correct about the lengths of the interconnects, but I will counter that with the statement, that the influences of these lengths are calibrated out by the VNA.
Quote from: Chet K on 2020.09.18, 17:27:52
Both should be close to exact opposites as possible on the Smith chart.
He is correct that the maximum power transfer occurs when the impedance of the source and load are complex conjugates...but we need to know them first before we can calculate the conjugate and right now Itsu is getting bogus reactance (X) and resistance (R) values at the 5MHz.
The matter of the maximum power transfer is further complicated by the fact that in NMR we are NOT after the maximum power dissipation in the coil/antenna (as it would be in a Ham radio).
Instead, we are after the maximum H-field amplitude in the radial direction and that means that near fields and reactive currents count towards that goal.
Maybe one of the RF connectors is open (inside or outside of the VNA). An Ohmmeter might be able to detect it.
I measured the paths from the nanoVNA up to the coil with a DMM and all connectors measure OK (0 Ohm through, infinity inner to outer).
The nanoVNA TX port measures 58 Ohm, the RX port 50 Ohm.
I used several different (length of) coax from nanoVNA to the coil, but the screenshot signals does not change from the last one shown above.
Itsu
Quote from: Itsu on 2020.09.18, 19:29:37
I measured the paths from the nanoVNA up to the coil with a DMM and all connectors measure OK (0 Ohm through, infinity inner to outer).
The nanoVNA TX port measures 58 Ohm, the RX port 50 Ohm.
No easy answers then >:(
I surmise that, when you measure between the center conductors of the two semi-rigid coax cables (the ends that plug into the VNA) then you get 0.5Ω.
How does it look when you solder a 510Ω SMD resistor in place of your coil ?
A 510 Ohm smd resistor he! I will take one from my stack of 510 Ohm smd resistors :D
Quote from: Itsu on 2020.09.18, 20:00:11
A 510 Ohm smd resistor he! I will take one from my stack of 510 Ohm smd resistors :D
Yeah, like from this resistor kit
470Ω will be fine, too.
Can your VNA display phase angle?
Can you make a phase vs. frequency plot like this one (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36043) ?
I used a 560 (558 measured) Ohm metal film resistor instead for now.
First screenshot is with te saver PC program,
second screenshot from the nanoVNA itself like the earlier screenshots.
Itsu
Quote from: verpies on 2020.09.18, 20:29:22
Can your VNA display phase angle?
Can you make a phase vs. frequency plot like this one (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36043) ?
Yes it can, from the coil you mean?
Here the plots from S21 phase and Lin mag from the saver program (upper 2 graps) and the nanoVNA itself from the spaced coil (series measurement):
Quote from: Itsu on 2020.09.18, 20:29:46
I used a 560 (558 measured) Ohm metal film resistor instead for now.
Must be good enough
Quote from: Itsu on 2020.09.18, 20:29:46
First screenshot is with the saver PC program,
The resistance (Series R) is 601.85Ω, which is in the general ballpark of 560Ω.
The capacitance (Parallel C) is small but not small enough - it should be close to zero for a resistor.
The inductance (Series L) is negative :o ...but that is nothing next to the units of Reactance (X), which are displayed as Farads ! :o :o :o
Quote from: Itsu on 2020.09.18, 20:29:46
second screenshot from the nanoVNA itself like the earlier screenshots.
The reactance (X) is close to zero Ohms, as it should be for a good resistor ...we also have an improvement in the Units of Reactance department.
But the real resistance (R) is displayed as 68.5Ω. That's almost an order of magnitude smaller than the value displayed by the PC program ! ...and wrong, too.
Quote from: Itsu on 2020.09.18, 20:43:35
Here the plots from S21 phase and Lin mag from the saver program (upper 2 graps) and the nanoVNA itself from the spaced coil (series measurement):
This just gets better and better.
In this plot (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36315), the phase is negative( less than 0º ) from 0.1-10MHz - that's correct for an inductor (ask ELI the ICE man).
But wait!, in the previous plot (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36297) the reactance was negative for
the same coil from 0.1-10MHz, which means capacitive behavior. Shouldn't the phase be greater than 0º for a capacitor ? ...please check my homework !
You've got to plot the negative S21 reactance and negative S21 phase on the same screen and show it to the developers.
I am posting S21 Reactance and S21 Phase plots of a real capacitor and a real inductor measured by a real VNA (using this fixture (https://www.overunityresearch.com/index.php?action=dlattach;topic=3924.0;attach=36067)) in case anyone has any doubts how they are supposed to look like.
Note the opposite signs of the reactance and phase for both of them.
I have to sit down and take a good look at my calibration setup as the more i fiddle around the more confused i get.
I will take that 560 Ohm resistor as a base and aim for a resulting value of around 560 Ohm not 68 is it shows now.
By the way, the series X and parallel X sometimes displays as Henries or Farad, so probably a bug.
Itsu
Guess i have to skip those S21 measurements as they seem not reliable.
Will start tonight with some TDR measurements.
Itsu
Using my picopulser, i see a reflection at 9.8ns which disappears when turning the pot.
It however disappears in the noise, so hard to know when the correct impedance match is there.
Also, when the pot is at 0 Ohm, the reflection pulse is positive instead of negative which it should be.
Anyway, 9.8ns for a round trip means 4.9ns for the cable/coil length.
Speed of light in vacuum is 299 792 458 m/s = 0.299792m/ns
0.299792 * 4.9 = 1.47m.
the cable/coil is 1.62m long, so velocity factor should be 0.90.
The resistance measured when the reflectionpuls disappears in the noise is around 175 Ohm, so close to the pots upper limit.
Guess something is not right here, so will recheck tomorrow.
That velocity factor is plausible with the air dielectric.
Also, the characteristic impedance of 175Ω is not that far off from my coil's, which is not air gapped.
At 8.9ns round trip time the transmission line half-wave resonance (when the reflection can be 180º out o phase) would occur at 56MHz ...and 28MHz is half of that.
This still does not explain that peak in the H-field's amplitude at 5MHz.
Could your VNA's S21 measurement be that unreliable? You can verify it with a long FG's sweep and scope.*
* Calibration of that method can be done by clamping the i-Probe on a piece of SHORT Litz wire shorting the FG and saving the result of this FM sweep as csv. Later, the FM sweep of the coil with the H-probe can be normalized to that saved calibration csv.
Here a video of the TDR measurement setup: https://www.youtube.com/watch?v=q9PsTZWzdpQ
The H-field measurement i need to repeat as the "thru" calibration done was not correct.
I will use your calibration method mentioned above both for the nanoVNA as the FG/scope.
Not sure what you mean with saving as CSV file as the MDO3054 scope seems to have no save to CSV possibility (the SA does).
Itsu
Quote from: Itsu on 2020.09.20, 12:12:33
Here a video of the TDR measurement setup: https://www.youtube.com/watch?v=q9PsTZWzdpQ
The H-field measurement i need to repeat as the "thru" calibration done was not correct.
I will use your calibration method mentioned above both for the nanoVNA as the FG/scope.
Not sure what you mean with saving as CSV file as the MDO3054 scope seems to have no save to CSV possibility (the SA does).
Itsu
Itsu,
The MDO series will save waveforms as a csv file in the following manner- First select the "Menu" button on the front panel and then press "Save Waveform". Plug a USB drive into the front panel and then from the Save Waveforms menu, select which channel you wish to save with the "A" knob and then select the "File" destination with the "B" knob which will save the waveform in a .csv format on your USB drive.
Regards,
Pm
PM, O0 thanks,
had to rename the .csv file to .rar to have it uploaded here.
After download, rename to .csv and it should work
Its just a test, so the waveform is from my picopulse test screen.
EDIT: i renamed the file to .pdf to avoid truncating
Itsu
Used the above suggested H-field probe thru calibration and made a nanoVNA sweep, see screenshot:
Itsu
Quote from: Itsu on 2020.09.20, 12:12:33
Here a video of the TDR measurement setup: https://www.youtube.com/watch?v=q9PsTZWzdpQ
Yes, the polarity of the reflected pulse seems to be off. That would explain why the peak was at 28MHz instead of 56MHz. This is weird and I'd like others to verify this conclusion.
Also, the amplitude ratio of the incident picopulse to the reflected one appears smaller than in my TDR experiments here (https://www.overunityresearch.com/index.php?topic=3924.msg83295#msg83295) and here (https://www.overunityresearch.com/index.php?topic=3924.msg83297#msg83297) and here (https://www.overunityresearch.com/index.php?topic=3924.msg83166#msg83166). I used a scope with 50Ω inputs to do the first two measurements and I think you are using HighZ inputs that are not compensated by a scope probe.
Quote from: Itsu on 2020.09.20, 12:12:33Not sure what you mean with saving as CSV file as the MDO3054 scope seems to have no save to CSV possibility (the SA does).
I did not know it cannot save the acquired data. I guess I will have to make Imagemagick count the pixels to obtain the calibration curve...
Click on the images below, to see them animate:
Yes, the amplitude of the reflected pulse seems low and has the wrong polarity somehow.
I have meanwhile changed the coils SMA connector to a BNC one to ommit the sma2bnc adapter, i used both the scope probes 50 Ohm and 1MOhm setting, but no significant change was seen,
so i guess the before measured 0.90 velocity factor and cable impedance of 175 Ohm holds.
This video is from the H-Field probe test mentioned 2 posts above:
https://www.youtube.com/watch?v=tuDoNOm8ZMA
Itsu
FG sweeps of the pancake coil when pot at 0 Ohm:
1st 10KHz to 100MHz (resonances at 8, 42 and 96MHz)
2th 10KHz to 50MHz (resonances at 8 and 42MHz)
3th 10KHz to 10MHz. (resonance at 7.9MHz)
FG has one big turn 5cm away from the pancake coil, scope probe at coils connection.
.CSV files attached (rename them to .csv after downloading).
Quote from: Itsu on 2020.09.20, 20:12:58
FG sweeps of the pancake coil when pot at 0 Ohm:
Apparently, it is an X amplitude vs. frequency relationship of the pancake coil.
What is X ?
Quote from: Itsu on 2020.09.20, 20:12:58
FG has one big turn 5cm away from the pancake coil, ...
Is it an inductive turn that is supposed to be a transmitter of an H-field ?
If "yes" - do you know its inductance and the transmitted H-field's amplitude vs. frequency relationship ?
Quote from: Itsu on 2020.09.20, 20:12:58
...scope probe at coils connection.
What is the capacitance of the probe ? Is it a HighZ probe ?
I have the feeling is misunderstood you with this FG sweep setup, but nevertheless, here some answers:
QuoteApparently, it is an X amplitude vs. frequency relationship of the pancake coil.
What is X ?
X at the 10MHz sweep is 52V, see screenshot.
QuoteIs it an inductive turn that is supposed to be a transmitter of an H-field ?
If "yes" - do you know its inductance and the transmitted H-field's amplitude vs. frequency relationship ?
Its more a capacitive coupled loop i think. Induction measured is 0.6uH @ 100KHz
QuoteWhat is the capacitance of the probe ? Is it a HighZ probe ?
3.9pF and 10Mohm.
QuoteI have the feeling is misunderstood you with this FG sweep setup, but nevertheless, here some answers:
I now used the H-field probe as with the nanoVNA to measure the H-Field with the scope.
See setup in picture.
Calibration was done with small single loop litz directly on the FG, see screenshot 1
Measurement was done with a 50MHz sweep, see screenshot 2.
We see a peak at 17Mhz and 32MHz.
Quote from: Itsu on 2020.09.21, 16:33:44
I now used the H-field probe as with the nanoVNA to measure the H-Field with the scope.
See setup in picture.
Calibration was done with small single loop litz directly on the FG, see screenshot 1
Measurement was done with a 50MHz sweep, see screenshot 2.
We see a peak at 17Mhz and 32MHz.
The hookups are OK, but there is some nasty aliasing of the RF signal with the scope's sample rate, which makes it difficult to process the csv files numerically.
Please test for the aliasing first with the FG connected directly to the scope. The signal should not disappear like the attached one at 3.10E-02s due to aliasing.
10s sweep time is fine, but please repeat it with an integration time set below the half of scope's sampling rate... that setting might be near the "bandwidth limiting" settings or sample averaging in acquisition menu.
For example my scope has this RMS mode in which up to 16M samples can be squared and added up inside the FPGA on the fly and then only their average (sqr actually) is written to scope's memory. I have seen other scopes having a similar acquisition mode where the absolute values of the samples are averaged arithmetically instead, and then only their average is written into scope's memory*. If your scope has a similar mode - please use it.
With such memory saving acquisition modes, e.g. 10G samples do not need to be written into scope's memory when the scope samples at e.g. 1Gsps for 10s and the high (1GSps) sampling rate prevents aliasing of all signals below 500MHz.
Also, some i-Probe amplifiers have an RMS mode and their output in this mode only needs to be low pass filtered at kHz frequencies to avoid aliasing.
*
In analog world this is equivalent to a high-frequency rectifier followed by an RC integrator.
OK, will have to do some investigation into this, thanks.
No such thing as " RMS mode" in my scope as far as i can find.
I do however can increase the "record length" from the present set 10K up to 10M, but doing so creates a large file like 166M compared to the earlier 195K.
Guess i have to find a compromize between record length and aliasing.
Will start with 1M record length etc. but this file is already 15.8M big, so only accessable from my FTP i guess (also the 10M record length file (166M) is there)
I uploaded the 100K record length file (1.5M) below again as .pdf, perhaps thats already useable.
Itsu
Quote from: Itsu on 2020.09.22, 11:30:30
No such thing as " RMS mode" in my scope as far as i can find.
What about some maximum or peak hold mode that takes the max of the last e.g. 1M samples and writes only that to the memory ?
Also regular sweep averaging might work if you trigger each sweep on a
noisy trigger signal ...it will take a long time but the file size will me small.
Quote from: Itsu on 2020.09.22, 11:30:30
I do however can increase the "record length" from the present set 10K up to 10M, but doing so creates a large file like 166M compared to the earlier 195K.
I could write you a small command line utility that, averages the absolute values of N samples and writes the result to a small output file.
Quote from: Itsu on 2020.09.22, 11:30:30
Guess i have to find a compromize between record length and aliasing.
I don't think you can avoid aliasing this way because if you want to do a 10s 0-50MHz sweep just like before, then according to Nyguist you need a minimum of 100MSps sampling rate to avoid aliasing.
100MSps * 10s = 1G samples ...and I doubt your memory is that deep.
Quote from: Itsu on 2020.09.22, 11:30:30
I uploaded the 100K record length file (1.5M) below again as .pdf, perhaps thats already useable.
I'll try to process it
The HP i-Probe amplifier, which I was repairing once, used the 6-pin chip AD8361 (https://www.analog.com/media/en/technical-documentation/data-sheets/AD8361.pdf) to generate its RMS output the analog way.
Quote from: verpies on 2020.09.22, 14:04:29
What about some maximum or peak hold mode that takes the max of the last e.g. 1M samples and writes only that to the memory ?
Also regular sweep averaging might work if you trigger each sweep on a noisy trigger signal ...it will take a long time but the file size will me small.
I could write you a small command line utility that, averages the absolute values of N samples and writes the result to a small output file.
I don't think you can avoid aliasing this way because if you want to do a 10s 0-50MHz sweep just like before, then according to Nyguist you need a minimum of 100MSps sampling rate to avoid aliasing.
100MSps * 10s = 1G samples ...and I doubt your memory is that deep.
I'll try to process it
I used "peak detect" then the 10K samples which seems much smoother.
File sizes are the same (166K for this one), so attached below.
Quote from: Itsu on 2020.09.22, 15:23:53
I used "peak detect" then the 10K samples which seems much smoother.
File sizes are the same (166K for this one), so attached below.
Wow, that is a weird one!
What is that frequency jump around 1s ?
Did the FG output that waveform or is it an aliasing artifact ?
I guess i did not waited long enough for triggering completed.
Here a 100K record length with peak detect waiting long enough:
Itsu
:( See attached.
Hmmm, i used a lower FG sweep time (200ms, scope 20ms) now and this looks better to me, see graph.
Not sure what is causing this truncating? after 1s, the FG or the scope.
Changing it from .csv to .pdf and back is not causing it as the original .csv also has it.
Below attached a 10Khz to 50Mhz sweep in 200ms with 100K record length with peak detect.
Itsu
Triggering from a noisy signal and averaging sweeps at low kSps also averages aliasing errors and produces small filesizes.
The FG can generate noise on a 2nd channel. It can be added resistively to the trigger signal.
Look how unfortunate triggering and low rate sampling in the upper case results in sampling all zeros (blue arrows)
In the lower case, the sampling starts at a different phase and results in different sampled points.
When the maximums of these samples are averaged over multiple sweeps, then the aliasing errors can be averaged out.
All that is needed is for the trigger signal's phase to be varied (by sawtooth, triangle, sine or by noise).
Ok, but i trigger from the FG external trigger point which is fixed (square wave) as i understand.
Anyway, here a 2s sweep (scope 200ms) which shows similar at the 1s point:
Quote from: Itsu on 2020.09.24, 09:30:27
Below attached a 10Khz to 50Mhz sweep in 200ms with 100K record length with peak detect.
And below are the first 32k samples of that sweep with their absolute values averaged out.
It looks better except for some weirdness at the beginning.
However you shortened the sweep length to 200ms to fit into memory and this is too much of the sacrifice.
Quote from: Itsu on 2020.09.24, 09:46:54
Ok, but i trigger from the FG external trigger point which is fixed (square wave) as i understand.
Yes, but you can resistively add a noise (or slow ramp) to that square wave so that the sampling starts several nanoseconds later (or earlier) for every sweep.
This way, when you just keep the peaks (maximums) from the multiple sweeps, the aliasing errors will disappear ...even with low sample rates and long sweep durations (still small file sizes)
I added noise from my 2th FG channel via a tee to the square trigger signal in various strength.
But the artifact after 1s still remain.
Looking at the data from the csv file i see the 4th digit behind the decimal point is being ignored after the 1s mark, see red digits in the picture (data changed into "numbers")
After 0.0099 and 0.0100 (1s) it should continue with 0.0101, then 0.0102 etc. not with repeated 0.0100 then skip to 0.0110 etc. i think.
Quote from: Itsu on 2020.09.24, 14:23:25
I added noise from my 2th FG channel via a tee to the square trigger signal in various strength.
That should have varied the starting phase of the sampling sequence.
Maybe the square wave is too steep. Try adding a small RC integrator to slow down the edges of that square trigger signal.
The only way this can not work, is if your trigger subsystem is a slave to some coarse time quanta at low sampling rates.
Quote from: Itsu on 2020.09.24, 14:23:25
After 0.0099 and 0.0100 (1s) it should continue with 0.0101, then 0.0102 etc. not with repeated 0.0100 then skip to 0.0110 etc. i think.
When I look at the raw .csv file, I see all timestamps expressed in scientific notation, e.g.: "1.99998e-01".
How did you get to the purely decimal point number format ?
QuoteThat should have varied the starting phase of the sampling sequence.
Maybe the square wave is too steep. Try adding a small RC integrator to slow down the edges of that square trigger signal.
The only way this can not work, is if your trigger subsystem is a slave to some coarse time quanta at low sampling rates.
Ok, i can try that
QuoteWhen I look at the raw .csv file, I see all timestamps expressed in scientific notation, e.g.: "1.99998e-01".
How did you get to the purely decimal point number format ?
I tried to explain with: (data changed into "numbers") above, but as i have a Dutch Excel version, i am not sure how it is named:
Go to cel properties (right mouse on A colom), then 2th from above (getal = number), then change to 4 decimals, see picture.
Itsu
Quote from: Itsu on 2020.09.25, 07:34:25
I tried to explain with: (data changed into "numbers") above, but as i have a Dutch Excel version, i am not sure how it is named:
Go to cel properties (right mouse on A colom), then 2th from above (getal = number), then change to 4 decimals, see picture.
In my Excel it looks like this:
...but it only affects how the number is displayed. It does not affect the underlying value.
So the whole A-colum with scientific notation does not change to decimal when doing so?
I made a short video on what i did: https://www.youtube.com/watch?v=K6ZCBf43LDw
Itsu
Quote from: Itsu on 2020.09.25, 14:51:15
So the whole A-colum with scientific notation does not change to decimal when doing so?
No, after importing external data, Excel stores the values in binary form internally.
Your video shows only the change of the display format - this change does not alter the binary values stored internally, it only changes their presentation on the display. See the attachment.
The significant digit truncation happens in the procedure which imports the data from the external csv file into Excel.
Just compare the number of significant digits in the
CSV file (https://www.overunityresearch.com/index.php?topic=3924.msg84454#msg84454) displayed by
Notepad++ (https://notepad-plus-plus.org/downloads/) with the number of significant digits imported by Excel (you can keep the scientific notation display format, to see the difference directly).
After that import is completed, all subsequent operations do not lose any information and the display format does not affect the values which Excel has stored internally.
QuoteNo, after importing external data, Excel stores the values in binary form internally.
Your video shows only the change of the display format - this change does not alter the binary values stored internally, it only changes their presentation on the display. See the attachment.
OK, yes the data stays the same, only the presentation changes from scientific to in this case decimal.
So the original data from the Tek scope already has the problem after 1s.
QuoteThe significant digit truncation happens in the procedure which imports the data from the external csv file into Excel.
Just compare the number of significant digits in the CSV file displayed by Notepad++ with the number of significant digits imported by Excel (you can keep the scientific notation display format, to see the difference directly).
Below screenshot is from the original file from the Tek scope opened with Notepad++
Around the 1s mark (red horizontal line) it goes from 10ms steps to 100ms steps (and repeating that step 4 times).
Excel has not touched this file yet.
Quote from: Itsu on 2020.09.26, 07:48:15
Below screenshot is from the original file from the Tek scope opened with Notepad++
Around the 1s mark (red horizontal line) it goes from 10ms steps to 100ms steps (and repeating that step 4 times).
Excel has not touched this file yet.
You are right, the time stamp resolution is too low for this sample rate and so they repeat in the csv file.
I think a bug report is in order.
OK, but this won't be a showstopper, Right?
Itsu
No, it is only an annoyance.
As far as showstoppers go, I nominate the pancake/toroid crosstalk.
OK, so what do you suggest, throw in the towel or go for a second pancake coil build to do some measurments on?
Quote from: Itsu on 2020.09.28, 20:09:47
OK, so what do you suggest, throw in the towel or go for a second pancake coil build to do some measurments on?
It think this pancake coil will not get any better. Its SRF is high enough that it can be lowered with external capacitances.
So the next steps are:
1) build another pancake to create a bucking H-field when connected in parallel and to minimize the E-field occuring between the two pancake coils.
2) wind the toroidal coil with the same winding technique as the pancake coils in order to avoid circumferential induction in it (this is the major contributor to the H-field crosstalk).
3) surround the toroidal coil with a grounded E-field (only!) shield as described
here (https://www.overunityresearch.com/index.php?topic=3924.msg82938#msg82938) in order to minimize the E-field crosstalk.
I will do the same.
Great, that will keep me occupied for the rest of the winter.
Quote1) build a mirror image pancake
So physically different (mirrored) from the first one?
Quote from: Itsu on 2020.09.29, 07:41:18
So physically different (mirrored) from the first one?
No, because the coil can be flipped around to make a bucking field. These pancake coils essentially do not have a direction and a hot/cold side because reversing their connections or flipping their sides predominantly changes the direction of the H-field generated by them.
Practically, some of the turns of these coils cross "above" and some other turns cross "below" the former. The idea of the mirror or side flip is for the "above turns" to
mesh with the "below turns" of the 2
nd pancake, so that the axial distance between the two pancake coils remains as constant as possible.
Right, so only the crossing scheme will be different from the first coil.
Started to build a second former using 1mm fishing line etc. to create the 2th pancake coil.
I moved the part about the defective Manson SPS4900 PS to a new thread here: https://www.overunityresearch.com/index.php?topic=3960.msg84636#msg84636
Itsu
I finished my 2th, spaced, weave 1 pancake coil (left) which is mirrored compared to the right coil.
16.14uH @ 100KHz compared to 16.17uH for the first one.
Itsu
Quote from: Itsu on 2020.10.15, 20:29:30
I finished my 2nd spaced, weave 1 pancake coil (left) which is mirrored compared to the right coil.
16.14uH @ 100KHz compared to 16.17uH for the first one.
So you don't need to flip it and you've got the first point from the list below, behind you.
Quote from: Itsu on 2020.09.28, 20:09:47
So the next steps are:
1) build another pancake to create a bucking H-field when connected in parallel and to minimize the E-field occuring between the two pancake coils.
2) wind the toroidal coil with the same winding technique as the pancake coils in order to avoid circumferential induction in it (this is the major contributor to the H-field crosstalk).
3) surround the toroidal coil with a grounded E-field (only!) shield as described here (https://www.overunityresearch.com/index.php?topic=3924.msg82938#msg82938) in order to minimize the E-field crosstalk.
Quote2) wind the toroidal coil with the same winding technique as the pancake coils in order to avoid circumferential induction in it (this is the major contributor to the H-field crosstalk).
Mammoth task requiring lots of patience...........
Quote from: Itsu on 2020.10.17, 20:35:24
Mammoth task requiring lots of patience...........
I know, I've done it but I put some gaps between the turns (like with the pancakes) so it goes faster.
Finally finished the weaved coil around the watertube.
Litz wire
Fres = 34MHz
L= 11uH @ 100KHz
R= 0.4 Ohm
Now looking into screening it.
Itsu
Quote from: Itsu on 2020.10.29, 15:31:25
Finally finished the weaved coil around the watertube.
Litz wire
Fres = 34MHz
L= 11uH @ 100KHz
R= 0.4 Ohm
That's a high Fres, but that's good.
What S21 do you get for the circumferential induction with a loop of the same diameter, like on the picture below?
How does that compare to the previous coil construction, when the distance is kept the same.?
I made some S11 measurements of this weaved tube coil, see screenshot below.
Seems my Fres from yesterday (34MHz with FG /scope) was on a harmonic.
Problem again with the S21 measurements is the calibration.
I tried some setups, but non showed any sensible data (to me).
Itsu
I made an S21 calibration using a SOL and for thru a short between port 1 and port 2, then inserted the single loop to port 1 and the tube coil to port 2.
Below are some S21 measurements with a distance of 2cm between coils.
Itsu
How does that S21 compare to the toroidal coil of the old construction ?
Well, i have to check the old data as i rewound it using the spaced weaved litz wire.
Itsu
It seems i never made such measurements on the old toroidal coil.
I can make similar toroidal coil on a spare tube and redo that measurement....
Itsu
I made a new toroidal coil on a spare tube similar as the old toroidal coil and used the same setup as the new weaved spaced tube coil, see screenshot of the same data.
Itsu
If you look at the two superimposed |S21| graphs (at the same scale) then the difference in the circumferential current induction is significant between the old toroidal construction and the new. This induction is the major contributor to the undesirable crosstalk. The reminder appears to be of capacitive nature since it increases with frequency.
Great, so your weaved construction method payed off to minimize the crosstalk.
And screening it, will further decrease it.
Itsu
I was trying all sorts of easy screening methods to avoid using verpies his suggested method, but they all failed, so i finally decided to give verpies his idea a shot, see here:
https://www.overunityresearch.com/index.php?topic=3924.msg82938#msg82938
The result looks promising, so now have to apply severall layers of varnish and see if it will hold its shape.
Its 3 layers of magnet wire on a flexible hose.
Itsu
Quote from: Itsu on 2020.11.09, 21:37:53
The result looks promising, so now have to apply severall layers of varnish and see if it will hold its shape.
What diameter is this wire?
How long did it take you to wind the 3 layers of this wire?
The diameter of the hose seems a little too small to create an overlap of the 2 hemi-toroids created after the circumferential cuts...but I might be wrong just by looking at the photo.
Varnish the wires first on the surface and cut them later after the varnish dries.
Do not let the varnish soak through the wires so much that it glues them to the hose
*, unless you were proactive and wrapped the hose with a saran wrap (https://www.amazon.com/saran-wrap/s?k=saran+wrap) first (or coated the hose with wax).
Do not use a saw with teeth to cut the wires because teeth will fray them. Use a small and thin abrasive cut-off disk (e.g. for the Dremel tool) or a very very sharp knife.
After separating the Litz Cloth (formed by the varnished wires) from the hose, varnish it again from inside and don't forget about the cut edge, (the one that will be overlapped and not be soldered), so it is varnished in such manner that the exposed ends of the cut wires are not shorted electrically. We do not want to create any conductor loops that will oppose the AC magnetic field.
Note, that some varnishes are more rigid than the others.
*
I used a flexible styrofoam rod which completely dissolves in acetone instead of the rubber hose and an epoxide based varnish which is immune to acetone, so I did not have any stiction problems.
Wire is 0.2mm, it toke 20 mins or so to wind once i had the winding rig set up.
I now too think the diameter of the hose is to small, but it was all i had and went with it.
The wire is directly onto the hose, so i will have to see how it comes out.
I expect the inner layer will not be adequately varnished and could cause problems during cutting or lateron, so will see how that goes.
I will use a dremmel cut-off disk.
Many thin layers of varnish will be better than a few thick ones, so will take some time.
Itsu
After 7 layers of varnish i used the dremmel cutting tool to slice it in half.
Getting it loose from the tube was somewhat cumbersome, but doable, some loosened up wires stayed behind, so first thing i did was varnish the inside to fixate it, see picture.
Soldering the outer rings will be the next challence.
Itsu
Quote from: Itsu on 2020.11.16, 14:35:09
Soldering the outer rings will be the next challence.
If you want, you can also solder the inner rings instead of the outer rings. The ground connection placement doesn't make much difference for an E-field shield so it is up to you.
Whichever ones you chose, make sure the wires on the opposite rings are not shorted together. You can fray them a little and varnish their ends to isolate them. We do not want Eddy loops anywhere!
Soldering the edges is a mess, the heat melts the varnish before it melts the green solderable coating of the magnet wire causing the half rings to fall apart in multiple pieces.
It smokes and smells like hell and the burned varnish prevents solder from attaching to the exposed copper, see picture.
I can't see how to attach the both halfs together using more solder as the halfs will crumble apart while applying heat.
Not sure this is doable.
Itsu
Quote from: Itsu on 2020.11.18, 15:56:51
Soldering the edges is a mess, the heat melts the varnish before it melts the green solderable coating of the magnet wire causing the half rings to fall apart in multiple pieces.
That is a problem indeed.
The epoxide varnish, which I use, does not melt and holds the wires together despite the temperature. It is the same stuff which is used in PCB soldermasks.
I give up on making a screen for the tube coil.
I tried severall approaches, i sacrificed a kacher/Tesla secondary coil.
The varnish/coating of the magnet/litz wire makes it somehow impossible to solder it again as the solder won't flow and/or stick to the cleaned blank copper.
I used heavy solder irons and normal light ones, but non works.
So untill i find another solution this has to go on the backburner for now.
Itsu
Quote from: Itsu on 2020.11.18, 15:56:51
Soldering the edges is a mess, the heat melts the varnish before it melts the green solderable coating of the magnet wire causing the half rings to fall apart in multiple pieces.
It smokes and smells like hell and the burned varnish prevents solder from attaching to the exposed copper, see picture.
I can't see how to attach the both halfs together using more solder as the halfs will crumble apart while applying heat.
Not sure this is doable.
Itsu
Well with ltz wire don't try the back burner it vaporizes the fine copper strands ;) but wife's nail varnish remover works on most things,
Yeah, i tried acetone, but it did nothing to remove the magnet wire coating.
Try flow solder and flux, heat with a hot air gun when the two parts are together
As used with soldering SMD components.
Regards
Mike 8)
I tried flux, and used a 120W soldering iron, but its not flowing / sticking, just smoking, smelling and burning.
Unfortunatly i have no heat gun to try.
Itsu
Some enameled wire has high temp coating that would be almost impossible to tin in this config.
Some winding wire like from jaycar actually act as solder flux and it would go fairly smooth.
Another thing is the so called "lead free" solder which can be less than useless for this.
Every time i buy solder advertised as 60/40 ...half the time it goes in the bin with along with an ebay refund.
Jaycar do have the right 60/40 solder if that helps.
Thanks 3D,
i am using "lead free" solder since a few years and really had to adapt to it.
So it might have something to do with that too, so i will be looking for some old fashion 60/40 solder lead solder.
Itsu
Hi Folks,
On flux you mean pine resin?
Here is what I mean: https://www.amazon.co.uk/Natural-ROSIN-COLOPHONY-Incense-Solder/dp/B01N11EUFS (https://www.amazon.co.uk/Natural-ROSIN-COLOPHONY-Incense-Solder/dp/B01N11EUFS)
Thanks,
Gyula
Almit Flux T435:
https://dokumen.tips/documents/almit-t435.html
Itsu
Quote from: gyula on 2020.12.04, 12:57:27
Hi Folks,
On flux you mean pine resin?
Yes I have that in small tubs, works well normally and is non acid unlike the others. It is solid until you touch with the tip of the soldering iron.
But what I use for smd is a soler paste in a tube.
Regards
Mike 8)
Here is what I mean: https://www.amazon.co.uk/Natural-ROSIN-COLOPHONY-Incense-Solder/dp/B01N11EUFS (https://www.amazon.co.uk/Natural-ROSIN-COLOPHONY-Incense-Solder/dp/B01N11EUFS)
Thanks,
Gyula
Hi Itsu and Mike,
Okay, thanks for the answers.
Itsu, I have often used a contact spray, TUNER 600, when I quickly wanted a clean surface for soldering. It dries in seconds, leaving no residue (and the surface this spray is applied on cools down).
Here is a data sheet on it: http://www.dact.com/Tuner_600_data_sheet.pdf (http://www.dact.com/Tuner_600_data_sheet.pdf)
Please note that this spray does not remove enamel from the wire, it should be removed first by other means, unfortunately.
A possible source for TUNER 600: https://www.soselectronic.com/products/kontakt-chemie/tuner-600-200ml-33196 (https://www.soselectronic.com/products/kontakt-chemie/tuner-600-200ml-33196) At other places it tends to be more expensive.
Gyula
Thanks Gyula,
i have "Kontakt 60", which is a similar contact spray i guess, but as you mentioned it will not clean off any enamel or varnish.
Itsu
I was able to make some shielding around the weaved tube coil using some litz wire, see pictures
Doing same measurement as in post #659 above but using the scale as in verpies his post #664.
Screenshot 1 shows the outcome (worse seems to me) without the shield grounded.
Screenshot 2 shows the outcome with the shield grounded.
There are a lot of differences, but not any better in my opinion as the unshielded outcome in post #659
Inbetween those measurements, there was a VNA upgrade to be able to vary the transmittion level (+1dBm, -1dBm, -4dBm and -10dBm).
No idea what level i was on, now i have set it to -1dBm so probably the 2 situations are not comparable :(
Itsu
As my shielding is asymmetrical (short side curls in, long side extends to the middle) due to avoiding the ends touching, i tried swapping over the weaved coil, so now the long extended shielding side is exposed to the transmitting loop, and the picture changes.
Below screenshot 1 is with flipped over coil in the same scale as my post above.
We now see a flatlining over the first 17Mhz or so on the S21 graph.
Screenshot 2 is a zoomed in shot of S21 graph.
Grounding the shield makes things worse
Itsu
I folded over the long side of the shielding inside over the short side making sure they did not contact each other and taped it all around the tube.
Now measurements show an almost equal damping from both sides (S21 Gain of -62dB).
I have sandwiched the tube coil between the 2 spiral weaved coils for further maesurments
Itsu
I finally was able to put all items of this setup together, see picture.
I have the 2 weaved pancake coils put in parallel bucking (4uH @ 100KHz) 2cm apart.
The tube coil also weaved measures 11uH @ 100KHz.
The tube is filled with double distilled water and the tube coil screen is grounded.
Running at 4.250MHz for now (sine wave DC) from my FG into the Amplifier which pulls 12.7V @ 1.2A (16W).
Input circuit is balun across a 10nF cap then a series trimmer cap 100pF with across it a fixed 220pF ceramic (total 320pF) to the both paralleled pancake coils.
Output circuit is a 100pF trimmer cap parallel to the tube coil to a parallel 670KOhm load resistor, see diagram.
Output measurement shown in screenshot below, yellow = voltage across the 670K resistor (11.8V rms), green = current probe in this 670K load resistor (1.7mA rms), red is math function voltage x current (2mW) .
Strange is that the voltage and current are not in phase. Also, P=U²/R would result in 0.2mA........
I will do some tuning and adjustments to hunt for anomalies
Video here: https://youtu.be/5IeLHGW7Qco
Regards Itsu
Itsu,
That looks like a 90 degree phase between voltage and current. The ratio of rms voltage to rms current gives a capacitive reactance of 6.9Kohms. At your frequency that works out to a 5pF capacitance, which would be the scope probe.
Smudge
Hi Smudge,
Ok, sounds plausible, the phase difference is about 84°, but why does the P=U²/R gives a 10 times lower value at 0.2mA?
Itsu
Hi Itsu,
Maybe the 670K resistor isn't 670K.
Smudge
Smudge, you are right, its 690K :P
Hi Itsu,
Would you check the resistance or impedance of the 690 k resistor with your LCR meter at 100 kHz?
On your question: why does the P=U²/R gives a 10 times lower value at 0.2mA? Smudge meant the resistor cannot be 670 kOhm because the measured current and voltage (if we are to believe the measurements and we surely can) shows it otherwise. The 84° phase difference (current leads voltage) certainly indicates the resistor, together with the voltage probe is a capacitive load, as he also mentioned.
The 4.25 MHz frequency is very high for such type and many kOhm value resistors from the self inductance and capacitance point of view.
If you have 3 pieces of 220 kOhm similar size resistors, try to connect them in series and replace the 690 Kohm single resistor and see how the phase angle, current and voltage changes if any.
OR, just omit the 690 K resistor and see the voltage and current without it. Use say two 3 cm long pieces of wire to replace the legs of the resistor, voltage probe would clip across these open wire ends and the current probe would clamp around one wire like it was for the resistor. (the load in this case would be the probe input impedance (parallel RC) as specified for that voltage probe)
Gyula
Gyula,
thanks for your insights.
The used 690K resistor measures 450mH @ 100KHz, which i find hard to believe.
Concerning the "P=U²/R gives a 10 times lower value at 0.2mA?" (must be 0.2mW!!) i gather that the scope math shows the correct calculated value (2mW), and the "P=U²/R" goes wrong due to the 84° phase shift.
But if i incorperate this 84° phase shift by multiplying the 0.2mW by Cos84 i get an even worse outcome (0.02mW).
A 100 Ohm 1% induction free resistor as load gives more normal (no phase shift) results, so it must be the used 690K with its reactance that plays up.
I will do your tests later tonight.
Regards Itsu
Hi Itsu,
Thanks for doing the 100 Ohm load test check.
I cannot recall the type of your LCR meter, does it have series or parallel equivalent impedance selection choice? And in resistance mode does it use normal DC current only?
Normally, as the frequency increases from the kHz to the MHz range, a film or metal film resistor gradually becomes increasingly reactive (mostly capacitive but even a turning point for inductive nature may happen), especially for the higher values like a few kOhm and higher. Check a few kOhm, then say some ten kOhm and then some hundred kOhm metal film types at 100 kHz with your meter, just for fun to see this behaviour for yourself.
I think this formula P=U²/R is valid for resistances without any reactance and you can use a Z impedance instead of the R resistance if the impedance is real.
Greetings
Gyula
Quote from: Itsu on 2021.01.26, 11:51:29
Gyula,
thanks for your insights.
The used 690K resistor measures 450mH @ 100KHz, which i find hard to believe.
Concerning the "P=U²/R gives a 10 times lower value at 0.2mA?" (must be 0.2mW!!) i gather that the scope math shows the correct calculated value (2mW), and the "P=U²/R" goes wrong due to the 84° phase shift.
But if i incorperate this 84° phase shift by multiplying the 0.2mW by Cos84 i get an even worse outcome (0.02mW).
A 100 Ohm 1% induction free resistor as load gives more normal (no phase shift) results, so it must be the used 690K with its reactance that plays up.
I will do your tests later tonight.
Regards Itsu
Itsu,
Actually, with the scope measured rms voltage and current being 11.83v and 1.714ma respectively, to arrive at the mean output of 2.406mw would require a phase angle of arccos(2.406e-3) = 89.862 degrees. This phase angle could be assumed from the traces. Pout = 11.83*1.714e-3*cos(89.862) = 2.408mw.
Also, the current lead of ~90 degrees simply means the network connected between the probes is capacitive at the frequency used no matter what the indicated resistor value is.
Pm
Quote from: gyula on 2021.01.26, 14:03:08
Hi Itsu,
Thanks for doing the 100 Ohm load test check.
I cannot recall the type of your LCR meter, does it have series or parallel equivalent impedance selection choice? And in resistance mode does it use normal DC current only?
Normally, as the frequency increases from the kHz to the MHz range, a film or metal film resistor gradually becomes increasingly reactive (mostly capacitive but even a turning point for inductive nature may happen), especially for the higher values like a few kOhm and higher. Check a few kOhm, then say some ten kOhm and then some hundred kOhm metal film types at 100 kHz with your meter, just for fun to see this behaviour for yourself.
I think this formula P=U²/R is valid for resistances without any reactance and you can use a Z impedance instead of the R resistance if the impedance is real.
Greetings
Gyula
Gyula,
My LCR meter is an Agilent U1733C, and it does have the series/parallel selection choice.
Not sure about the resistance mode being DC current only.
Knowing the impedance of this 690K resistor @ 4.250MHz would be nice, but neither my LCR meter nor the little VNA is able to measure it somehow.
Itsu
Quote from: partzman on 2021.01.26, 14:25:49
Itsu,
Actually, with the scope measured rms voltage and current being 11.83v and 1.714ma respectively, to arrive at the mean output of 2.406mw would require a phase angle of arccos(2.406e-3) = 89.862 degrees. This phase angle could be assumed from the traces. Pout = 11.83*1.714e-3*cos(89.862) = 2.408mw.
Also, the current lead of ~90 degrees simply means the network connected between the probes is capacitive at the frequency used no matter what the indicated resistor value is.
Pm
PM,
thanks, so the scope math shows the correct power level (2.4mW), but the problem is the P=U²/R calculation, which is only 0.2mW, and even when using the 89 or 84° phase shift does not correct this (gets worse).
Only when taking a value of 69K then it starts making sense (without the COS phi).
Anyway, i will not use the P=U²/R*COS(phase shift) for output power calculations.
Itsu
Gyula,
QuoteOR, just omit the 690 K resistor and see the voltage and current without it.
Use say two 3 cm long pieces of wire to replace the legs of the resistor, voltage probe would clip across these open wire ends and the current probe would clamp around one wire like it was for the resistor.
(the load in this case would be the probe input impedance (parallel RC) as specified for that voltage probe)
Thats what i did, and below the resulting traces, see screenshot 1
Yellow probe is specified as 3.9pF@10MOhm.
2th screenshot is with a 100 Ohm 1% induction free resistor as load.
Itsu
Hi Itsu,
Okay, thanks. I have found and will read the user manual for the LCR meter later on, just for curiosity.
Normally the higher than some ten kOhm RF impedances mean the upper limit of measurement ranges for such meters, only the high-end instruments may be able to measure say up to 100-120 kOhm like the good old HP4193A, this works up to 110 MHz. (I did not study the latest vector impedance meters though.)
Your little VNA surely has its own upper range limits too.
The 1st screenshot above for the yellow probe shows amplitude and phase relations comparable to that of the 690 kOhm resistor.
Here is an impedance calculator for parallel RC network https://keisan.casio.com/exec/system/1258032649 (https://keisan.casio.com/exec/system/1258032649) and it gives Z = 9.55 kOhm absolute value for your probe (10MOhm/3.9pF) used as a load at 4.27 MHz frequency (phase angle is -89.945 degree).
Gyula
Hi Gyula,
interesting this calculator, i would have thought that the impedance would be on top of the resistance, but it decreases from 10MOhm to 9.5KOhm.
So when paralleling the 10MOhm probe and the 690K resistor, it results to a 645.5K resistor and the 690K resistor adds about 4pF to the capacitance (according to my LCR meter), totaling to about 8pF.
This results according to your calculator into (645.5K / 8pF @ 4.25MHz) a Z=4.7K @ -89.5°
Itsu
Yes. And add to this that we do not know the equivalent circuit values of the 690 k resistor itself which also influence the 690 k value at the 4.25 MHz frequency. I mean the Z=4.7 k you just wrote should be "something else". This is because you have got the measured data (the phase angle, current and voltage) involving the 690 k, and Smudge calculated the "something else" value as Z= 6.9 kOhm but he used an 'eyeballed' 90 degree phase shift from your scopeshot.
Here is the behaviour of thick film chip resistors in the function of frequency (size 1608): https://www.quora.com/What-are-the-characteristics-of-a-5k-resistor (https://www.quora.com/What-are-the-characteristics-of-a-5k-resistor)
The frequency dependence of metal film types like the 690 k you happen to use is much worse than the flat chip resistors.
Gyula
Gyula,
well, if i now use my found 4.7K impedance on the P=U²/R * COS Phi, then i get:
11.8²/4.7 * -89.5 = 2.5mW which is very close to the scopes calculated 2.4mW in post #688.
Problem solved or is this coincidence?
By the way, my used 690K resistor is a carbon ¼W type.
Itsu
Okay Itsu, the problem is solved I think. The P=U²/R * cos(Phi) formula is an Ohm's Law variaton of the P=U*I*cos(Phi) formula so it can be used too.
Edit: Carbon resistor: okay, then it may indeed have the 600+ kOhm resistance and the scope probe capacitance shunts it at the 4.25 MHz frequency.
Gyula
Using the diagram below, i am trying to put as many power into the paralleled coils, but allthought the amplifier when running on a 50 Ohm dummy load puts in about 6 Watts @ 4.25MHz, when connecting to the paralleled coils, i measure no more then about 200mW going in.
Both the input and output trimmers are set approx. half way causing a peak in output power (a sort of resonance at 4.25MHz), but there still must be a form of mismatch.
Using a balun between amp. and step up cap (10nF) makes little difference.
Looking for the correct method of inputting more power into the paralleled coils.
Itsu
Hi Itsu,
Would like to ask the DC resistance of the pancake coils measured between points A and B, see attached schema.
In a video you mention about 4 uH for the pancake coils, do you mean it across points A and B ? Were the magnets and the ground wire on the top present when you measured the 4 uH?
Please also tell the inductance of the output coil, you may have included it already but I missed it. You need to disconnect the capacitors that are now present, for inductance measurements of course.
The capacitive divider could match the pancake coils but it is hard to find the correct ratio. Perhaps first you could use a negative resistance oscillator between points A and B and connect a tuning capacitor also across points A and B which would resonate and oscillate the two pancakes around the 4. 25 MHz frequency. And when this cap value is known then the members of capacitive divider could be figured out perhaps more easily.
For a negative oscillator I refer to an earlier circuit of last year where an n- and a p-channel JFET or an n-channel JFET and a pnp transistor is combined into the two terminal device (Lambda diode), you surely recall it. When such device oscillates your full circuit between points A and B, you can measure and tune it to around 4.25 MHz.
Added: a few uH variable coil could be used between the hot output of the amplifier and the center point of the capacitive divider, this would increase the impedance which is now between the center point and point B, I mention this because I believe the 10 nF loads down the amplifier output excessively. Or you could reduce the 10 nF to say 2.2 nF or 3.3 nF etc.
Gyula
Hi Gyula,
remember, the coils on the drawing are shown in series, but they presently are in parallel (bucking).
Below is measured with the magnets, the ground wire on the top and tube/output coil present, no caps:
The DC resistance of the pancake coils measured between points A and B is 0.4 Ohm.
The inductance of the pancake coils measured between points A and B is 4.64uH
The DC resistance of the tube/output coil is 0.4 Ohm
The inductance of the tube/output coil is 10.8uH
I do remember the negative resistance oscillator, i will look it up and see i can build one.
EDIT: found the link (private thread) here: https://www.overunityresearch.com/index.php?topic=3910.msg81273#msg81273
in there there is this link to a negative resistance webpage: http://www.zen22142.zen.co.uk/Theory/neg_resistance/negres.htm
I also will try out lowering the 10nF and/or adding a "few uH variable coil".
As i understand it from verpies/Smudge, the goal now is not to get the pancake coils radiate, but only to have the H-fields penetrate the water inside the tube to get the protons resonate which then is being picked up by the tube/output coil.
Thanks, itsu
Hi Itsu,
Thanks for the details. I modified the schematic to show the paralleled pancake coils, is that correctly drawn now? So the 4.64 uH refers to the paralelled pancakes, right?
For capacitive divider, try to reduce the 10 nF to as low as say 820 pF as a start and find resonance by peaking the trimmer cap. Roughly a 300 pF tuning cap would be needed across the paralelled coils and this 300 pF should come from the two seriesly connected capacitors in the divider.
I return later on.
Gyula
Gyula,
QuoteI modified the schematic to show the paralleled pancake coils, is that correctly drawn now? So the 4.64 uH refers to the paralelled pancakes, right?
Thats correctly drawn, and yes, the 4.64 uH refers to the paralelled pancakes. O0
Itsu
Hi Itsu,
Here are some capacitor value pairs in pF for the divider as a guide (but this maybe late for you):
820:480 1000:440 1500:380 2200:350 3300:340, higher values within a pair mean the bottom cap values like the 10 nF shown in the drawing.
Any pair of these give a 300 - 305 pF resultant tuning cap for the paralleled coils to resonate around 4.25 MHz.
You found 6W power output from the amplifier directly in a 50 Ohm dummy load: this then means 17.3 V RMS (48.8 V peak to peak) across the 50 Ohm resistor.
So if you monitor the RF amplitude across the paralleled coils by a HV scope probe, the amplitude should be much higher than this when the trimmer is tuned to resonate the coils), depending on the ratio of the capacitors in the divider. And how much of this can couple to the output coil is another question of course.
So the goal would be to achieve as high voltage amplitude across the paralleled coils as possible (even 15x 20x 30x higher than what comes out from the amplifier when driving the cap divider input) at resonance. It will not be close to the 5-6W but perhaps would approach that.
Gyula
Hi Gyula,
thanks for the info, i have put together a divider with a 120->370pF variable cap and a fixed 2.2nF smd, so it will be close to your 2200:350 one.
Will see how that works out.
Itsu
Some input measurements with the new input matching circuit, see diagram.
Tuned to resonance on 4.25MHz which means almost all capacitance (trimmers) used.
Screenshot shows the input signals (voltage yellow, current green and power red) coming from the PA.
Video with the new setup and measurements here (also with voltage probe on pancake coils): https://youtu.be/V8L6-ujsZ7k
Itsu
This is a progress for sure and perhaps testing the 3300pF : 340 pF cap ratio would show the tendency to go either higher or lower ratios. Just put some nF capacitor, one at a time, across the 2.2 nF chip cap and reduce one of trimmer caps little by little while watching either the calculated power on the scope where you have now the 4W or watch the increase on the power meter which showed the 1.5W. I think you will need to increase the nF values (and slightly reduce the trimmer caps value) to increase the power towards the 6W which is possible with the dummy load.
Addition:
I think you can trust much better in your scope measurements than in your power (and swr meter) regarding the power levels (4W vs 1.5W).
Have you happened to check the power+swr meter versus the scope power measurement when the 50 Ohm dummy load was used? I mean whether the power meter when showed the 6W, what the scope RMS voltage was across the 50 Ohm?
Regarding the swr meter, it indicates roughly 1.4 in this latest video, what did it show when the dummy load was used? IT should show near 1.0 even if there is no full scale in the forward power direction.
The ratio of the stepped up voltage to the input RF voltage seems to be 2.4 (24V RMS/10 V RMS) This ratio could surely be increased, meaning this is the direction to go towards better matching.
Here are some more divider cap values to choose from: 4700pF : 330pF 5600pF : 320pF 6800pF : 320pF 8200pF : 315pF 10nF : 313pF 15nF : 310 These latter values may seem extreem but perhaps if you see an increasing voltage step up ratio it should go together with input power increase too, measured at the capacitor center tap input. So the 6W power may be better approached: question is whether this 4W power that you have now is enough for the effect or not yet.
Thanks,
Gyula
Hi Gyula,
i will add some caps to the 2.2nF one to see how it behaves.
I too think the scope will be a better indicator of the power input then the SWR meter due to the phase shift i see between voltage and current.
I will do some measurements on the SWR meter, problem is to clamp the current probe.
It did show a SWR of 1:1 when using the dummy and the 6W.
The present 70Vpp across the pancake coils with the 4W input is still way of the severall hundred volts Smudge mentioned in his PDF:
"The drive coils will have RF voltage across them that is several hundred volts."
Regards Itsu
Hi Itsu,
When you terminate the swr meter with the 50 Ohm dummy load, it would be enough to use the voltage probe because the scope will surely measure correctly the RMS voltage across it and then use the formula P=U²/R. This then gives a power level, if it is indeed 6W, then the RMS voltage should be around 17.3 V (48.7 Vpp) across the 50 Ohm. I understand the current probe is a problem to clamp and if the voltage level is close to the calculated one, then no need for the current probe in this case.
Gyula
Quote from: Itsu on 2021.01.29, 09:23:18
The present 70Vpp across the pancake coils with the 4W input is still way of the severall hundred volts Smudge mentioned in his PDF:
"The drive coils will have RF voltage across them that is several hundred volts."
When I wrote that I envisioned the two pancake coils to be in series, not in parallel. So your 70Vpp looks good to me.
Smudge
I managed to clamp the current probe around the dummy load center lead.
When Amp. set to 15V (1A) it generates 6W on the SWR meter at a 1:1 SWR (dummy measures 51 Ohm).
Screenshot shows in yellow the voltage across the 51 Ohm (19.49V rms), in green the current through it (354mA rms) and in red the power consumed by it (6.89W).
P=U²/R with this data gives 7.4W.
Video here: https://youtu.be/J9T3DY0usKQ
Regards Itsu
Quote from: Smudge on 2021.01.29, 14:55:03
When I wrote that I envisioned the two pancake coils to be in series, not in parallel. So your 70Vpp looks good to me.
Smudge
Hi Smudge, ok, yes of course, so we are set to go already.
I will try to maximize the input matching circuit as outlined by Gyula, then go hunting for anomalies.
Itsu
Hi Itsu,
Okay on the wattmeter calibration check, so it shows a lower watt value than reality around this 4 MHz frequency. Such power + swr meters are good enough but one cannot expect much precision from them and the better power meters are more expensive. Thanks for doing and showing these tests.
Gyula
I made some measurements while varying the input caps ratio.
input matching circuit
trimmer caps (pF) fixed cap (nF) Pancake voltage (Vpp) V rms Vpp SWR meter (W)
120/370 1.1 89 15 48 3.5
120/370 2.2 106 9.5 27 1.4
120/370 3.3 70 5.5 16 0.4
The trimmer caps settings for max. pancake coil voltage are not similar as for max. input voltage!!
The trimmer caps are adjusted anywhere inbetween their ranges, no specific measurements done.
So it looks like that the initial used 2.2nF cap was the optimum one if looking at the pancake coil voltage (106V), see screenshot 1.
The screenshot 2 is from the same setup (2.2nF) but with signals on the SMA input connector, yellow voltage, green current and red power (seems somewhat high).
Itsu
Hi Itsu,
It turns out now that your picking the 2.2 nF capacitor value (instead of the 10 nF) was a lucky one. It is likely that if the 2.2 nF were variable, say +/- 300 pF or so around the 2.2 nF then such fine tuning would give a few more peak to peak volts but this is not really needed. Smudge wrote that the 70 Vpp already looked good and now you have 106 Vpp.
Regarding the values of the trimmer capacitors, they are not important to measure because at the 2.2 nF capacitor at the bottom the trimmers must have about 350 pF together to arrive at the 302 pF resonating value for the paralleled coils.
Regarding your observation that the max pancake coil voltage does not coincide with the maximum input voltage (I express your sentence differently but I think this is what you meant?), it may be caused by the following:
I think that under the 2.2 nF fixed capacitor values the stepped down resonant impedances are getting higher than 50 Ohm hence manifesting lighter loads (hence higher voltages) and above the 2.2 nF fixed cap values the stepped down impedances are getting below 50 Ohm hence manifesting heavier loads (hence lower voltages) for the power amp output.
Putting this otherwise: the power amp is the generator and loading it below or above its own output impedance its output voltage changes with respect to the power matched case (when it has half of the voltage compared to its unloaded output voltage).
Thanks for all your kind efforts.
Gyula
Gyula,
QuoteIt turns out now that your picking the 2.2 nF capacitor value (instead of the 10 nF) was a lucky one.
the picking of the 2.2nF cap was not just luck, i happen to have some of these 2.2nF SMD multi layer ceramic caps which are i think excellent for these kind of applications.
QuoteRegarding your observation that the max pancake coil voltage does not coincide with the maximum input voltage (I express your sentence differently but I think this is what you meant?)
Not sure if that is expressed correctly, as i wanted to say that in one setup situation, like the 2.2nF setup, the tuning of the caps for max. input on the sma connector is different as for tuning the max input on the pancake coils.
I don't think it changes the load on the PA when moving the voltage probe from one point to the other.
Anyway, i guess i will do some slow sweeping across a frequency range say 4 to 4.5Mhz,watch the output 690K resistor output for anomalies while slowly tuning step for step the input caps.
Thanks, Itsu
Hi Itsu,
On the 'lucky pick' of the 2.2 nF I meant you did not choose say 1.5 nF or 2.7 nF even if you had had several capacitor values.
The latter values would have given less amplitudes across the pancakes versus the 2.2 nF as you found out in your tests with the 1.1 or 3.3 nF caps.
You wrote earlier:
"The trimmer caps settings for max. pancake coil voltage are not similar as for max. input voltage!!"
You explained it too:
"... I wanted to say that in one setup situation, like the 2.2nF setup, the tuning of the caps for max. input on the sma connector is different as for tuning the max input on the pancake coils."
Sorry if I did not express it correctly. I think the explanation for your findings is that the trimmer capacitor(s) not only vary the input impedance of the divider (which loads the PA output) but tunes the resonant frequency of the overall pancake tank at the same time. This then involves the differing voltage amplitudes manifesting across the input of the divider (i.e. across the 2.2nF) versus the maximum voltage possible at resonance across the tank.
You also wrote:
"I don't think it changes the load on the PA when moving the voltage probe from one point to the other."
I agree: the placement of the voltage probe cannot cause any change, its own 4-5 pF is negligible versus either the 2.2 nF or the resulting 302 pF tank capacitor value. The load on the PA changes whenever the trimmer capacitor(s) are varied.
Is this okay now? 8)
Thanks,
Gyula
Gyula,
QuoteI think the explanation for your findings is that the trimmer capacitor(s) not only vary the input impedance of the divider (which loads the PA output) but tunes the resonant frequency of the overall pancake tank at the same time.
This then involves the differing voltage amplitudes manifesting across the input of the divider (i.e. across the 2.2nF) versus the maximum voltage possible at resonance across the tank.
Right, so aming for max. voltage across the pancakes (tuning for resonance) will be the goal, despite it will cause a slight impedance mismatch on the PA.
Okay O0
Itsu
Yes, that is ok, always adjust the trimmer (or trimmers) for max voltage measured across the pancake coils.
When the FG gives 4.25 MHz and the pancake tank is also tuned to max amplitude which is say 106 Vpp, the 3 dB bandwidth of the tank could be estimated by detuning the FG to a lower and an upper frequency at which the thank voltage reduces to 0.707 x 106 = 74.9 Vpp for both frequencies. The difference of these two frequencies will give the loaded bandwidth of the tank circuit, a useful detail. If the lower frequency is say at 3.9 MHz and the upper frequency is say at 4.6 MHz, then their difference is 0.7 MHz so the 3 dB bandwidth would be B = 0.7 MHz. This then would give the loaded Q of the pancake tank in this example as QL = f0/B = 4.25/0.7 = 6
Gyula
Gyula,
i tried to measure the 3 dB bandwidth, but it seems there is a second peak near by in frequency, so the results are not really clear.
So i first made a 2Mhz to 6Mhz sweep of the pancake coils response while tuned for max. voltage on 4.25Mhz using the trimmer caps, see screenshot.
It shows a broad resonance area with some dips and peaks from which i mentioned the 2 strongest.
In the below video i did another long (2min) sweep of the output coil (690K resistor) from 4 to 4.5Mhz while again tuned for 4.25Mhz and i noticed a jitter around the 4.25Mhz.
I think this jitter might be from the these dips and peaks around resonance.
Not sure where this ragged resonance area comes from, probably the 2 pancake coils are not exactly tuned for resonance and kind of interfere with their resonance points.
Video here: https://youtu.be/K9SoBdijA7o
Itsu
When i tune the trimmers for max signal while using the 2 to 6Mhz sweeping, i get a strong (170Vpp) resonance peak lower in frequency around 3.5Mhz (trimmers all 3 on max. capacitance), see screenshot.
Also here we see these bites out of the trace, pointing to me to the both coils interfering with their individual resonance points.
Why we have such a much stronger resonance peak lower in frequency i can't explain as i would have thought that using the trimmers we could get this strong resonance peak anywhere in the range of the caps.
Perhaps the output coil at the same frequency inhibits the resonance to build up (Q).
Itsu
Hi Itsu,
Thanks for doing all these tests. My thoughts: Chances are that the pancake coils have their own resonant frequencies by themselves and they are coupled and both are included in forming a parallel LC circuit too.
There may be a certain influence on them from the output coil too but it may not be significant.
If the self resonance of the pancake coils happens to be in the 3 to 4 MHz range accidentally, this may give such behaviour you found by the frequency sweep. Whether this oddity is a bad or neutral thing from the searched effect point of view I cannot judge it.
Gyula
Hi Itsu,
The anomaly you are looking for is a strong resonance peak, so that 3.5MHz peak may be important. Is it still there when the water is removed? If it can be identified with the presence of the water it could be what you are looking for. After all your hard work you deserve a breakthrough. My previous estimate of the expected proton resonance frequency was based on an estimate of the static field from the magnets, and that may have been wrong. It would have been nice if you had a miniature proton spin magnetometer that enabled you to measure that static field....Oh but of course that is what you are building, silly me! The early work on NMR searching for the predicted nuclear resonance back in 1948 nearly failed because the frequency was just outside the range being searched. Here's hoping for that breakthrough.
Smudge
Gyula, Smudge,
the 3.5Mhz strong peak is still there without the water.
So i will expand the sweeping range somewhat (3 to 5Mhz f.i.) to see if another strong peak appears with the water added only.
Itsu
The last few days i was busy with characterizing the below shown setup by sweeping frequency between 2 and 6Mhz and adjusting the input and output trimmer caps.
Each found peak and/or dip in amplitude was checked both with and without the distilled water present.
The screenshot below shows one such double peak just below 4Mhz (center) with water (white) and without water (yellow).
Up till now no difference between with and without water in peak/dips was noticed.
I tried both with the 690K Ohm resistor (179Vpp) as with a 100 Ohm 1% resistor (29Vpp).
Regards Itsu
Hi Itsu,
Not sure at what sweep rate you did this. The whole object of this unusual water geometry is to minimise the frequency dispersion of the individual proton precessions, and that would result in a very small bandwith of the NMR resonance. It could be so narrow that it will only be found with a very slow sweep across the frequency range. I can't offer a feel for how slow this should be, but I see it as a laborious exercise like looking for a needle in a haystack. I think you are looking for a line width that is much smaller than the typical resonances that you are seeing, so you must be on the lookout for a sudden jump then fall over a very small change in applied frequency.
If this CW approach fails to find the proton resonance it may be necessary to try the pulsed approach looking for a free induction decay or a spin echo. If successful this will tell you the proton frequency and then you can narrow down the CW approach. The pulsed approach applies pulsed modulated RF, a pulse of certain width or certain number of cycles. There should then be a free induction exponential decay of RF following that pulse. The spin echo approach requires a second RF pulse some time after the first one, again of a certain width, and that causes the RF to build up then decay again as a spin echo. I would need to look into this further to find the pulse widths required. Does your FG have the ability to pulse modulate the output RF?
I am most grateful for your time spent on this, thank you.
Smudge
Hi Smudge,
i did a 200ms sweep when tuning the input / output trimmer caps to see how they influence each other, then switched over to 10s sweeps to do the step by step movement of the trimmer caps.
These first rough sweeps did not show much, so i will now move to some more narrower (3 to 5Mhz) sweeps which could be longer in time like 1 minute or more.
I have a fairly advanced FG, so i can modulate the output RF, yes.
Thanks, regards Itsu
Quote from: Itsu on 2020.12.02, 21:09:15
So untill i find another solution this has to go on the backburner for now.
Oh!, I can see it is off the back burner now. I didn't know about it since I was not checking the forum for a long time.
Kudos to Gyula for all that practical RF advice and to Itsu for not giving up.
I'd like to notice several elephants in the room:
1) The NMR is so narrow that you need to resolve down to 100Hz or even lower. The MHz wide sweeps are too wide to notice it.
2) If the RF circuits are tuned precisely to the NMR frequency, how will you distinguish between them ringing and the NMR ?
3) The goal is to maximize the amplitude of the current flowing in the pancake coil's windings, because it is the current that generates magnetic fields (not voltage!). Since the current amplitude increases together with the voltage amplitude, tuning for maximum voltage at the coils accomplishes the same result, although maximizing the voltage is not the ultimate goal in itself.
4) The SWR between the PA and "tuner*" should be minimized and the SWR between the "tuner*" and the pancake coils should be maximized. This also means that the current/voltage phase offset between the "tuner*" and the pancake coils should be maximized.
5) The RF power between the PA and the "tuner*" should be very similar to the RF power between the "tuner*" and the pancake coils **, despite the SWRs being wildly different.
6) The grounding point of the Litz shield should make sense in the RF circuit's topology. It should ground away the capacitively coupled energy between the pancake coils and the toroidal sensing coil. This means RF ground, not necessarily Earth ground.
7) The Litz shield grounding wire should be of very low inductance so it offers very low impedance path to the unwanted capacitively coupled energy of the electric field. This energy should not be coupled into the toroidal coil's circuit !
* By "tuner" I mean the capacitive matching circuit, which is between the PA and the pancake coils.
** Any difference between these two powers will be due to dielectric losses in the "tuner's*" capacitors, resistive loses in the wiring and in the "tuner*" and due to the RF radiation resistance (minimal EM radiation by the pancake coils will be indicated by the maximized SWR between the "tuner*" and the pancake coils and maximized i/v phase at the coils). We really do not want the pancake coils to work as good antennas that radiate EM radio waves away
Hi verpies,
thanks for the comments, i will take a deeper look to them later today.
The grounding now is on my house earth system on the attic, so no good RF ground.
I do see a hugh difference in signals on the output coil with or without the ground attached, like 200Vpp with and 450Vpp without.
Also there are other peaks on different frequencies with or without.
But this problem could be due to the grounding of the scope probe which is on the same ground as the house grounding.
Itsu
Verpies:
Can you explain what is the difference between an Earth ground, compared to an RF ground.
NickZ
PS. I have given up on OverUnity.com.
Seams as only the trolls are left there, without any control or moderation over what has been going on there.
So, I hope they don't come looking to disrupt this forum, as well.
Just a warning...
verpies,
Quote1) The NMR is so narrow that you need to resolve down to 100Hz or even lower. The MHz wide sweeps are too wide to notice it.
i already went down to 500KHz sweeps because also Smudge mentioned something like that.
Now limiting to 100KHz sweeps.
Quote2) If the RF circuits are tuned precisely to the NMR frequency, how will you distinguish between them ringing and the NMR ?
good question, i understand/guess a narrow(er) spike/peak ontop of the resonance peak would appear.
Quote3) The goal is to maximize the amplitude of the current flowing in the pancake coil's windings, because it is the current that generates magnetic fields (not voltage!). Since the current amplitude increases together with the voltage amplitude, tuning for maximum voltage at the coils accomplishes the same result, although maximizing the voltage is not the ultimate goal in itself.
Ok, so monitoring the pancake coils current together with the output coil voltage would be needed and possible without groundloop problems.
Quote4) The SWR between the PA and "tuner*" should be minimized and the SWR between the "tuner*" and the pancake coils should be maximized. This also means that the current/voltage phase offset between the "tuner*" and the pancake coils should be maximized.
Hmmm, the first part is easy as the SWR meter is inbetween the PA and "tuner".
The second part not, as its hard to measure.
Quote5) The RF power between the PA and the "tuner*" should be very similar to the RF power between the "tuner*" and the pancake coils **, despite the SWRs being wildly different.
Ok, thats somewhat easier to measure using the SWR/Power meter and the scope math.
Quote6) The grounding point of the Litz shield should make sense in the RF circuit's topology. It should ground away the capacitively coupled energy between the pancake coils and the toroidal sensing coil. This means RF ground, not necessarily Earth ground.
Ok, so i now have used a 10cm long, thick litz wire from grounding shield to PA/SWR meter outer coax shield, which dropped the output coil signal from 200Vpp to about 80Vpp depending on tuning.
Quote7) The Litz shield grounding wire should be of very low inductance so it offers very low impedance path to the unwanted capacitively coupled energy of the electric field. This energy should not be coupled into the toroidal coil's circuit !
Right, but also this is hard to measure.
Meanwhile sweeping along..........
Itsu
Quote from: NickZ on 2021.02.06, 14:30:33
Verpies:
Can you explain what is the difference between an Earth ground, compared to an RF ground.
It is the point to which the Litz shield attaches. The purpose of this atypical shield is to block the capacitive coupling of the electric field between the pancake coils and the toroidal coil (while leaving the magnetic field unaffected) so the RF ground should be capacitively symmetrical with respect to the toroidal coil terminals.
Also, one should be careful to measure only the toroidal coil's differential signal and not the common mode signal, so the grounding of the scope matters, too.
The common mode interference signal, which can be picked up by an Earth-grounded scope, can be attenuated with multiple ferrite beads (or toroidal RF cores) which encompass BOTH wires connecting the toroidal coil. These beads significantly increase the inductance of these wires, so they present a high inductive reactance path to the common mode signals. ...and because these beads encompass BOTH wires - they do NOT attenuate the differential signals.
Come to think of it, the common mode ferrite beads (or toroidal RF cores) would also attenuate the undesirable common-mode signals between the "tuner" and the pancake coils.
P.S.
I understand what you mean about the O.U. forum and welcome. Things are different here.
Quote from: Itsu on 2021.02.06, 16:22:21
Hmmm, the first part is easy as the SWR meter is inbetween the PA and "tuner".
The second part not, as its hard to measure.
You can use the i/v phase offset to gauge the SWR between the "tuner" and the pancake coils, because the larger this phase offset becomes (ideally 90º), the higher the reactive current becomes between the "tuner" and pancake coils and the SWR also becomes larger there.
Large Reactive Current (ideally: PowerFactor=0) is desirable at this point in the circuit. Large Real Current would be bad because it would mean that EM waves are being radiated or resistive heating or dielectric heating is happening.
@Gyula
How is your experience with common-mode chokes at these frequencies ?
Also, what do you think about an addition of a shorted turn, which would employ the Lenz law against common-mode RF signals, too... ?
Quote from: verpies on 2021.02.06, 17:09:16
You can use the i/v phase offset to gauge the SWR between the "tuner" and the pancake coils, because the larger this phase offset becomes (ideally 90º), the higher the reactive current becomes between the "tuner" and pancake coils and the SWR also becomes larger there.
Large Reactive Current (ideally: PowerFactor=0) is desirable at this point in the circuit. Large Real Current would be bad because it would mean that EM waves are being radiated or resistive heating or dielectric heating is happening.
Sweeping from 3.5 to 4.5Mhz.
At 3.9Mhz:
the SWR meter shows 6W @ SWR 1:1.4 (max. / min.)
The math shows ELI phase (103°) see screenshot 1 (measuring at pancake coils yellow voltage, green current)
The sweep shows a resonance peak around 3.9Mhz see screenshot 2 (measuring at output coil)
Itsu
The strange thing about the above first screesnhot is that when calculating the power from the shown pancakes voltage and current we get a negative 17W.
This probably is due to the fact we are measuring "inside" the parallel input circuit.
Itsu
Quote from: Itsu on 2021.02.06, 18:28:40
The math shows ELI phase (103°) see screenshot 1 (measuring at pancake coils yellow voltage, green current)
Looks like your phase measurement is reversed 77º → 103º
Hmmm, i measured phase: yellow -> green = 103°.
When measuring phase: green -> yellow i get -103°.
So still 103°, but i know what you mean it should be 77° as more then 90° would not be possible for lagging or leading phases in capacitive or inductive circuits.
Itsu
Quote from: Itsu on 2021.02.07, 11:00:21
Hmmm, i measured phase: yellow -> green = 103°.
When measuring phase: green -> yellow i get -103°.
So still 103°, but i know what you mean it should be 77° as more then 90° would not be possible for lagging or leading phases in capacitive or inductive circuits.
Itsu
Itsu,
Check your current probe direction in the circuit. If the current was 180 degrees reversed, then the phase would be ~77 degrees and the input power would be positive. However, input current to input voltage phase >90 degrees is possible with the right circuit configuration.
Pm
Hi PM,
i did check the probe as i found the negative power strange, but to me it is in the right direction, see diagram below, where in green the current probe position / direction.
Indeed if i reverse the probe direction, the current will be leading (ICE) by 77° and the calculated power will be positive.
So am i thinking wrongly that the current probe in the drawn position is the correct one?
Added a picture of the setup including current probe and the thick litz wire for grounding the shielding litz to the outer coax from the amplifier.
Itsu
Quote from: Itsu on 2021.02.07, 15:44:59
Hi PM,
i did check the probe as i found the negative power strange, but to me it is in the right direction, see diagram below, where in green the current probe position / direction.
Indeed if i reverse the probe direction, the current will be leading (ICE) by 77° and the calculated power will be positive.
So am i thinking wrongly that the current probe in the drawn position is the correct one?
Added a picture of the setup including current probe and the thick litz wire for grounding the shielding litz to the outer coax from the amplifier.
Itsu
I agree with you. The current probe direction is correct in the schematic so what you see is what you have! :) I don't understand it though!
Pm
Hi Verpies,
Thanks for your kind words. Regarding your questions, I have good experience with common mode chokes, I used ferrite toroids on the outside of coax cable to reduce current in the outside braid.
At these frequencies involved here the material of ferrite beads or toroidal cores should still have to have low loss and ample cross section area for the differential RF currents, especially when higher power levels of some ten or hundred watts are involved. The use of several individual cores is preferred to a single one core, i.e. stacked one after the other. this reduces core losses when high differential currents are to be attenuated.
I agree with the suggestion in your reply#739 on using two common mode chokes at the input and at the output of the pancake coils as you indicated in the photo. It will involve a small retuning of the trimmers. I cannot exactly recall Itsu's toroidal core arsenal, maybe an OD of 1.5 or 2 cm from 3F4 or similar ferrite material if he has two of them would give some kOhm common mode reactance with max 10-15 turns which are easily windable.
The permeability of such cores may range from some hundred to a few thousands, the 3F4 has about u=900.
About the idea of the addition of a shorted turn to reduce common mode RF signals: interesting idea, I never tried it. My initial guess would be that one or only a few shorted turns would have small attenuating effect (this also depends on the frequency), but many, shorted single turns, probably isolated electrically from each other would start blocking differential currents. The Triax coax type has a continuously braided and conducting shield around the two inner conductors and I wonder how this would compare to your idea of shorted individual turns.
Gyula
Quote from: verpies on 2021.02.06, 17:48:46
@Gyula
How is your experience with common-mode chokes at these frequencies ?
Also, what do you think about an addition of a shorted turn, which would employ the Lenz law against common-mode RF signals, too... ?
Quote from: partzman on 2021.02.07, 16:48:41
I agree with you. The current probe direction is correct in the schematic so what you see is what you have! :) I don't understand it though!
Pm
I verified with manual sweeping that the above used setup where we see negative power, is real and repeatable.
Below (and above) resonance, we have around 80-90° phase shift between voltage and current (ELI) and positive power going into the pancake coils.
Around resonance we see a quick shift to around 107° phase shift and a negative going power.
Above resonance things change back to "normal" again.
Itsu
So the pancake coils are feeding energy into the "tuner" :o ?
Does the presence or absence of water have any effect on this negative power?
Smudge
Perhaps more importantly, at this resonant point can power be extracted via a lower value load resistor on the toroidal coil and does the negative power on the pancake coils then reduce? If so is it possible to get zero power at the pancake coils while extracting power from the toroidal coil? You may then have achieved the dream.
Smudge
Not sure what is going on, yesterday (and like in post #743) i had this repeatable negative power behaviour: https://youtu.be/kI8QkjPicqo
This morning with the same setup its different (normal?): https://youtu.be/jD7Gh5ObRV0
I think i can have either a defective current probe or the magnetic field of the strong magnets is off-setting it.
The 359 means the amplifier is not recognizing the attached probe.
Degausing the probe away from the magnets is OK, but upon approaching the magnets the 359 returns off and on.
Not sure if this defect / off-set can affect the phase being measured only.
Perhaps i should try using a csr.
Itsu
Hi Itsu,
Perhaps first you could arrange repositioning the current probe as I show in this picture with the piece of (green) wire, so the probe could get further away from the ring magnet. The length of the new wire would not disturb tuning noticeably.
Gyula
Note:
The negative power can also come from the fact that the cosine of angles above 90° is negative mathematically and the oscilloscope follows the math rules. cos(103°)= -0.2249
So supposing the current probe is not influenced by the ring magnet as much as say causing higher than 90° phase angle appear on the scope, just by the unwanted magnetic bias of the probe, the negative power may also come from the negative cosine value.
Partzman wrote yesterday: "However, input current to input voltage phase >90 degrees is possible with the right circuit configuration." I also mean the 103° phase angle could also be reality, this latter will turn out with the further tests.
Gyula
Quote from: gyula on 2021.02.08, 11:27:11
Perhaps first you could arrange repositioning the current probe as I show in this picture with the piece of (green) wire, so the probe could get further away from the ring magnet.
...or get a hacksaw and cut a piece of a steel water pipe large enough to contain the entire current probe inside it. The steel should effectively screen out any stray external magnetic fields (especially the HF ones).
This might come handy in future projects, too.
With the pandemic, lockdowns and all, we all can use the exercise ;)
(https://i.imgur.com/ksrdbTm.gif)
Quote from: gyula on 2021.02.08, 11:54:12
The negative power can also come from the fact that the cosine of angles above 90° is negative mathematically and the oscilloscope follows the math rules. cos(103°)= -0.2249
Yes, that was my initial take on that, too, because cos(77°) = -cos(103°).
However, Itsu is insisting that sometimes the current really is lagging behind the voltage by more than 90°.
Perhaps his current probe is getting saturated by the external magnetic field or his scope makes a phase measurement error when the signal level is too low or a common-mode signal is confusing the scope.
Repositioning, shielding the contactless current probe should eliminate such modes of error. Substituting a CSR for the current probe should do it, too. Finally, common-mode signals can be attenuated by a CMC as you've described.
Also, in theory, at the LC resonance frequency, the amplitude should be the highest and the the current should lag the voltage by 90°. However, Itsu's system does not manifest the highest amplitude at 90°.
Thanks guys,
I have arranged the current probe like Gyula suggested.
I think also there is the matter with how the scope is calculating the phase, like both Gyula and verpies mentions, >90° then power gets negative.
Only, it should not get >90°.
But using a 1 Ohm 1% induction free csr does not make more sense to me.
Using that csr, the power into the pancake coils is 9W (stays positive), but is hard to believe as the difference with the current probe (0.5W) is hugh.
Something is not right here.
Video here: https://youtu.be/prx00V2WI5I
Itsu
Hi Itsu,
IF you agree, perhaps check the current probe right at the output of the function generator if you still have the dummy load with the short wire loop as you used at the output of the power amplifier with the current probe.
Perhaps you could insert the same 1 Ohm csr in series with the 51 Ohm dummy load and also check the phase angle that way too. Adjust the FG to its highest output possible, it would be terminated still correctly by the dummy load (and the 1 Ohm in series with it).
You could do this same test at the output of the PA but the output of the PA may bring in a certain reactance while the FG output cannot. You did this dummy load test last week at the PA output and the current probe gave correct current phase vs the voltage across the dummy load.
Gyula
Quote from: Itsu on 2021.02.08, 16:12:20
Thanks guys,
I have arranged the current probe like Gyula suggested.
I think also there is the matter with how the scope is calculating the phase, like both Gyula and verpies mentions, >90° then power gets negative.
Only, it should not get >90°.
But using a 1 Ohm 1% induction free csr does not make more sense to me.
Using that csr, the power into the pancake coils is 9W (stays positive), but is hard to believe as the difference with the current probe (0.5W) is hugh.
Something is not right here.
Video here: https://youtu.be/prx00V2WI5I
Itsu
Itsu,
What is your calibration for CH2(blu) that is used to measure the CSR. Your scope shows 500ma/div with a peak current of ~800ma. Is this correct?
Pm
PM,
not sure what you mean by "calibration for CH2(blu) that is used to measure the CSR", but i suspect the mV to mA translation which is 1:1 (1A/V).
The 500ma/div is correct, not sure where you see the peak current of ~800ma.
Gyula,
I will be using a 50 Ohm 1% induction free resistor with a series 1 Ohm 1% induction free csr to do some measurements, and it shows some results i do not expect to see, like the amplitude of the signals and the phase offset on 4MHz of this induction free csr.
Using the above 51 Ohms and 3.46V rms i would expect to see 67.8mA rms, but i do not.
Itsu
Okay Itsu, take your time and next time please tell some measured data.
The 50 or 51 Ohm termination across the FG output should halve the open circuit output voltage of course and the current and voltage of the 51 Ohm should be in phase. You will surely find the problem.
Last week you found almost fully in phase current and voltage for the dummy load at the output of the PA.
Thanks for all the efforts.
Gyula
See above, here the video: https://youtu.be/ggAyTt3Y05U
we see 51 Ohm and 3.46V rms across it so i expect to see 67mA rms of current.
Instead, the current probe (green) shows 63mA rms (close) and the csr current (Blue) shows 84mA rms (way off imo).
Also @ 4Mhz we have on the csr a considerable phase offset.
Going to 1Khz improves concerning the csr phase difference as its now almost in phase.
But the amplitude and/or rms value is still to high in my opinion.
The current probe shows a much to low value and very off phase, so i suspect this probe to be defective.
I just found out that this probe does not measure any DC currents, so the DC part of it might be the problem. >:(
Itsu
Yes, the current probe got defective. And the 1 Ohm is not 1 Ohm at 4 Mhz but higher than that inductively. This latter explains the "9W" going into the pancake coils, it is obviously wrong.
Regarding the current probe, try to degauss it a few times repeatedly, perhaps it helps. But if it cannot measure DC current as you wrote, then it got defective unfortunately. Perhaps there is a helpline or forum discussion for such failure?
Quote from: gyula on 2021.02.08, 22:15:48
Yes, the current probe got defective. And the 1 Ohm is not 1 Ohm at 4 MHz but higher than that inductively.
I concur.
Also, I think that the inductance of the interconnects and the capacitance of the probes should not be neglected.
With all the 0.1 Ohm csr tests i have done lately using my nanoVNA in my "workbench" thread i also had analyzed my 1 Ohm csr and i remembered that the inductance in the 10KHz to 10MHz range did not increase that much as the 0.1 bare metal csr's.
Therefor i expected that my 1 Ohm csr would be stable up to the used 4MHz.
To be sure i now checked again the used 1 Ohm csr again using the "Shunt-Thru" measurement method, and it confirmed only a slight increase in impedance on 4Mhz, see green marker in the below screenshots.
It shows that the impedance increases from 927 mOhm on 10KHz to 1.031 Ohm on 4MHz to 1.223 Ohm on 10MHz.
This means a 9% increase on 4MHz compared to 10KHz, so the seen amplitude increase and phase shift at 4MHz compared to 1KHz on my 1 Ohm csr might be plausible when including the "inductance of the interconnects and the capacitance of the probes".
Looking for a new current probe as i think the hall sensor in my present one is toast.
Itsu
Meanwhile sweeping in 100KHz chunks across a 3.5 to 4.5MHz range.....
Itsu
Quote from: Itsu on 2021.02.09, 15:46:08
Meanwhile sweeping in 100KHz chunks across a 3.5 to 4.5MHz range.....
Did you add the CMCs to decrease the common-mode cross-talk and make NMR signal relatively larger ?
No, not yet, will look into the CMC usage today, see if i can make some and install them.
Itsu
I have lots of different ferrite toroids, but most are of unknown origin.
I still have these small black 10x6x4 toroids for my nano pulser toroids in the dally setup, and some white TN10/6/4-3E25 (U = 5500)
Guess i will use them to start with and put some 10-15 turns on them.
Itsu
Hi all,
Back in 2011 when we did Lithuania experiment, the frequency of ferrite was measured by coils of 90 degrees arrangement.
We had flat coil made from foil around yoke and the standard coil on top of it.
Then after giving white noise to coil from foil I looked on peaks in spectrum analizer for possible points where material of ferrite did resonate - https://www.youtube.com/watch?v=kGGCG2i8u8U
Hopefully this approach would be useful to you as well when looking for material resonant frequencies.
Cheers!
I used the white TN10/6/4-3E25 with 10 turns each (both measured about 270uH @ 100KHz) as a first CMC on the output coil, see picture.
Screenshot 1 shows the difference in output signal without (purple) and with (white) cmc for a 3.5 to 4.2MHz sweep. ====>>>> wrong!!, should be:
Screenshot 1 shows the difference in output signal with (purple) and without (white) cmc for a 3.5 to 4.2MHz sweep.
Screenshot 2 is at a different vertical/division setting of the purple trace.
Itsu
Hi Itsu,
That is 10 x voltage gain at the output, very nice and it was a good idea from verpies to bring up the use of CMCs.
I wonder whether you had to readjust the trimmer capacitor to peak it around the 3.8 MHz or you did not need to do that?
Thanks for your efforts.
Gyula
QuoteHi Itsu,
That is 10 x voltage gain at the output, very nice and it was a good idea from verpies to bring up the use of CMCs.
I wonder whether you had to readjust the trimmer capacitor to peak it around the 3.8 MHz or you did not need to do that?
Thanks for your efforts.
Gyula
Gyula,
no, no adjustment was needed.
I will make another one with the black ferrite to see how that behaves.
Itsu
Ooops, it seems i was on a false peak after adding the cmc.
I did had to retune the output cap, even so that it now has a stronger output of 92Vpp then without the cmc :(
See screenshot sweeping from 3.5 to 4.5MHz now
Itsu
Oh, okay.
You may wish to make the CMC even more balanced by moving the turns of one of the windings on the core just a little while watching the inductance on an L meter and achieve the minimum inductance possible ( say some 10 nH ), see the attached photo I edited. IT does not matter which side of the CMC you short circuit the wires and measure on the other side.
This way the CMC will have the smalles detuning effect on the output coil.
Of course, this is may need to secure the windings so that the turns should not move after the fine adjustments and you put the CMC back into the circuit.
But I think the CMC is already good enough. O0
Gyula
Quote from: T-1000 on 2021.02.11, 00:48:09
Then after giving white noise to coil from foil I looked on peaks in spectrum analizer for possible points where material of ferrite did resonate - https://www.youtube.com/watch?v=kGGCG2i8u8U
Hopefully this approach would be useful to you as well when looking for material resonant frequencies.
Yes, this one of the methods of stimulating the material. All of the being:
1) CW
2) FM
3) Nanopulses
4) Noise
Quote from: gyula on 2021.02.11, 22:44:17
You may wish to make the CMC even more balanced by moving the turns of one of the windings on the core just a little while watching the inductance on an L meter and achieve the minimum inductance possible ( say some 10 nH ), see the attached photo I edited.
Yes, this is a good test to maximize the effectiveness of the CMC and to make sure that the windings are not wound in the wrong direction (it is the easiest mistake to make, while constructing a CMC).
The white cmc (measured 268 and 272uH) gives a resulting 2.15uH when added see pic. 1.
The black cmc (measured 232 and 228uH) gives a resulting 2.02uH when added see pic. 2.
Squeezing together and/or spreading the turns makes very little difference (10nH or so).
So i guess they are wound in the correct way.
The question might be if they are suitable for the used frequency.
The screesnhots show the nanoVNA sweep (10KHz - 10MHz) in this shorted condition and confirm the resulting 2.xxuF inductances.
Itsu
Hi Itsu,
Thanks for the CMC measurements. Your output coil has 10.8 uH inductance as you wrote and the 2 or 2.1 uH resultant inductance of the CMC adds to that, this is why the output trimmer needs to be retuned. This small inductance increase for the output coil does not seem to cause any problem, what is more, by retuning the circuit it gave higher output voltage as you found.
May I notice that the inductance difference between the two coils within the same core sounds rather high, i.e. 268 uH vs 272 uH. This difference can only be reduced if a quarter or a half turn is removed from the 272 uH coil or adding similarly that amount to the 268 uH coil, to have them very close to each other, within say 0.5 uH. Then the 2 - 2.1 uH measured differences will be much smaller.
But please deal with such nuances only if you feel like to, the turns should be fixed mechanically in the end, after the windings trimming, otherwise any small bending of the wire ends may change the full windings inductance on that side.
The 3E25 core material for the white core is not favorable for the 4 MHz range, above some hundred kHz it gradually becomes lossy and the initial permeability of 6000 reduces to a very low value.
I do not know the material for your black core.
Question is how these specs may influence the goal with this setup, if negligibly, then no need to further bother with the CMCs and they are already good enough.
Gyula
I tried to match the black cmc coils to be equal, by adding about 1/8th of a turn to the lowest valued one, but even when within 1uH, the short test keep showing 2uH which might indicate the limit for my LCR meter.
They now both measure 231uH, and i have superglued the ends, so when enough dried i will see what the output does with this black cmc.
My expectations would be to have a lower output signal with the cmc in place then without but the last test with the white cmc did not meet that expectation.
Is my expectation wrong?
Itsu
I think the first thing now would be to find an answer on why the 10x gain in voltage amplitude has come about what you reported in your reply #773?
Maybe the output trimmer tuning got somehow off-tuned and then after the CNC insertion you returned and got 10x as many voltage.
Gyula,
i see what you mean now with "10x gain in voltage.....etc." and what is causing my confusion.
I screwed up again in that reply #773, as the 2th sentence should read:
Screenshot 1 shows the difference in output signal with (purple) and without (white) cmc for a 3.5 to 4.2MHz sweep.
^^^ ^^^^^
Sorry about that.
So when adding the cmc, the signal was 10x smaller due to the fact that the output resonance point had shifted way lower in freqeuncy and we were left with the 10 lower input resonance peak only.
Lateron i retuned the output circuit so the 2 resonance points are combined again resulting in a small increase in total amplitude over the without cmc situation.
Itsu
Okay Itsu, I understand, thanks for explaining.
Would be interesting to see how the black CMC would behave at the output now that you trimmed both windings on it to 231 uH. (By the way, is the type of the black core known?)
And regardless of that, would you connect the white cored CMC to the input of the pancake coils? (try to trim the windings on it also closer to each other if possible).
This way all the 3 coils assembly would become a balanced, quasi fully ground independent circuit. (Even though the CMC core(s) may not function correctly around 4 MHz.)
Thanks for your kind efforts.
Gyula
Quote from: gyula on 2021.02.12, 17:47:52
May I notice that the inductance difference between the two coils within the same core sounds rather high, i.e. 268 uH vs 272 uH.
Me thinks that hundreds of µH for a shorted CMC is too high.
What is the inductance of one winding (q.g. the green one) of the CMC while the other winding is open (e.g. the red one) ?
Hi Verpies,
Yes, that would be too high. But he wrote this:
The white cmc (measured 268 and 272uH) gives a resulting 2.15uH when added see pic. 1.
The black cmc (measured 232 and 228uH) gives a resulting 2.02uH when added see pic. 2.
And he showed the inductance for the shorted coils as around 2 uH, see the pictures in Reply 780 above.
EDIT: Sorry I misunderstood you. Itsu measured the individual windings while the other winding in the same core was open. The results are what I quoted from him in italics.
Gyula
Gyula, verpies,
The black toroid make is unknown.
After some trimming, the white cmc is now very equal too (265 / 265 uH)
And yes, white cmc has 265 / 265 uH measured when its opposite coil was open.
black cmc has 231 / 231 uH measured when its opposite coil was open.
I installed both cmc's now, the white at the input, the black at the output, see picture.
I found both resonance points and combined them close to 4MHz, see screenshot.
We now have a peak of 30V or so, so much less then the 91V with only the white cmc on the output.
itsu
Dear Itsu,
In post 773, you edited the mistake and it says now that
Screenshot 1 shows the difference in output signal with (purple) and without (white) cmc for a 3.5 to 4.2MHz sweep.and the amplitude of the white waveform was 80.4 Vpp without the white CMC.
Now you wrote:
Quote from: Itsu on 2021.02.13, 17:08:31
...
We now have a peak of 30V or so, so much less then the 91V with only the white cmc on the output.
I think you may have meant "...so much less then the 91V
without the white cmc on the output." ?
Anyway, Would you mind removing the white CMC and use only the black one at the output?
This is to check whether the black core CMC performs similarly bad as the white one at the output.
One more test would be good to see: I think you have some Amidon RF toroids with at least 2 cm OD or higher which nicely work around 4 MHz and make a CMC for the output from such core. These cores has much less permeability though, this means that very likely at least 20 turns are needed to get say at least 30-40 uH for each winding, perhaps that would be good enough for a CMC.
Thanks for your kind efforts.
Gyula
Hi Gyula,
well, things get confusing now, i agree.
To reiterate:
post #773 without white cmc (white trace) 80.4V @ 3.5 to 4.2MHz
with white cmc (purple trace) 10.8V @ 3.5 to 4.2MHz => but output circuit OFF tune now!!
post #776 ooops, corrected the output circuit tuning, now:
with white cmc (purple trace) 91.6V @ 3.5 to 4.5MHz
^^^^^^
post #788 white and black cmc trimmed each to be equal value (white 265/265uH, black 231/231uH)
Both cmc's installed, black on output circuit, white on input circuit.
Both circuits tuned to max. at around 4Mhz.
We now have a peak of 30.5V, so much less then the 91.6V with only the white cmc on the output see post #776
Now, when doing a slow sweep (20s) with the both cmc's installed, something got smelly and it turns out to be the input white cmc which gets hot.
I reduced the sweep time to 10s and now its acceptable to 29°C at the resonance peak.
As can be seen in the above picture, the white input cmc is very close to the magnets which probably have a huge influence (saturation) on it, but not sure thats the reason for the heating up.
I will remove the white input cmc now, retune and see what the peak value is then.....
Looking for the Amidon Rf toroids too O0
Itsu
I removed the white input cmc (only black output cmc installed), and retuned the both circuits to max peak @ 4MHz:
with Black cmc (purple trace) 54.8V @ 3.5 to 4.5MHz
SO:
without cmc, tuned to max. at 4MHz: 80.4V
with white output cmc, tuned to max. at 4MHz: 91.6V
with black output cmc, tuned to max. at 4MHz: 54.8V
with black output cmc, AND white input cmc tuned to max. at 4MHz: 30.5V (white cmc gets hot).
Found a T106-2 Amidon toroid.
Itsu
Okay Itsu, now you cleared the issue, I disregarded your post # 776, unfortunately.
It is sure also that the white core has losses at RF frequencies like 4 MHz, it is made for 300 kHz or less as its data sheet includes. Its closeness to the magnets may also influence the loss but basically the core material is to blame. The few watts output at the PA driving the white core already heats it up.
Thanks for the test on the black core CMC, it is surely a low frequency core too, seems to be worse than the white core.
The T106-2 has an AL of 135, it would mean 400*135= 54 uH from 20 turns for each winding. Hopefully 25-30 turns may fit on for each half side. IT is a good RF core.
Gyula
OK, great, i will see how many turns i can get onto it.
Itsu
Quote from: Itsu on 2021.02.13, 20:48:45
OK, great, i will see how many turns i can get onto it.
Itsu
Sorry Itsu but I considered first this page for the A
L http://www.amidoncorp.com/t106-2/ (http://www.amidoncorp.com/t106-2/) where A
L is 135 but it needs to be divided by 10 if I use A
L =135 so 20 turns gives only
5.4 uH, unfortunately. 40 turns gives 21.6 uH using this calculator here: https://toroids.info/T106-2.php (https://toroids.info/T106-2.php) The suffix-2 in the T106-2 means permeability of 10 for that material.
So perhaps a higher permeability RF core would be better...
Ok, no problem, so a T200-2 won't be any better, what about a T225-8?
It says Al=425 only.
Itsu
Well the AL = 425 is wrong it is only 42.5 nH/N2. It means that 20 turns gives 400*42.5=17 uH, 30 turns gives 38 uH and 40 turns gives 68 uH.
My data comes from this data sheet, page 13: https://www.tme.eu/Document/82fc93ef93d05a020041f9203a13131a/MICROMETALS-Tseries.pdf (https://www.tme.eu/Document/82fc93ef93d05a020041f9203a13131a/MICROMETALS-Tseries.pdf)
OK, so looking for a Txxx-6 or so i guess, or do you have a suggestion?
Itsu
Hi Itsu,
First I thought of your core arsenal to use what is already available and I recalled some types. We need to realize, the RF powdered iron cores have low permeability hence relatively low AL factor, unfortunately.
Do you happen to have ferrite toroids from Amidon like the FT series? i.e. FT140-43 https://toroids.info/FT140-43.php (https://toroids.info/FT140-43.php) or FT240-43 https://toroids.info/FT240-43.php (https://toroids.info/FT240-43.php)
Perhaps Verpies could suggest other types too.
Gyula
Gyula,
I got some FT-3K50TS series toroids that i know of, but i might have more but most are of unknown origin.
Most are coloured, a few black like this hugh T107/65/18 3F4 Ferrite core.
I can order some FT140-43 / FT240-43 eventually.
Itsu
Hi Itsu,
It is okay that you can order the FT cores but I cannot tell how the lack of using such CMCs in this setup may influence the real goal of these tests.
You can use the FT-3K50TS and if they do not have too low permeabilty, then they could probably make better CMCs than the ferrite material manifested so far. Just try to wind 10 turns on them and measure inductance.
These cores are for CMCs.... see here: https://www.hitachi-metals.co.jp/e/product/finemet/cmc/core/ft-3k50t_f.html (https://www.hitachi-metals.co.jp/e/product/finemet/cmc/core/ft-3k50t_f.html)
Gyula
OK, i have:
FT-3K50T F3020CS 18uH@100kHz Nano and
FT-3K50T F6045GS 24uH@100kHz Nano
Will start with 10 turn coils.
Itsu
Perhaps the bigger one can accomodate 30-40 turns on any one of its half core area. 30 turns would give about 21.6 uH, 40 turns would give about 38.4 uH.
Hmmm, one of the smaller ones had some windings on it, 7 turns and 12 turns and they measure 1mH and 3.5mH @100KHz (1mm od wire).
Itsu
Well, it means that the correct data sheet for the core type is not found yet. Your measurements show that this core must have very high permeability.
From its OD, ID anf height data the permeability can be calculated if needed.
First you wrote FT-3K50TS for its type and I then gave link to FT-3K50T F series by mistake, and there was no data for FT-3K50TS. I think this is the Finemet type which was finished by Hitachi years ago and this was a topic already somewhere here.
I suggest to make two windings with identical number of turns on it and check how it behaves in the setup.
EDIT: from your L meter display in the upper right corner the Q quality factor is 1.27 at the 100 kHz test frequency. This low Q may or may not forecast a bad behaviour for the core at 4 MHz, we shall see.
Thanks, Gyula
Thanks Gyula,
the below attached data sheet shows a permeability of about 2000 at 4MHz, see Fig.2
Itsu
Okay Itsu, that is fine. Now it remains to be seen how it really behaves. You will have some difficulty to trim the turns to get a reasonably close inductance value between the two windings.
I put up 2 windings of 8 turns each on my finemet core, see picture.
They measure 979uH @ 100KHz each with the other coil open.
In series it measures 5.22uH @ 100KHz.
Without any cmc in the circuits and resonance tuned to 4MHz i have the white trace (94V).
With the finemet cmc in the output circuit and again tuned to resonance at 4MHz i have the purple trace (110V).
Sweeping from 3.5 to 4.5MHz.
Itsu
Itsu, that is great news. O0
Perhaps you could use the other Finemet core as a CMC at the output of the PA to drive the input of the pancake coils. (Or put this CMC to the front to see any behaviour. )
Thanks,
Gyula
Hi Gyula,
how is it great news as WITH cmc the output signal gets stronger?
I understood from verpies his Post #739, see quote below, that a cmc would attenuate (unwanted) signals, so i would expect to see less (only the wanted) signal with a cmc installed.
QuoteThe common mode interference signal, which can be picked up by an Earth-grounded scope, can be attenuated with multiple ferrite beads (or toroidal RF cores) which encompass BOTH wires connecting the toroidal coil. These beads significantly increase the inductance of these wires, so they present a high inductive reactance path to the common mode signals. ...and because these beads encompass BOTH wires - they do NOT attenuate the differential signals.
Where do i go wrong?
Itsu
Perhaps try to ask this question from yourself: why is it that the scope shows less output voltage without the CMC? C.C
Very likely an asymmetrical measurement setup (i.e. without a CMC or CMCs in this case) uses up power what the CMC can confine and route to the load.
Very likely, with no CMC, some of the output signal can leak to the enviroment via the ground of the instrumentation, this may also mean we do not always get full output in certain setups, this much depends on the particular measurement circumstances.
This is my take on this.
Gyula
Gyula,
ok, i will ponder on that question some more later :-\
For now i have added a 2th finemet cmc in the input circuit (just before the pancake coils).
It has similar 8 turns as the 1st one, but measures 1714 and 1715uH @ 100KHz :D
The below screenshot shows the output when tuned again to 4MHz resonance with both finemet cmc's installed (48.6V).
Sweep again from 3.5 to 4.5MHz in 20s (no heating up of the input cmc now).
Itsu
Okay Itsu, I can ask these:
Has the SWR changed when you connected the 2nd CMC to the front of the pancake coils?
I assume you had to retune the trimmer capacitors after inserting the 2nd CMC?
What is the output when you replace the two CMCs with each other?
Gyula
Quote from: Itsu on 2021.02.15, 21:39:14
...
ok, i will ponder on that question some more later :-\
...
I meant it would be also a valid question, sorry.
Addition to my previous post:
Would you test the 2nd CMC at the output too? (without using the 1st CMC)
Gyula
Hi Gyula,
I did not check on the SWR lately, but will do that later today.
I always retune both sides trimmer caps when changing anything just to check i am on resonance in both circuits, so i assume it was needed here to, but not sure.
I will swap over the CMC's and check the 2nd CMC at the output without using the 1st CMC also later today.
Regards Itsu
QuoteHas the SWR changed when you connected the 2nd CMC to the front of the pancake coils?
Checking the SWR with the 2th finemet cmc on the input (and with 1st finemet cmc on output) shows about 1.5W going in at resonance and a low (1:1.3 or so) swr (not enough RF for full scale set).
QuoteWhat is the output when you replace the two CMCs with each other?
When swapping the 1st cmc with the 2th, the same output is seen (~50V), no retuning needed.
Note: yesterday i measured the 2th finemet cmc (8 turns each) to be 1714uH, but it also showed a value of 1.096uH each sometimes.
I never found out why these different values were shown, but it settled finaly to the strange high 1714uH value.
Today i meaured both cmc's again and the 2th now measures a steady 1096uH for each.
The 1st cmc now shows a steady 978uH each, so no change there.
I must assume that with both cmc's having the same amount of turns ( 8 ) their inductances should be very close, so the now measured 978 and 1096 are probably the real ones.
Not sure still why i measured the 1714 most of the times for the 2th cmc yesterday.QuoteWould you test the 2nd CMC at the output too? (without using the 1st CMC)
Removing the 1st cmc from the input, now only the 2th cmc on the output, needed to retune.
We have 126V now with the 2th cmc on the output (was 113V with 1st cmc there), SWR about 1:1 with 4W in on resonance see screenshot.
Swapping back to the 1st cmc again on the output and retuning (no cmc on input) we again have 114V with swr 1:1 at 4W.
So the 2th cmc with the slightly higher inductance (1096 v 978uH) gives higher output.
Itsu
Quote from: gyula on 2021.02.15, 21:26:02
Perhaps try to ask this question from yourself: why is it that the scope shows less output voltage without the CMC? C.C
Very likely an asymmetrical measurement setup (i.e. without a CMC or CMCs in this case) uses up power what the CMC can confine and route to the load.
Very likely, with no CMC, some of the output signal can leak to the enviroment via the ground of the instrumentation, this may also mean we do not always get full output in certain setups, this much depends on the particular measurement circumstances.
This is my take on this.
Gyula
OK, concerning your question above, so you mean a groundloop problem.
I know i have my grounded FG as source and use my grounded scope on the output, so i changed to the differential probe measurement method for measuring the output.
Yellow and blue probe on each side of the 690K resistor, and the math function calculating blue - yellow, see screenshot.
The same output as 1 post / screenshot earlier (126V) now shows 205V!
Guess i have to revisite all my measurements done up till now using the differential probe method.
Itsu
Hi Itsu,
First on you last but one reply, my answer:
It looks like when there is a CMC at the input of the pancake coils, then SWR is around 1:1.3 and without it the SWR improves. You mention about 1.5 W going in with the 2nd CMC at the input: the CMC seems to influence input matching unfavorably. And then you find 5 W input without any CMC at the input (all these shown by the power and SWR meter, right? you do not have a current probe).
This may also mean the lower input power (whatever causes it to be lower) is to blame to have the smaller voltage of 50 Vpp or so at the output.
Regarding your Note: As far as I know, these types of Finemet cores are not made of solid core material but of very thin Finemet laminated pieces of layers on top of each other to make up for the h size (height). And these laminations are put into a plastic case holder to give them a certain rigidness. I think it is possible that when you handle this core, the laminations inside may change their position a little and due to the high permeability, the effect may manifest as you experienced. Perhaps measure this core again with the L meter and try to squeeze the core gently across its height by your fingers. Or try to flex the body of the core also very gently to see any change in the L value. This core has the higher OD versus the 1st CMC, right? so it may be more vulnerable to such small deformations. The rigidness of the plastic case also counts.
On you latest post above:
It is okay that you revisit some measurements but perhaps try to start with checking for instance, with the usual single voltage probe, the output open voltage of your FG set to a known value and then terminate it by 50 Ohm dummy load and check again etc.
Then the power amp may come because it brings in another 'ground' from the DC power supply and so on.
Question: did you find the 205 Vpp with or without any CMC ? :D
Thanks,
Gyula
QuoteFirst on you last but one reply, my answer:
It looks like when there is a CMC at the input of the pancake coils, then SWR is around 1:1.3 and without it the SWR improves. You mention about 1.5 W going in with the 2nd CMC at the input: the CMC seems to influence input matching unfavorably. And then you find 5 W input without any CMC at the input (all these shown by the power and SWR meter, right? you do not have a current probe).
Yes, i agree, and yes all these shown by the power and SWR meter (so not very reliable), no current probe right now.
QuoteThis may also mean the lower input power (whatever causes it to be lower) is to blame to have the smaller voltage of 50 Vpp or so at the output.
Right, the lower the input, the lower the output.
QuoteRegarding your Note: .......... This core has the higher OD versus the 1st CMC, right?
It seems there is something going on inside this toroid, but i cannot invoke it by squeezing or knocking, its just there, then its not, and it needs the LCR meter to disconnect then reconnect to show up.
This core has the same OD as the first one, the two are identical.
I do have 2 other bigger OD Finemet cores, but did not use then yet.
QuoteOn you latest post above:
It is okay that you revisit some measurements but perhaps try to start with checking for instance, with the usual single voltage probe, the output open voltage of your FG set to a known value and then terminate it by 50 Ohm dummy load and check again etc.
Then the power amp may come because it brings in another 'ground' from the DC power supply and so on.
The PA is fed from a PS with floating return line, so not grounded.
QuoteQuestion: did you find the 205 Vpp with or without any CMC ?
The 205V was found with the same situation as when i found the 126V, meaning with the 2th CMC on the output.
So i just switched from a single probe setup (ground lead) to a differential probe setup (no ground).
See the below diagram on the groundloop situation when using a single (grounded) probe.
I tried using a battery operated FG, but the PA did not like this floating input as it started to oscillate.
Itsu
Quote from: gyula on 2021.02.14, 21:25:32
It is okay that you can order the FT cores but I cannot tell how the lack of using such CMCs in this setup may influence the real goal of these tests.
Quote from: Itsu on 2021.02.15, 21:01:30
I understood from verpies his Post #739, see quote below, that a cmc would attenuate (unwanted) signals, so i would expect to see less (only the wanted) signal with a cmc installed.
Where do i go wrong?
The ideal goal without water is to obtain 0V amplitude at the output (the toroidal coil) when a source of the signal is applied to the input (pancake coils). Of course that ideal can only be approached.
Part of the signal appearing at the output comes from the common-mode admittance between the input an output. The role of the CMCs is to attenuate that as much as possible, leaving the unobstructed path only for differential signals.
When the E- shielding is done correctly, the S21 series measurement (https://www.overunityresearch.com/index.php?topic=3924.msg83438#msg83438) between points A and B should be very low, indicating that no common mode signal is getting through the inter-winding capacitances (C
IW) and ambient (C
AMB).
By shorting the terminals of the coils during this measurement it is assured that no current flows in the coil windings and no magnetic flux is generated/received by the coils and all energy between the input and output is transferred through the stray capacitances and the E-field, which should be effectively shunted to the RF ground by the E-shield ...and CMCs (or beads) should further attenuate whatever remains. Do they?
Quote from: gyula on 2021.02.16, 18:58:36
It looks like when there is a CMC at the input of the pancake coils, then SWR is around 1:1.3 and without it the SWR improves.
We have to be very careful when using the phrases "SWR improves" or "better SWR" in reference to this system, because this is not an RF transmitter!
Ideally, the SWR between the PA and the "tuner should be 1:1, and the SWR between the "tuner" and the Pancake coils should be
1:∞.
...and the voltage amplitude between the toroidal coil's terminals should be 0V without water.
Of course this "ideal" can only be approached.
Verpies,
thanks, i will setup this "S21 series measurement" without any water to get a base line and add the cmc's to see any difference.
Itsu
Itsu, thanks for your answers and all your efforts.
Verpies, okay and thanks for the additional explanations.
Gyula
@Gyula, Smudge
You are welcome to improve on my diagram in "Cap coupling2.png (https://www.overunityresearch.com/index.php?topic=3924.msg87826#msg87826)" to show other modes of energy transfer in this system. Perhaps you can envision other paths through which this energy can be transferred from input to output by non-H fields.
The E-field transfer through the CAMB and the free floating conductor at the top of my diagram does not have to be the only one.
A first "S21 series measurement" using as "plane of reference" the points A and B in the diagram (A = port 1, B = port 2 of the nanoVNA).
So no CMC's, (input) tuner or output circuit used.
First screenshot is with the toriod shield floating (not connected)
Second screenshot is with toroid shield connected to nanoVNA outer connectors.
Itsu
Quote from: Itsu on 2021.02.17, 15:16:35
First screenshot is with the toriod shield floating (not connected)
Second screenshot is with toroid shield connected to nanoVNA outer connectors.
You can recognize the capacitive coupling by the initial 90º phase-shift.
The difference between she shield connected and not, is stark at 4MHz: S21 Gain is -10.447dB with the shield floating and -39.252dB with the shield connected, which is almost a -30dB difference ...or 1000x less CM power reaching the toroidal coil.
Amplitudewise (│S21│ @ 4MHz) it is 0.3V/V vs. 0.011V/V which is a 27x smaller voltage amplitude when the shield is connected.
So the E-Shield is working but it could work better. Mine gives me -55dB CM attenuation @ 4MHz. Your notch in the S21 Gain is surprising at approximately 2.6MHz. This notch is -77dB deep and should be investigated.
Maybe the coil leads or shield's grounding lead inductance is resonating in tandem with the C
IW capacitance ? Try to determine what factors affect the position of this notch.
And i thought -30dB was very good.
Anyway, see picture below, when i change the 15cm long thick white litz wire going to the top of the shield with a 5cm copper strip, then this notch moves to ~5Mhz @ -75dB.
So i guess its the influence of this shield connection lead that is causing the notch.
With some tuning i might get this notch exactly on 4MHz.
Itsu
I added the CMC-2 (1096uH each) to the output coil and toke 3 measurements;
SC1 floating shield
SC2 shield connected using the 15cm litz wire
SC3 shield connected using the 5cm copper strip.
Notice the notch is gone now.
Itsu
I have minimized the lead from the nanoVNA to the shield to be a 2cm copper strip.
Now again without any CMC, tuner or output circuit, the below screenshot is the result.
Itsu
Still with the minimized 2cm copper strip to shield, but now with cmc-1 on imput and cmc-2 on output:
Itsu
So the CMCs give you almost 17dB more of common mode rejection. That means that they are working but because the │S21│ curve is sloping up, it means that the capacitive transfer of energy overcomes the CMCs reactance at higher frequencies (just like in LC circuit). How do ferrite beads work for your CM rejection at higher frequencies ...or toroidal cores used like beads (½ turn) ?
P.S.
I took my computer apart and I was offline.
I used 2 ferrite beads (the 6 holes ones) with 1 turn each at the output (adding to the cmc) and gained only -0.5dB or so @ 4MHz, no gain at higher frequency seen.
Then moved both to the input (adding to the cmc), and gained -1dB @ 4Mhz, and also no gain at higher frequency seen (-55dB @ 10MHz).
Also adding some small ferrite toroids (od 7mm) used like beads (½ turn) does not add anything.
SC shows besides the 2 cmc's on in- and output, also the 1 turn ferrite beads on input and the ½ turn toroids on output.
Itsu
One thing i did not test using verpies his proposed "S21 series measurement" test is the black and white little cmc's.
So the setup is now like the picture here: https://www.overunityresearch.com/index.php?topic=3924.msg87707#msg87707 but only with swapped cmc's (black at input, white at output).
Also these 2 cmc's together with the shield show a respectable CM attenuation (-61dB @ 4MHz).
The shield still connected to the nanoVNA using a 2cm copper strip.
Itsu
Quote from: Itsu on 2021.02.21, 19:58:05
Also these 2 cmc's together with the shield show a respectable CM attenuation (-61dB @ 4MHz).
I think so, too. -61dB is more than million times power transfer attenuation from the input to the output.
So what is your differential S21 Gain when the shield is grounded in the same manner and these CMCs are in place?
Not sure what you mean as the above situation IS with the 2 little cmc's installed and the shield grounded to the nanoVNA casing using the 2cm copper strip.
Itsu
Quote from: Itsu on 2021.02.22, 10:06:50
Not sure what you mean as the above situation IS with the 2 little cmc's installed and the shield grounded to the nanoVNA casing using the 2cm copper strip.
You are referring to common-mode S21 gain/loss.
I was referring to differential S21 gain/loss ...all other things being equal.
OK, i will test that tonight, but i expect it will be similar as the outcome seen in that link i posted above.
Itsu
Quote from: Itsu on 2021.02.22, 10:16:40
I will test that tonight, but i expect it will be similar as the outcome seen in that link i posted above.
I expect it to be very different because the pancakes' H-Field will transfer energy if it is not perfectly orthogonal to the toroidal coil and without baluns you need to connect one side to the RF ground like in Diag.3
* or Diag.4
* (whichever works best).
*
Of course, CMCs can be put in lieu of Ferrite Beads ...or in addition to them. Also, notice their placement, i.e.: Diag.1 vs. 2 and Diag.3 vs. 4.
For the Common mode measurements, i used the Diag.2 setup.
Now for the Differential Measurement, i used the Diag. 4 setup as it is the most convenient, but the nanoVNA does not like that as i got:
- WARNING - Got a non plausible data value: (25.612972260 -2.380255222)
then terminated the measurement after 10 warnings
It does gave a solution which worked:
You can disable data validation on the device settings screen.
which i did, but probably invalidated the data received.
Anyway, this is what it recorded then, see screenshot below.
Will try the Diag. 3 setup....
Itsu
Quote from: Itsu on 2021.02.22, 20:26:43
I used the Diag. 4 setup as it is the most convenient, but the nanoVNA does not like that as i got:
I'll wait for Gyula's opinion about not including the ground in the CMC. I suspect he will have a lot to say about it...
Hi Verpies,
Thanks for placing confidence in me, unfortunately I have not done such sophisticated common mode or differential mode measurements.
Anyway, I think that it would be better to include the CMCs and use Diag. 3, this way any ground loop between the sig. generator and the oscilloscope would see a relatively high impedance hence ground separation.
Gyula
It is worth remembering that the grounds are connected at the VNA (or scope) anyway.
The scope and FG are, but the nanoVNA is not grounded to the house earth system.
Itsu
I used the Diag.3 setup, which compromised the short 2cm copper strip from VNA to shield, and again the nanaVNA (saver programm) still did not like it:
- WARNING - Got a non plausible data value: (33.164096832 -3.170856953)
So again i had to "disable data validation on the device settings screen".
The below screenshot shows the resulting graphs.
Itsu
Quote from: Itsu on 2021.02.23, 16:01:21
- WARNING - Got a non plausible data value: (33.164096832 -3.170856953)
So again i had to "disable data validation on the device settings screen".
You have positive voltage gain because of the turns ratio between the pancakes and the toroidal coil. Of course that would not matter if the H-Field generated by the pancakes was completely orthogonal to the toroidal coil.
Q @Smudge: "Why is Itsu's toroidal coil picking up so much magnetic flux variations from the pancakes ?" From the previous experiments we know, that his E-filed coupling is attenuated by more than million times, so it must be H-field coupling. The downward slope of the S21 Gain supports that conclusion.
Dave Jones made a video titled "How NOT To Blow Up Your Oscilloscope", but he forgot to make an analogous one about VNAs, ...TLDR: Inline coaxial attenuators.
Quote from: verpies on 2021.02.23, 19:02:27
Q @Smudge: "Why is Itsu's toroidal coil picking up so much magnetic flux variations from the pancakes ?" From the previous experiments we know, that his E-filed coupling is attenuated by more than million times, so it must be H-field coupling. The downward slope of the S21 Gain supports that conclusion
Could it be th feed to the center of the pancake coils? If there is a small feed along the axis that will create a circular H field even though it doesn't pass through the ring.
Smudge
Itsu has previously described that the pancake coils are connected in a a parallel bucking mode. If true, then IMO you are experiencing what I had experimentally discovered as described in my post # 548 quote-
FWIW, I ran an experiment with my two flat pcb coils in parallel but wired in bucking mode such that the voltage differential between opposite coil turns is zero. In this mode, I could measure appreciable voltage/current in an H field sense coil of any shape with differential probes. The problem is, the equivalent inductance is now ~1/2 each coil thus raising the resonance frequency.
A possible solution may be a current multiplier as described in the following link-
https://www.accelinstruments.com/Applications/WaveformAmp/Magnetic-Field-Generator.html
Here, the resonance could be controlled with Cs and Cp. Just a thot.
Regards,
Pm
The toroid coil meets the criteria of an "H field sense coil of any shape" that I describe.
Pm
Quote from: Smudge on 2021.02.24, 18:57:07
Could it be the feed to the center of the pancake coils? If there is a small feed along the axis that will create a circular H field even though it doesn't pass through the ring.
Which feeds do you have in mind?
Please refer to the photo below:
The red ones are the feedlines to the pancake coils.
The green ones are the outputs out of the toroidal coil.
The blue one is the grounding for the E-shield between the pancakes and toroidal coil (the shield is covered in yellow-green tape).
Quote from: partzman on 2021.02.24, 20:51:12
...I ran an experiment with my two flat pcb coils in parallel but wired in bucking mode such that the voltage differential between opposite coil turns is zero. In this mode, I could measure appreciable voltage/current in an H field sense coil of any shape with differential probes.
How much of it was E-field coupling and how much H-field coupling ?
Quote from: verpies on 2021.02.25, 19:07:09
How much of it was E-field coupling and how much H-field coupling ?
I would have to re-visit the experiment but at time I logically concluded from my testing that it was predominately H-field coupling. I can't remember exactly why I came to that conclusion though!
Pm
Allow me to physically define my experiment mentioned above for reference. Viewing the inside face of the left pancake coil (rectangular pcb in my case), we start at the inside center and wind clockwise. Viewing the inside face of the right pancake coil, we again start at the center and wind clockwise. These faces are brought together and placed parallel to each other. The top finish winds are connected together and the bottom start winds are connected together and grounded. A voltage is applied to the connected start windings and a section of a sense coil is placed between the pancake coils and centered in reference to the start wind connection. The sense coil axis is parallel to the rectangular wires or traces in my case.
In this configuration, the gradients of the volts-per-turn on each end of the sense coil cancel. IOW, the E-field and A-field are self cancelling. The sense coil is also floated above ground by measuring each end with two 10M ohm/3.9pf probes differentially to minimize any parasitic currents.
The sense coil is loaded with say a 1k ohm resistor and the Lenz effect is noted at the input. What is causing this induction in the sense coil? It has to be the H-field IMO.
Regards,
Pm
Quote from: verpies on 2021.02.25, 19:03:09
Which feeds do you have in mind?
See attached image.
Smudge
Quote from: Smudge on 2021.02.26, 16:28:43
See attached image.
I see, but my feed wires were twisted around each other and later I switched to using a coax in place of the twisted pairs.
I think Itsu keeps them together, too unless he inserts a probe over one wire only.
These wires conduct currents in opposite directions so their H-Fields should cancel, when they are close to each other.
See attached image.