PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2024-03-29, 10:27:40
News: If you have a suggestion or need for a new board title, please PM the Admins.
Please remember to keep topics and posts of the FE or casual nature. :)

Pages: 1 2 [3] 4 5 6 7
Author Topic: Magnetic Delay Transformer  (Read 25342 times)

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056
OK I see your problem.  Graham had the benefit of an expensive multi channel scope so he had all scope probes permanently connected, he didn't have to swap ground connections about.  Also he could use differential measurements using two probes so that ground loops were not a problem.

In your case in order to keep things the same for both input and output power measurements the ISO and balun/choke are redundant and you may as well just connect the FG directly to the device.  Also I can't see how you can use two ground points at opposite ends of the CSR, surely those two scope ground connections are shorting out the CSR.  You need a different set up with just one ground point.  For the output power you don't need a CSR.  My two images below show a permanent ground connection between primary and secondary, with the scope probes grounded there.  For the input power you need the math shown on the image there, CH2-CH1 of course gives you the voltage across the primary to be multiplied by the current.  For the output power the math channel uses the load resistance.  (If you are worried about the inductance of the load resistance then use the CSR and connect in the same manner as for the input power).  I show dummy probes which are simply a resistor and capacitor in parallel (e.g. 10M and 20pF or whatever your probes are) to simulate the presence of the absent probes.  All this ensures that to the best of your ability the system is identical for both measurements.

I see your power measurements are noisy.  Maybe this is because you are using a 1M load which will not consume much power.  I think that 1M is OK when you are exploring input impedance, but maybe a lower value would be better for power measurements.

Hope this helps.

Smudge

Smudge,

thanks for the resonse, but i think we are miscommunicating somehow, probably i am not expressing myself clear enough in English.

I will never use 2 two ground points at opposite ends of the csr, i meant to show the 2 possibilities i have on each csr to put a ground lead, but not 2 at both ends.

I tried your both input and output schemes, but the thing i want to avoid is the groundlink between the 2 coils L5 / L6, i think they will cause measurement errors (it shows up on my scope), so therefor i measure input OR output, not both at the same time.
If Graham does so, i have my doubts about the results.


Anyway, the 1MOhm resistor indeed causes very low current, almost unmeasurable (300uA) for my current probe so i will try some lower values.

Here a video i made yesterday omitting the ISO xformer / balun and taking some power measurements.
Some things puzzle me, like why does the input power turns negative at 2.8 / 2.9Mhz, and why do i not see a 0° phase at the output signal (V and I)?
https://www.youtube.com/watch?v=7o6JZluTAjM

I will try a lower resistor, like 100K to see if it improves.

By the way, when using your output calc CH1²/R for the value in the video, i get 11² (rms) /1.022M = 121uA

Itsu
   

Group: Moderator
Hero Member
*****

Posts: 1828
Smudge,

thanks for the resonse, but i think we are miscommunicating somehow, probably i am not expressing myself clear enough in English.

I will never use 2 two ground points at opposite ends of the csr, i meant to show the 2 possibilities i have on each csr to put a ground lead, but not 2 at both ends.

OK, sorry for the misconception.  I now see you have a hall sensor current probe so you don't need the CSR, but maybe your probe can't handle the high frequency so you need the CSR for accurate measurements.

Quote
I tried your both input and output schemes, but the thing i want to avoid is the groundlink between the 2 coils L5 / L6, i think they will cause measurement errors (it shows up on my scope), so therefor i measure input OR output, not both at the same time.
Yes I know that.  And to keep things more or less exactly the same for both measurements I suggested using the dummy probe.
Quote
Here a video i made yesterday omitting the ISO xformer / balun and taking some power measurements.
Some things puzzle me, like why does the input power turns negative at 2.8 / 2.9Mhz,
Well that is exactly the effect we are looking for.  If the input resistance (real part of the input impedance) goes negative then that is what happens, the unit feeds back power instead of consuming it.  So I am glad you have found this, even if it is a false reading.  I suggested the 1M load because I had the feeling that the negative input resistance would show up better without a serious load on the secondary.  Maybe we should concentrate on that and use that power feedback somehow, and forget about taking power from the secondary.
Quote
and why do i not see a 0° phase at the output signal (V and I)?
Well looking at the video I see generally a 180 degree phase which simply means you have your current probe the wrong way round.
Quote
By the way, when using your output calc CH1²/R for the value in the video, i get 11² (rms) /1.022M = 121uA
You mean 121uW, but yes that's about right.  With such low output power the input is really just feeding losses (or maybe not if that negative power is real).  If your other output power measurement (average of V*i) is giving a much larger answer look for a math error (like using 1K instead of 1M).  But I think I would concentrate on that negative input power, maximise that to see (a) whether we can get self oscillation and (b) whether we can get usable power out somewhere (maybe from the secondary) when it is oscillating.

Smudge
   

Group: Moderator
Hero Member
*****

Posts: 1828
Itsu,
If your negative input power is real, then your thingy is feeding power into the internal 50 ohm impedance of the FG.  You are measuring V*i as power, could you try the math channel measuring V/i as resistance?  Then is there a frequency where that negative R is a maximum?  If so then you could try disconnecting the FG, putting a resistor near that value across the primary, then giving the circuit a kick from the FG to get it started and see what happens.  You may be closer to getting self oscillation than you realize.

Smudge
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

I did use 1 Ohm CSR's when using the ISO xformer / balun setup, but switched over to the
current probe (DC to 50MHz capable) using the battery operated FG (to minimize ground leads).

The output signals (V and I) look more to have 90° or so phase shift, not 180° which is strange to
me when being in parallel resonance.

Indeed, i mean 121uW for output power.

Anyway, i will redo the input measurements and use V/I (resistance) to check for maximum negative R tonight.

Looking at the video of the input measurments, it looks like the I signal is shifting from leading to
lagging the V signal pointing to shifting from a capacitive reactance to inductive reactance when going
from 2.5Mhz to 2.8Mhz.

More later,  Itsu
   

Group: Moderator
Hero Member
*****

Posts: 1828
Looking at the video of the input measurments, it looks like the I signal is shifting from leading to
lagging the V signal pointing to shifting from a capacitive reactance to inductive reactance when going
from 2.5Mhz to 2.8Mhz.
Were it just a reactance change the math power would not become negative.  No, I think there is more to it than that, and I am excited that this is showing what I have predicted all along.  There will be a frequency where that negative power is a maximum and IMO that will coincide with the frequency where the input resistance is maximum negative.

With regard to the output waveforms you are right, the current is at 90 degrees to the voltage (I got confused and looked at the math trace).  Maybe this has something to do with your current probe being so close to your variable capacitor.  There will be displacement currents flowing from plate to plate and if some flow through the current probe this will give a false reading.

Edit.  I might add that because of the time delay along the magnetic path, what seems to be resonance at the primary may not be resonance as seen at the secondary.

Smudge 
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

I used the same setup again, only increased the frequency to 3MHz to have my variable cap
variable plate in the middle at resonance.

I went up to 3.7Mhz showing the negative wattage using the math calculating V*I as before, see screenshot 1
I then changed the calculation over to V/I (resistance) and got the signals as shown in screenshot 2

Decreasing the vertical scale on the Math trace still does not put it into screen limits, so not sure
how to interpret this signal.

Video here: https://www.youtube.com/watch?v=_9vRq1r6P3U

Does the fact that all the way up to 5MHz the wattage and resistance stay negative says something?
I mean its not just a small area where this happens.

Concerning the output V and I not being in phase, i tried a longer loop on the 1MOhm resistor putting
the current probe about 10cm above the var cap, but still the V and I stays at 90° when in resonance.

Itsu
   
Group: Professor
Hero Member
*****

Posts: 2992
  Those are strange results, all right.  How can you have negative resistance and negative wattage?
   

Group: Moderator
Hero Member
*****

Posts: 1828
They are not strange to me, negative R and negative power are predicted using classical transmission line theory if the line has a reactive Z and is terminated by a capacitance.  It was silly of me to ask for the math channel to compute V/i as that produces infinities when the current goes through zero.
Smudge
   

Group: Moderator
Hero Member
*****

Posts: 1828
I have been doing some calculations using data from Itsu's first video where the negative power was near maximum.  The numbers were:-
Power W = minus 10.55 milliwatts
Voltage V = 1.667 volts rms
Current Itot = 37.7 milliamps rms
The ratio of W/V gives me the real component of current which calculates at Ireal = minus 6.329 mA.
Then V/Ireal gives us the negative input resistance which is R = minus 163.4 ohms.
We can calculate the angle between Ireal and Itot as theta = cos-1(Ireal/Itot) giving theta as 80.336 degrees.  And Pythagorus gives us the imaginary (reactive) component of current as Iimag = 37.165mA.
These vectors are shown in the image below.  It is seen that the voltage leads the current by 99.66 degrees, i.e in excess of 90 degrees.  It is that angle exceeding 90 degrees that accounts for the negative input resistance and the negative power.  Negative input power simply means that the device is feeding back energy to the source, and not consuming energy.  Of course there is also that reactive current where energy sloshes back and forth but the average power transfer there is zero.

We can use the ratio of voltage to reactive current to get the input reactance which calculates at X = 44.854 ohms.  Since the voltage leads the current this is an inductive reactance omega*L, so we can calculate the input inductance as L = X/omega giving an input inductance of 2.462 microhenries at the 2.9 MHz frequency.  The image below shows the input as seen by the FG as that 2.462 uH in parallel with the negative R of minus 263.4 ohms.

If we wish to make this system self oscillate (which we do) we must use something that will supply that reactive but wattless current Iimag.  We can't use the FG because that has an internal resistance that dissipates power.  But we can use a capacitor, and of course if that C is resonant with the input L then it provides this current automatically.  That value of C calculates to be 1.224 nanofarads.  The input circuit then looks like that shown below and that should self oscillate provided that the ESR of the capacitor doesn't swamp the negative R.  Transferring the shunt R to a series r in a resonant circuit is done by r = L/(C*R) and this gives an equivalent negative series r as minus 7.63 ohms.  So if the ESR of the 1.224nF capacitor is less that 7.73 ohms we could win this battle.

Smudge

 
   

Group: Moderator
Hero Member
*****

Posts: 1828
I used the same setup again, only increased the frequency to 3MHz to have my variable cap
variable plate in the middle at resonance.

I went up to 3.7Mhz showing the negative wattage using the math calculating V*I as before, see screenshot 1
I then changed the calculation over to V/I (resistance) and got the signals as shown in screenshot 2

Decreasing the vertical scale on the Math trace still does not put it into screen limits, so not sure
how to interpret this signal.

You won't get it within screen limits as it goes to plus and minus infinity! Sorry I asked you to do it.  If the math doesn't get some sort of saturation then the average showing negative R might be useful.  But we can calculate another way as in my previous post.

Quote
Does the fact that all the way up to 5MHz the wattage and resistance stay negative says something?
I mean its not just a small area where this happens.

That is predicted to happen so it's good.

Quote
Concerning the output V and I not being in phase, i tried a longer loop on the 1MOhm resistor putting
the current probe about 10cm above the var cap, but still the V and I stays at 90° when in resonance.
I would forget about the output there and concentrate on that negative power at the input.  If this is all genuine then the whole point of the exercise is to use that input and get it to self oscillate.  If I were doing this I would remove the 1Meg load and just have the secondary shunted by the capacitor.  I would try a small fixed capacitor as that large variable of yours could somehow provide a radiation path back to the source and it would be good to eliminate that possibility.

Smudge
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Impressive calculations and analysis, i can't argue against it :-)

I thought for the negative resistance to use the "between 2 cursors math" setup and avoiding the zero crossing
but it will only calculate a half cycle or so.

 
So i will loose the the 1M resistor, find a fixed cap on the secondary (~260pF) and a low ESR cap
of about 1.22nF for the input.


I have no time today/tonight, so it will take some time to come back.

Itsu
   

Group: Moderator
Hero Member
*****

Posts: 1828
My calculations were for the smaller capacitance value on the secondary you used earlier.  A higher value there will require a recalc.and that will result in a different value up front.
Smudge
Edit: see next post as you didn't use a smaller value.
« Last Edit: 2019-10-13, 10:29:14 by Smudge »
   

Group: Moderator
Hero Member
*****

Posts: 1828
Hi Itsu,
Looking back I see that your 10mW negative power at 2.9MHz did use 260pF across the secondary.  If you can repeat that with a fixed value capacitor then that would be good.  If the input tuned with a 1.22nF capacitor is going to self oscillate it will require a trigger to get it going.  I don't think you can use the FG directly as its internal 50 ohms will consume that 10mW of free power.  You can get over this by using two capacitors in series with one of them much greater than 1.22nF, but the series value is still that wanted 1.22nF.  I have used a 10 to 1 ratio for those two capacitors in the diagram below.  The FG can then be fed onto the larger value capacitor.  If you keep your scope probes in place you should witness what appears to be a very very high Q resonance as you adjust the frequency as the input circuit is no longer shunted by the FG 50 ohms.  Using more than a 10 to 1 ratio for those two capacitors takes the FG further away from shunting the circuit giving higher Q, and hopefully you will eventually reach self oscillation where the FG can be disconnected.  I have shown a switch in the FG connection so that you can quickly disconnect it if needed.
Smudge
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Smudge,

updates seen, i will go back to 2.5MHz resonance and measure the var cap capacitance exactly as i
made a guess (260pF) about it.
Then use a fixed cap of that measured value across the secondary and loose the 1M resistor.

If i again get the negative wattage above resonance i can continue with setting up the input side.

I have some 2.2nF Multilayer Ceramic Capacitors MLCC caps, so 2 in series will get me close to your
calulated 1.22 / 1.34nF value.

Itsu
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Turned out that the var. cap was set at 288pF when at resonance on 2.5MHz.

So i use a 270pF (269pF measured) MLCC cap instead making resonance at 2.56MHz.
I removed the 1MOhm resistor.

Installing the input circuit as per diagram above, i have:

2x 2.2nF in series making 1.040nF (measured), 1x 270pF parallel to above making a total of 1.304nF (measured).

The series cap across the FG is made of 1x 10nF plus 1x 3nF parallel totalling 12.7nF (measured)

It seems that now the secondary resonance point shifted to 2.96Mhz (17Vpp)
We also see a very strong resonance point at 590KHz (51Vpp) probably from the primary LC, but i will disregard it.

Hunting for negative wattage above resonance shows now very little (-244uW) negative wattage around 3.1MHz see screenshot.

Trying to let is selfresonate fails up till now.

Video here: https://www.youtube.com/watch?v=DKrE8-agtRs

Itsu
   
Group: Experimentalist
Hero Member
*****

Posts: 1665
Itsu,

In your latest screen shot in the post above, there is considerable positive offset in the current probe waveform.  This is quite visible but is also evident by the varying peaks in the math waveform.  If the current waveform (usually the problem) is not offset by any appreciable amount, the power waveform peaks will be at relatively equal peaks.

It may not be the case but since the offset is positive, it could be possible your actual input power is at a higher negative value than displayed.

Pm
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

PM,

yes, at these low current settings (5 / 10mA/div.), my current signal tends to start floating a bit.
But if i take a screenshot at the right moment, like below, it gives a similar picture/value's.
It stays in the -200 to -250uW's.

Itsu
   

Group: Moderator
Hero Member
*****

Posts: 1828
Turned out that the var. cap was set at 288pF when at resonance on 2.5MHz.

So i use a 270pF (269pF measured) MLCC cap instead making resonance at 2.56MHz.
I removed the 1MOhm resistor.

Did you then move your probes to the primary to check that you still got somewhere near minus 10mW at somewhere near 2.9MHz?  You don't show this in the video.  If those numbers have changed significantly then the calculations that I did won't hold.  That initial exercise was to rule out any radiation feedback from the large variable C, i.e a stray E field coupling back from that capacitor to the front end.  If the "good" result of minus 10mW is mainly due to some radiation feedback it doesn't necessarily mean that it can't be repeated using smaller capacitors, it just means we would have to supply that feedback path.

Assuming that you did do this we continue.
Quote
Installing the input circuit as per diagram above, i have:

2x 2.2nF in series making 1.040nF (measured), 1x 270pF parallel to above making a total of 1.304nF (measured).

The series cap across the FG is made of 1x 10nF plus 1x 3nF parallel totalling 12.7nF (measured)

It seems that now the secondary resonance point shifted to 2.96Mhz (17Vpp)
That's OK.
Quote
We also see a very strong resonance point at 590KHz (51Vpp) probably from the primary LC, but i will disregard it.

Hunting for negative wattage above resonance shows now very little (-244uW) negative wattage around 3.1MHz see screenshot.
I will go over my calculations again to see whether I have made an error.  In the meantime you might go back to you variable C so that you can adjust that slightly and see whether that -244uW can be increased.
Smudge
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Smudge,

i did briefly mention that the negative value's are still there using the fixed 269pF cap, around 1:14 in the video.

A quick check just now (cold equipment) showed around -6mW @ 3.5MHz (FG directly to the primary), but i will double check later today.

The 590MHz resonance point might be the primary L 49.2uH and the 1.3nF C.

Perhaps i should loose the breadboard with its stray capacitance etc.

Itsu
   

Group: Moderator
Hero Member
*****

Posts: 1828
Smudge,

i did briefly mention that the negative value's are still there using the fixed 269pF cap, around 1:14 in the video.
Sorry I missed that.  The audio on my old PC isn't working so I am using subtitles and in order to catch the best frame I played back in slow motion.

Quote
A quick check just now (cold equipment) showed around -6mW @ 3.5MHz (FG directly to the primary), but i will double check later today.
OK that's good

Quote
The 590MHz resonance point might be the primary L 49.2uH and the 1.3nF C.
I agree

Quote
Perhaps i should loose the breadboard with its stray capacitance etc.
Well it may be that it needs a more precise alignment between the notional 1.22nF at the front and the C at the secondary.  So it might be worth trying to use the variable C again and tweaking this looking for a sweet spot where the Q takes off.

Smudge
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Reinserting the var. cap on the secondary, using your input circuit and looking for a sweet spot, i find
best results when the var. cap is at its max. (345pF).

The negative input wattage is then -2mW @ 2.71Mhz, see screenshot.

The problem with this is that the resonance frequency is very close (2.69Mhz) and i have a negative
input wattage there as well of -1mW  :D

Itsu
   

Group: Moderator
Hero Member
*****

Posts: 1828
Itsu,
Something has worried me about that inductive input circuit.  You would normally expect the capacitive loaded transformer to have a positive (inductive) reactance below resonance, passing through zero at resonance and becoming negative (capacitive) above resonance.  Graham's measurements showed this classical S shaped curve.  Yet you are showing an inductive effect where the the voltage phase leads the current phase.  Then it struck me that the Hall effect current probe must insert some series inductance that could be significant at high frequencies.  Does the specification for your probe give you a value for this inductance?

If not could you please simply feed your FG onto a non-inductive resistor (say 100 ohms) and measure the voltage and current and calculated power in the way you have been doing?  That will tell us whether the probe's inductance is influencing the results, and if it is then my suggested input circuit from the FG is all wrong.  It will tell us what the probe's inductance is and then I can recalculate everything.  The image below shows the apparent input to the MDT as a series circuit equivalent to the parallel one, where the input has a series negative resistance of 7.43 ohms (using the 1.667V, 37.7mA, -10.55mW at 2.9MHz data that I previously used).  If the probe is responsible for that series inductance, then the image also shows the real input to the MDT as that -7.43 ohms in series with the expected input capacitance.  If I know the inductance of the probe I can then calculate that input capacitance and re-jig the manner in which to connect the FG.
Smudge
   

Group: Moderator
Hero Member
*****

Posts: 1828
It may not be an inductance effect of the current probe, it may be propagation delay.  Here is an extract from Tectronix "ABCs of Probes" (my underlining added).

Propagation delay is usually only a concern when comparative
measurements are being made between two or more
waveforms. For example, when measuring time differences
between two waveforms, the waveforms should be acquired
using matched probes so that each signal experiences the
same propagation delay through the probes.

Another example would be making power measurements by
using a voltage probe and a current probe in combination.
Since voltage and current probes are of markedly different
construction, they will have different propagation delays.
Whether or not these delays will have an effect on the power
measurement depends on the frequencies of the waveforms
being measured. For Hz and kHz signals, the delay differences
will generally be insignificant. However, for MHz signals the
delay differences may have a noticeable effect
.

And that noticeable effect could well be the appearance of the current waveform lagging the voltage waveform.  It could explain why the input to the MDT appears to be inductive when it should be capacitive.  If we can determine that difference in propagation delay by measuring a known resistor then I can recalculate things as stated in the previous post.

Edit: And my recalculation may well remove that negative power reading:'(

Smudge
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Smudge,

updates seen, i know about the delay of the current probe, and normally i try to compensate for this
delay by adjusting the voltage probes (±10ns) and or current probe (±10ns).

At 2.5Mhz there is slightly more then 20ns delay, so i set the voltage probes to +10ns and the current
probe to -10ns so the delay should be largely compensated for, but indeed not completly.

So let me measure the delay using a (normally i use a 50 Ohm 1% inductionfree) resistor at 2.5Mhz.

The Tektronix A6302 current probe (with AM 503B controller) specs do not mention any inductive influence,
see picture of its specs taken from the PDF below.

I will use a 100 Ohm 1% inductionfree resistor for making current measurements on the input later tonight.


Itsu
   

Group: Elite Experimentalist
Hero Member
*****

Posts: 4056

Using a 50 Ohm 1% inductionfree resistor to measure the propagation delay belween the voltage and
current probe at 2.5Mhz with a 10Vpp square wave.

Screenshot 1 is with both probes uncompensated,
screenshot 2 with voltage probe compensated with its max. +10ns, and the current probe with its max. -10ns.

So we are left with a 8.8ns propagation delay at 2.5MHz.

Itsu
   
Pages: 1 2 [3] 4 5 6 7
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2024-03-29, 10:27:40