OK so I have decided to try and build a Voltage trimmed 45.525 Mhz crystal oscillator & 16 Watt amplifier to drive a coil for NMR investigation.
I have ordered a cheap second hand crystal stamped 45.666MC and am hoping I can trim that with capacitance down to the required frequency.
First though I wanted to get the 16 Watt final stage built see attached circuit.
I already have a DDS and preamp from times past.
Today I wanted to try building the first step of the Amplifier, the input PI filter or CLC filter.
I only have a few variable capacitors that are adjustable up to about 80pf and want to find the required cap values to replace these with fixed values.
The input capacitor i had to parralel the variable cap with a 220pf capacitor (Little red Wima)
There are photos of the DDS preamp and the PI filter setup.
Also there are scope shots showing attenuation at different frequencies.
I was not sure how you adjust the 2 vriable caps, i was getting distortion on the incomming (Yellow trace) waveform at lower frequencies, i assume this is ok if the output waveform is good, i switched fft on so i could minimize the 2nd harmonic.
EDIT looks like i am getting just over 2 Watts in my 50 Ohm load
I will need to heat the iron filings at some point so have just ordered 1 of these, so I can pack the tube with iron and seal the ends with fire clay with a wire protruding from each end, I hope to use the irons resistance as the heater.
100mm Long
10mm External Dia
2.0mm Wall 6mm Internal Dia approx
Heat resistant to Over 1000 C
Heat Shock Proof
https://www.ebay.co.uk/itm/331525223850
Hi Peter,
If I understand correctly, your DDS drives a preamplifier which produces about 2W in a 50 Ohm load attached to the output of the input PI filter.
Well, if you wish to use the IRF610 power MOSFET for the 16W amplifier, then you need to include roughly 140 pF gate-source capacitance (Cgs) across the 50 Ohm load because this capacitance will be present when the MOSFET is included in the circuit.
So the input PI filter needs to be examined also if you simply add a 150 pF capacitor in parallel with the 50 Ohm, to represent the Cgs. 140 or 150 pF, not critical because it will change eventually in the function of the drain voltage anyway. Data sheet says 140 pF typical value at 25V DC drain-source voltage for an IRF610.
Regarding your 45.666 MHz crystal, you will probably have difficulties in pulling it down (roughly 141 kHz) with capacitance only, a series L inductance would most likely be needed. But it can be done for sure.
Here is an example for a good oscillator in which two 14050 kHz crystals (connected in parallel) are pulled between 14000 and 14065 kHz, with good stability. https://amqrp.org/projects/NB6M%20contestvxos/VXO%20Experiments.html (https://amqrp.org/projects/NB6M%20contestvxos/VXO%20Experiments.html)
Because the 45.525 MHz is roughly 3 times as high as 14 MHz, the 4.7 uH series coils could be linearly reduced by 1/3 i.e. use 4.7/3 uH for each but of course a single coil could be used there, the real uH value greatly depends on the crystals.
As you can see the 2nd schematic includes a single 2 uH coil only for the 2 paralleled 14060 kHz crystals. You may wish to have at least two 45.666 MHz crystals, it helps pullability (but should work with a single one too).
Notice the varicap tuning possibility at the bottom of the series coil(s), this you can use, so either a varicap or instead simply put there a good quality variable air capacitor of at least 40-50 pF with a turning knob.
Gyula
Quote from: Peterae on 2019.01.20, 22:10:52
I will need to heat the iron filings at some point so have just ordered 1 of these, so I can pack the tube with iron and seal the ends with fire clay with a wire protruding from each end, I hope to use the irons resistance as the heater.
That tube is pretty thin! I expected something over 10mm ID.
Anyway, I hope your iron filings are really a fine powder because large filings will have a problem with the skin effect just like solid iron.
Also, two common N40 NdFeB magnets without a yoke will provide only 500mT. See:
https://youtu.be/B015P0XFl9g?t=76
BTW: Heating the iron powder in an inert atmosphere for 2 days at 400C improves the iron's NMR line sharpness 3 times.
Hi verpies
The tube is thin but are we concerned about electrons exiting the iron, I am going to use the magnets as detectors by wiring them to a very sensitive charge detection amplifier, so if we do get beta particle collision to the magnets we will see a change in charge and know for sure the device is working.
The iron filings are not that small and I'm not even sure of their purity, I am looking for a source of better quality, I know that if I dissolve iron electrolytically in copper sulphate (pure fine copper displaces) and then dissolve zinc electrolytically I will get pure fine iron particles displaced but am not yet sure how easy it would be to wash, clean and dry them without oxidation (inert hydrogen atmosphere is needed really and sounds a bit dangerous to me), we are going to heat the iron filings to a temperature above their curie point so why would we get the skin effect??
we would be annealing the iron anyway as part of the heating cooling process.
Hi Gyula
Very interesting about the series inductor, I will have a play when it arrives.
and interesting about the capacitance of the fet this will be in series with the 10nF decoupling capacitor so I am looking at about 138pf, maybe I am going to have to leave a variable cap in the second cap position for fine tuning when the fet stage is fitted.
I was not sure if i should have a 50 ohm resistance across the input of the pi filter to terminate the output of the preamp, 2 watts is way more than needed to drive the irf610 says only 0.3 watts needed, i might be able to do that using the oscillator stage and the do away with the preamp altogether.
I have had success before using standard diodes as varicaps in a previous 45mhz oscillator i built and may try these again, it's a very old crystal, possibly second world war, i am hoping it's age may have lowered the cut frequency down a bit due to ageing.
I am wondering if It maybe a good idea to buy a dummy load and SWR meter to test the 16W amp stage before going on to work out the final load coil construction.
Hi Peter,
For the series coil an RF powdered iron toroidal core would be good like Amidon T-56-6 (yellow) and make say 30 turns to have a high Q 3.6 uH coil and make at least 1 tap on it, wire OD could be 0.3mm. Here is a calculator: https://www.changpuak.ch/electronics/amidon_toroid_calculator.php (https://www.changpuak.ch/electronics/amidon_toroid_calculator.php) Yesterday I quoted 2 uH series coil from the 2nd schematic but sorry it is 12 uH for the 14 MHz crystals so 1/3 of that for the 45MHz one will be close and with the tap you will have one more choice without unwinding.
Regarding the input capacitance of the MOSFET, it changes from within the same type MOSFET batch so do not care about the effect of the series 10 nF, it is really negligible. If you can leave a variable cap in the 2nd cap position, then do so to help vary the tuning/matching in any case.
If the output impedance of the preamp is far from 50 Ohm, then do you have any info what impedance is involved there? any preamp specification or maybe schematic? Not utterly important though (see below why), so no problem if you have no info.
I suggest leaving the preamp in the setup to protect the output of the DDS from any MOSFET "backfire" event, should such occur. If the 2W proves to be too high, then you can still use a passive resistor divider at the preamp input or output to reduce the RF level, the presence of such dividers can help matching between the stages to a certain degree too.
The varicap type shown in the link I gave yesterday is 1SV149 Toshiba. Works between 1V and 8V tuning voltages and has roughly 480 pF and 30 pF at those voltages, respectively. So quite a high capacitance ratio, it establishes the tuning range also for the crystal oscillator. OF course you can connect several varicaps or maybe standard diodes in parallel to increase their capacitance, the higher the capacitance change you can achieve, the wider the tuning range possibility for the crystal will be.
Yes a series inductance is needed when the frequency of a crystal is to be pulled downwards from its natural series resonance frequency. What is more, a tuneable coil would be needed but such is difficult to make to observe high Q in a solenoid form. This is why a variable capacitor (or varicap) is connected in series with a high Q coil, the idea is that by changing the capacitive reactance, the overall reactive impedance (series LC) changes and ideally the resulting reactance should remain mainly an inductive one to pull downwords the crystal.
A dummy load for 45-50 MHz could still be made from carbon resistors I believe, it maybe cheaper than an off the self one? Regarding SWR meter, well maybe it is not really needed: the load eventually will be an inductive coil, not a pure resistor. But it is up to you, to see matching at the output Pi filter first into a dummy load may be useful.
Gyula
No info on the preamp, seems to use a small chip as the main amp, looks like there's 2 series caps on the output with no termination resistor fitted.
thanks for your help gyula, we will see how it goes at the weekend ;)
I just had a go at building the Xtal oscillator.
I tried the circuit below, seemed easier to start with, I couldn't remember where I found it but I thought it was 50MHz but maybe not :(
I did not have 150pf caps for the base divider and used 220pf each cap instead, going to work out the impact of that tonight.
I used a BC337-16 instead of the 2n3904.
I soldered the crystal case to GND.
The sinewave is not perfect but a bit out of shape, the frequency is a third of the required frequency 15.21396MHz at 0volts and 15.2154 at 30V tuning voltage.
I used a hand made 830nH inductor.
When I put the output cap divider in place it stopped oscillating (22pf in series with 10pf) maybe the value are wrong for the required frequency? I think these need to be bigger values.
EDIT maybe the 220pf caps should be smaller for the correct frequency. I will try 47pf and 33pf caps tomorrow.
@Peterae
Is your planned RF amp capable of supporting a sequence of 45.525MHz bursts in case the more primitive CW NMR technique yields an output signal amplitude that is too low?
e.g.: if you put a quick on/off RF switch between the oscillator and the amp...
P.S.
Did you get my private message about the "other source" of an RF amp ?
Hi verpies
I am sure I can find a way to turn the oscillator on and off, maybe using a switchable buffer stage.
Yes sorry I meant to reply to your email, no need yet, let me try at 16 Watt and get the system built ;)
EDIT Looks like i have a 3rd Overtone Crystal
Ah this looks good for what I need
Quote from: Peterae on 2019.01.25, 19:45:13
Ah this looks good for what I need
If you don't want to bother : :)
https://www.ebay.com/itm/Vertex-Standard-VX4000L-6-Meter-Low-Band-37-50Mhz-Transceiver-70-watt-Yaesu/273667720755?hash=item3fb7de0633:g:wUkAAOSw5EpcFsyH
Cheap in the USA but almost doubled due to shipping to Europe >:(
Very Tempting F6FLT
Although I am enjoying myself so far and trying to learn at the same time.
I think I can put a tuned LC in the emitter to force 3rd harmonic, I will try tomorrow. 200nH and 63pf
Hi Peter,
You can see some more info on overtone oscillators here
http://www.fishpool.org.uk/overtone_oscillator.htm (http://www.fishpool.org.uk/overtone_oscillator.htm)
The 'trick' is to tune the coil in the collector to the needed overtone frequency with the trimmer cap (7-45pF), i.e. for 45.525 MHz. This way the transistor can amplify at that frequency only because at 45.5 MHZ the collector will have a higher impedance than what is in its emitter (220 Ohm||33 pF), so oscillation will be maintained.
There is the base-emitter capacitor missing here what is included in your finding in the above schema but you can test which is better.
It often helps if you connect a few pF (4.7 to 10pF) capacitor between the collector and emitter in the schema I show. The cap across the emitter resistor (33pF) can also be varied while watching the output on the scope.
Of course you need to connect the inductance I referred to earlier to pull the frequency downwards in series with the xtal and also with a variable cap or varicap just as shown in the schema in your reply #8 or in the lower right corner in the schema in your reply #11.
Gyula
Quote from: Peterae on 2019.01.25, 20:15:22
...
I think I can put a tuned LC in the emitter to force 3rd harmonic, I will try tomorrow. 200nH and 63pf
There is a 100 pF capacitor between the emitter and the ground, it is a low enough impedance for the third harmonics, I don't think this LC circuit is necessary. Moreover it could disturb the oscillator.
It is on the collector side that this tuned circuit is necessary, and it is there, it is the 250nH / 10->50pF, to be tuned on the harmonic, not on the fundamental. The other 250nH / 10->50pF connected to the output performs filtering and impedance matching, and the capacity that couples the two LC circuits is only 1pF to improve filtering. Increasing it may allow a better transfer of power, but we will have a little more unwanted products.
One of the advantages of this Clapp oscillator is that you can do almost anything you want on the collector side without disturbing the oscillator at the emitter.
Hi F6FLT
Only just seem your recommendations, thanks ;)
I think in that circuit, he is running a xtal at it's fundamental frequency and filtering the 3rd harmonic, I now think his circuit is no good for what I wanted hence your suggestion on modifications needed ;)
I now wonder if the circuit I built yesterday would have indeed worked properly if I had sorted the diode capacitance out as below.
I re-built the circuit as per gyula suggestion, see below circuit with an inductor and diode in the gnd side of the xtal.
Took a while to work out why I had a horrible waveform at the fundamental frequency, turned out my diode did not have enough capacitance for the circuit to work properly, ran out of time eventually.
Been scratching my head and am still scratching my head, I have 46Mhz in the scope shot, the crystal is 45.666Mhz, I seem to have gone up in frequency, I have seen it working at 45.657Mhz
Hi Peter,
I would suggest:
1) leave out the big coil in series with the xtal and also the varicap diode, and connect the xtal pin to the gnd directly, and see where it oscillates. (write down the frequency)
2) now lift up the xtal pin from the gnd and insert the coil in series, the coil's other leg goes to gnd, and see the oscillating frequency (write down the frequency)
3) lift up the series coil's leg and insert a variable cap of at least 40 pF or higher and see how now the frequency changes (write down the frequency range)
If your scope probe is directly tied to the collector shown as TP1 (and the gnd clip to the common negative rail), then the roughly 10-14 pF probe capacitance is built into the circuit. A better place is via output 5, via the series 560 Ohm + 10nF but the best would be an active buffer stage, probably your preamplifier would be good for this buffer stage? Hopefully it has not got a too low input impedance to load down the oscillator output too much to kill oscillation. Nevertheless, couple the oscillator output via the 560 Ohm + 10 nF to the input of the preamp and see the amplitudes at the input and the output of the preamp (and measure frequency at the output of the preamp).
Gyula
Thanks gyula
We will see how we get on tomorrow ;)
Hopefully if I can go up that much I can I can go down in frequency as well, seems very stable at 46Mhz.
I only had a 50-120pf variable cap for the collector and had to wind a small inductor to reach third order, I tuned it using my GDO and scoping across it, so is probably set a little high due to probe capacitance.I have been using a 100nF to feed the scope probe from the collector to stop the probe interfering with measurements.
I did play with the large inductor coil spacing and pulling the coil turns apart a little it dropped to 42.5MHz but seemed unstable at this frequency, so something strange going on at the moment, your suggestions above should shed some light i think.
Well, the 100 nF cap to feed the scope probe from the collector does not stop the probe interfering with the measurements, unfortunately. The best would be to use the preamp as I suggested above. If you do not yet wish to use it, then use either the 560 Ohm series resistor as shown in schema or even a higher, 1-2 KOhm series resistor and the 100 nF in series if you wish, The cap does only DC blocking, otherwise a short circuit at 45 MHz.
When you had the 42.5 MHz oscillations at the output, it was already an LC oscillator and the xtal did not work.
One more thing if I may: What may need to be changed (not to fight so much with the oscillator) is the big solenoid coil you wound for the xtal to pull it down. If you happen to have any Amidon or other powdered iron toroidal core for that coil (I mentioned the yellow one above with tap), that would be good because this big coil can couple easily to the collector coil and affect operation in an unwanted way.
It is okay that the axles of the two coils are perpendicular to each other but the big air core coil may have too big stray field around it, reaching the collector coil. This is why a confinement into a toroidal core would be useful.
IF you do not have such, try to order some, together with some trimmer capacitors. But I understand if you do not wish to invest too much into this project what is absolutly neccessary.
Gyula
I agree with Gyula. The setup must be adjusted and tested without varicap and large coil. Once everything is correct, the LCvaricap can be added (with a smaller size and a more confined field, for the reasons given by Gyula).
However, I don't really see the point of not having a quartz directly at the desired frequency. Quartzes operating at 45 MHz are commercially availabl (http://www.elecdif-pro.com/product_info.php?cPath=105_1005_1809&products_id=51629)e, they operate directly at this frequency even if they are cut for 1/3, 1/5 or 1/7 of the frequency (I even have some at more than 100 MHz). This results in a stronger signal directly at the right frequency.
Well, the frequency for the xtal needed is 45.525 MHz, the 45.000 MHz may not be good for the Meyer-Mace effect (at least as per the math formula). Such off the shelf xtals for 45.525 MHz are not readily available components, only in the relatively close vicinity like 45.666 MHz Peter found and tries to pull downwards.
I searched for off the shelf 45.525 MHz xtal (or close to this) but have not found.
Here is cheap 15.2 MHz xtal (roughly 1/3 of the target)
https://www.mouser.co.uk/ProductDetail/IQD/LFXTAL017300Bulk?qs=sGAEpiMZZMsBj6bBr9Q9aR%2fuGiDjvlISt9vCKfK%2fVG%252bPoJarV0hR0Q== (https://www.mouser.co.uk/ProductDetail/IQD/LFXTAL017300Bulk?qs=sGAEpiMZZMsBj6bBr9Q9aR%2fuGiDjvlISt9vCKfK%2fVG%252bPoJarV0hR0Q==) Using 2 or 3 pieces of such xtals in parallel the 45.525 MHz target can be achived with series LC tuning. This is for the case if the present 45.666 MHz cannot be pulled down enough.
Regarding RF toroid cores, here is a source but there surely are more of course:
http://powermagnetics.co.uk/pace-components/micrometals-iron-powder-Miscellaneous/rf-applications/t-44-17-micrometals-iron-powder-toroid (http://powermagnetics.co.uk/pace-components/micrometals-iron-powder-Miscellaneous/rf-applications/t-44-17-micrometals-iron-powder-toroid)
(I mentioned the T-56-6 already, it is also useful though recommended up to 40 MHz but still much, much better than an air core coil, the 40 MHz is not a rigorouos limit of course).
Gyula
First I disconnected the long inductor and grounded the xtal end, see attached waveform seems to consist of different amplitude waveforms of the same frequency that are DC offset.
It would not oscillate with just the long inductor connected to ground and could not seem to tempt it to oscillate.
Anyway before going further I will change my tuned collector to a toroid and a lower value trimmer and order some iron cores for the inductors.
I've just topped up with polyester, ceramic dipped caps and a trimmer kit.
https://www.ebay.co.uk/itm/700pcs-24-Values-Mylar-Polyester-Film-Capacitor-Assortment-Pack-Kit-DC-100V-F1E3/113149175254?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649
https://www.ebay.co.uk/itm/1X-1200-Pcs-24-Values-Ceramic-Capacitor-DIP-Monolithic-Multilayer-Ceramic-C-O1R8/283265583025?epid=20023293182&hash=item41f3f1c7b1:g:l2gAAOSw3LVb8Onk:rk:1:pf:0
https://www.ebay.co.uk/itm/80x-6mm-Plastic-Cover-Trimmer-Variable-Capacitance-Capacitor-Kit-for-Arduino/401615234996?hash=item5d822227b4:g:wZ8AAOSwB19b8Eou:rk:4:pf:0
I need to get a 50Ohm load for my amp 25Watt.
Question as we are going to try and drive the output coil at resonance I am going to need a tuning cap to be paralleled up to the final coil, I imagine this is going to need to be a high voltage vane type, what sort of voltage would be a good choice to order? I know this is going to be low pf as we want to get as many turns on the coil as possible.
Well, at least 500V rated variable would be needed as a minimum value, have not studied the choices. Unfortunately, the final coil's iron core will have a certain and yet unknown permeability at 45.5 MHz, so the maximum number of turns somehow will be limited just because it may let you using only a few pF (say less than 10 pF) tuning capacitor. Putting this otherwise: you are limited to use only as many turns as would let use a max 10 pF variable because the coil self capacitance and the stray capacitances will possibly be in the 5-7 pF range (a rough estimate).
Just curious: the scope probe was at the open end of the 560 Ohm+10nF when you took the scope shot, right? 8)
It is good you will have toroid cores, make sure to provide a tap or two for easing tuning. Until a clean 45.6 MHz output sinewave is achieved, by trimming the emitter capacitor too etc. do not use any series coil with the xtal.
Gyula
I may have used a 100n in series with a 560 Ohm, but yes this was in place and scoping at the end.
You say don't use a series inductor with the xtal, you mean only while we get a good 45.6MHz but it will be needed to lower the operating frequency?
Yes I do.
Re on the probe at the 560 Ohm + 10 or 100 nF end: for such probe loading the output, the output is virtually unloaded, so the LC circuit tuned to around 45.5 MHz should have a nice maximum voltage value as the trimmer cap is tuned.
Try to drive the input of your preamp via the 560 Ohm + 10 or 100 nF cap and see the amplitude at the input of the preamp. this would give an indication of the input impedance of the preamp. Try to tune the trimmer in the collector to maximum if needed.
Gyula
verpies
The on off bursts, do you have a time period for the on and for the off, just trying to plan ahead ;)
PS just ordered 500 grams of high purity 100 mesh iron powder.
Things to work out and do.
Get oscillator working at 45.525Mhz and be variable +/- 100Khz.
Build switchable buffer stage.
Complete Amp stage.
Find best coil diameter/ length for resonance at required frequencies.
Method of heating fine iron powder in quartz tube, preferably by passing a DC current through the iron powder.
Build very sensitive charge detector and connect to magnets to detect Beta/ electron impact.
Monitor current through iron tube as this maybe indicative that we have a working device.
static bias magnetic field, is now clear that a steel U shape is needed with magnets to concentrate field for the required strength of 600mT , experimentation required, I will probably use 100mm long N50 magnets.
Method of operation
Heat Iron above Curie Temperature and stabilized
Turn off Heating
Set test frequency
Turn on Frequency Bursts for a set time period.
Test for charge particles, according to Coleman/Gilespie device will run for a while.
Change to next test frequency.
Go back and start heating cycle.
Quote from: Peterae on 2019.01.27, 16:18:07
I need to get a 50Ohm load for my amp 25Watt.
Hi Peter,
You could consider such RF resistors https://www.ebay.co.uk/itm/153312707527 (https://www.ebay.co.uk/itm/153312707527)
You could mount it on a heat sink and fasten a BNC or SMA female socket to the heat sink.
Or use just a 30-40 cm long piece of coax cable and bolt its braid to the flange of the RF resistor (which is mounted by two bolts onto a heatsink) and solder the center conductor of the coax to the small tip of the resistor (there is 50 Ohm between this tip and the flange).
Otherwise, here is construction of dummy loads using such RF resistors:
https://pa0fri.home.xs4all.nl/Diversen/Dummyl/dummyloadeng.htm (https://pa0fri.home.xs4all.nl/Diversen/Dummyl/dummyloadeng.htm)
An off the shelf 25W dummay load: https://www.ebay.com/itm/263706066179
Gyula
Thanks Gyula just ordered the resistor, I have a sma lead and cpu heatsink ;)
Some of this stuff is going to take a while to arrive :(
Okay, and you would need some thermal grease to insure good heat conduction between the flange and the heat sink.
https://www.ebay.co.uk/itm/382244421173 (https://www.ebay.co.uk/itm/382244421173)
I found kinda professional mounting method, just for information to see, and read also the PDF file attached.
Gyula
Quote from: Peterae on 2019.01.27, 21:48:45
The on off bursts, do you have a time period for the on and for the off, just trying to plan ahead ;)
Microseconds for Pulsed NMR. I cannot give you a concrete interval because it depends on the strength of the RF field. The stronger - the shorter.
For CW NMR there are no bursts.
Quote from: Peterae on 2019.01.27, 21:48:45
static bias magnetic field, is now clear that a steel U shape is needed with magnets to concentrate field for the required strength, experimentation required, I will probably use 100mm long N50 magnets.
It helps but it is not absolutely necessary for iron and other ferromagnetic materials. Strong uniform DC bias field is necessary for paramagnetic materials, though.
Perpendicular Tx and Rx coils help a lot with isolation and sensitivity.
Impedance matching them is a bitch.
Using an RF switch to disconnect the Tx coil after a burst so the Tx amplifier's impedance does not load down the nuclear oscillations goes along way to improve the amplitudes at the Rx coil.
Quote from: verpies on 2019.01.29, 23:51:09
Microseconds for Pulsed NMR. I cannot give you a concrete interval because it depends on the strength of the RF field. The stronger - the shorter.
For CW NMR there are no bursts.
So is it ok to have an extra long break between pulses so that we make sure we are long enough say we were to guess worst case RF field strength and set time period to turn off the generator.
Quote
It helps but it is not absolutely necessary for iron and other ferromagnetic materials. Strong uniform DC bias field is necessary for paramagnetic materials, though.
What was the field strength you indicated that i am trying to obtain from the magnets, i know you posted a figure somewhere but cannot find it.
Quote
Perpendicular Tx and Rx coils help a lot with isolation and sensitivity.
Impedance matching them is a bitch.
Do i need a Rx coil? according to Meyer and Coleman my output power should appear longitudinally through the iron in half wave bursts.
Quote
Using an RF switch to disconnect the Tx coil after a burst so the Tx amplifier's impedance does not load down the nuclear oscillations goes along way to improve the amplitudes at the Rx coil.
The Tx coil is tuned, if we remember from McFreey maybe the tuned coil LC may pick up energy from the beta emissions and back feed or even self destruct if the feedback energy is too great,disconnecting the LC from the amplifier may make thismore likely.?
Quote from: Peterae on 2019.01.31, 19:10:00
So is it OK to have an extra long break between pulses so that we make sure we are long enough say we were to guess worst case RF field strength and set time period to turn off the generator.
Long breaks between RF bursts are OK.
Quote from: Peterae on 2019.01.31, 19:10:00
What was the field strength you indicated that i am trying to obtain from the magnets, i know you posted a figure somewhere but cannot find it.
For iron 750mT max. For other materials (paramagnetic) at least 1.2T which is easily achievable with two N52 Neodymium magnets.
Quote from: Peterae on 2019.01.31, 19:10:00
Do i need a Rx coil? according to Meyer and Coleman my output power should appear longitudinally through the iron in half wave bursts.
No, but it really helps if you want to make baby steps and get a good feel for the NMR alone.
Quote from: Peterae on 2019.01.31, 19:10:00
The Tx coil is tuned, if we remember from McFreey maybe the tuned coil LC may pick up energy from the beta emissions and back feed or even self destruct if the feedback energy is too great,disconnecting the LC from the amplifier may make thismore likely.?
IMO - less likely.
OK thanks for the figures
you say 750mT max whats the middle ground figure, I need something to aim for.
Been trying to work out how to seal the ends of my quartz tube and just found the answer.
1200 Deg C silicon glue
HIGH TEMPERATURE 1200'C ADHESIVE GLUE FOR EXHAUST FIREPLACE OVENS COLLECTORS 70 ml new
just need to work out how to apply it after forming a vacuum on the quartz tube, maybe I don't need a vacuum if I heat the iron hot enough and due to expansion I may get a vacuum or at least drive of any available oxygen when hot and hopefully the glue will go off fairly fast at an elevated temperature.
Quote from: Peterae on 2019.01.31, 19:38:08
you say 750mT max whats the middle ground figure, I need something to aim for.
Shoot for 600mT
Quote from: Peterae on 2019.01.31, 19:38:08
Been trying to work out how to seal the ends of my quartz tube and just found the answer.
1200 Deg C silicon glue - HIGH TEMPERATURE 1200'C ADHESIVE GLUE FOR EXHAUST FIREPLACE OVENS COLLECTORS 70 ml new
Does it make a hermetic seal so air does not get in over time ?
Quote from: Peterae on 2019.01.31, 19:38:08
just need to work out how to apply it after forming a vacuum on the quartz tube, maybe I don't need a vacuum
I find it much easier to push out the air with welding Argon than pulling a vacuum.
Quote
Does it make a hermetic seal so air does not get in over time ?
I wont know for sure until it arrives but
high adhesion
Quote
This a fireproof silicate adhesive designed for assembly and repairs of exhaust systems (collectors and exhaust silencers with holes in them), ovens, fireplaces, and damaged ceramic and metallic ducts – carrying away very hot combustion gas or fluid at a temperature over 1000°C. Resistant to fumes, dilute acids, water, chemicals and low temperature. It has very high adhesion and fast bonding time.
Here is a nice video about magnets in a yoke.
https://youtu.be/U8mnrO-gM8s
For good NMR experiments at least 20mm gap between 100mm O.D. disk magnets is convenient, so the magnetic yoke needs to be much bigger than the one shown in the video.
Also, the gap cannot be surrounded by aluminum spacers...or anything else that is conductive, because of RF eddy currents. Plastic in or near the gap is tolerable, but too much of it in the gap (or near it) limits the dexterous maneuverability and versatility of the magnet assembly.
(http://www.overunityresearch.com/index.php?action=dlattach;topic=3691.0;attach=30820)
(http://www.overunityresearch.com/index.php?action=dlattach;topic=3691.0;attach=30826)
The diagram above shows how a stator from a big old industrial motor can be converted to a magnetic yoke. The middle version (circular) and right version (semicircular) will work.
My local junk yard has many big old industrial motors with cracked casings, bent rotors ...but intact stators. They sell them by weight according to their scrap value.
3D print a hollow former and fill with iron resin mix, can be any size and not limited to the stator that needs to be found
I got some decent 10-40pf variable caps.
Still waiting on the cores but found some in my box that gave appropriate uH when a few turns were wound onto them.
I changed the fixed capacitor in the emitter to a 10-40pf variable and this cured the horrible waveform I had last time, seems this value is quiet critical for correct operation, there is a picture of the new setup with 2 variable caps, 1 in emitter and 1 in collector in parallel with the new 200uH inductor.
Scope shot 2 is with the xtal grounded one side
Scope shot 3 is with a 1uH inductor wound on a toroidal core in series with the xtal to ground
Scope shot 4 is with a 47pf cap in the ground leg of the xtals inductor leg.
Clean signal!
Now on the way to amplification?
Hi Peter,
Nice results.
You wrote 200 uH inductor, I guess you meant 200 nH? if that referred to the toroidal core shown with 2 turns?
For a start the core is good with its 2 turns but 200 nH would need at least 61 pF tuning capacitor for 45.5 MHz, so
maybe the 200 nH measurement is to be checked? The goal is to have 15-20 turns on a good core, not 2 or 3.
I suggest using at least a 600 nH coil and the 40 pF trimmer set to half way (20-21 pF) will give resonance at 45.5 MHz.
When this is done, a second coil of say 1 to 10 uH seems to be needed in series with the xtal in the ground leg. Maybe this xtal is not so active for pulling it down, I do not know, the 1 uH in series with it caused roughly 3 kHz less frequency only.
Gyula
Thanks Gyula
Yes sorry I made a mistake on the inductor value ;)
I don't seem to be able to get many more turns on that unknown core, if the inductance goes up it drops to fundamental frequency, I will now wait for the new cores before proceeding, maybe the xtal cannot be pulled down enough, we will then revert to the DDS board which is ok but I wish I had better control of it as the tft processor that came with it is a bit mickey mouse.
Bit worried about the iron powder, when I picked it up from the post office I faced questions of what it was, I hope there's no illegal uses GULP, and to make it worse I only ordered 500 grams which was plenty but now I am going to build a large core I estimate I will need another 1kg bottle of it, see pictures below of the former for the core.
I ordered 2 N52 60x20x10mm magnets to fit the core.
Quote from: Peterae on 2019.02.02, 17:46:04
Bit worried about the iron powder, when I picked it up from the post office I faced questions of what it was, I hope there's no illegal uses GULP,
Of course there is, but does that make you a criminal ?
There are criminal uses for dihydrogen monoxide, too...
BTW: If you mix iron powder with epoxy, its grains might become too far apart for some reactions and the epoxy itself will resonate, too....albeit at different frequency.
I've just been looking at the output of my dds and 1.5 Watt preamp, I tried scoping with no load and with a 50.11 Ohm load, although I am to believe this resistor has no inductance I can measure about 50nH on my meter.
Now I am not sure how valid this is with the present waveforms but I understand that I could find the output impedance of my preamp by subtracting the voltages when loaded and unloaded multiplying this by the load resistor and dividing by the loaded voltage.
The ffrequency was set to 45.250Mhz
So we have Zo = 50.11 * (23.6 - 9.99) / 9.99 = 68.26 Ohms
See scope shot 1 & 2
then I tried it with the 560 Ohm in series with a 0.1uF cap and connected the scope probe to the cap to try and isolate the probe.
see scope shot 3 & 4
So this time we have Zo = 50.11 * ( 6.48 - 3.67) / 3.67 = 38.36
Edit something strange going on, just going to recheck the scope readings again and will update.
OK the values seem to check out, amazed how I am getting such large pk-pk unloaded.
So I am just trying to get a feel for things for now.
I wound a coil which seemed to have a self resonant frequency of about 70Mhz in the hope that I can add a little capacitance to LC it at 45.25Mhz.
I bought some caps.
First test was to hook up my PI filter to the 50.11 Ohm load, see scope shot 1.
Second test was to hook my PI filter up and put the LC across the output, tune the pi and LC for maximum Amplitude
The piston cap made virtually no difference to the LC amplitude from fully wound in and fully wound out, value is 0.8pf to 10pf, see scope shot 2
Next i disconnected the Cap altogether leaving just the inductor see scope shope 3
Next I hooked a larger variable cap up across the inductor this time 10pf to 45pf this time I could definitely adjust the LC for a peak, see scope shot 5
I dont know how acurate this LCR meter is, certainly moves around a bit trying to measure nH inductance and just measured the tuned cap value in 5 and it measures 53.45pf (cap is 10-45pf but old stock) inductor measured 0.37uH
when I tap those LC values into a LC calculator I get 35.8Mhz plus I had my scope probe across the LC so something is out somewhere.
I think from the above tests I can add more turns to the inductor until I only need about 5-10pf piston cap for resonance.
Hi Peter,
First let me ask what kind of LC (or LCR) meter you have? I refer to the end of your first sentence where you wrote you measured 50 nH on your meter for your load resistor (in your reply #43).
At frequencies as high as 45.25 MHz a 10 pF capacitor (your probe probably has as a minimum?) has 351 Ohm reactance, this shunts your load and may greatly influence the result. What is also not good indication: the waveform changes as you change the measuring methods.
As a simple solution, try to use higher value resistors in place of the 560 Ohm to isolate better the probe. Say use 3 or 4 560 or even 1 or 2 kOhm metal film resistors in series, exact values are not important because the amplitudes the scope shows in such measurements are only comparative, need not be exact. Do not use a single resistor say 10 kOhm, just build it up from some series ones, to reduce their self capacitance. The individual values in the series resistor chain are not critical.
Will return to your other posts above tomorrow.
Gyula
The LCR meter is a TH2821a looks like the range is 0.1pf-999.9uF and 0.1uH-999.9H
So it looks like anything under 100nH is out of range, although capacitance should be ok, I must admit it's been ages since I looked the manual over but there is a series and parallel measurement mode and I have not been setting that and I just looked and it's been in series mode, maybe that should be parallel mode? I've put the link up for the manual, I also use it while plugged into the DC adapter as the battery has never been up to much, and the manual seems to indicate that sometimes this can interfere with accuracy, this was not a cheap meter but must be 5+ years old now, I will give the different modes a try next weekend.
https://d17bck4wpaw2mg.cloudfront.net/att/a/2/6/c/a26cq3/th2821a-manual.pdf
Maybe I should loosely couple the scope probe to the coil and then re do the unloaded voltage and loaded voltage and see it this works without actually the scope probe even being connected.
I just plugged my inductor dimensions into a calculator and it calculates out at 364nH which is quiet amazing as I measured 370nH.
It looks like my capacitance measurement is way off and I bet this is because I did it in series mode.
Hang on though I just realized the last variable cap of the pi filter is in parallel with the LC and appears to be set to maximum capacitance which if I remember was about 80pf although it could be higher, if that's correct I am running with 80pf+53.45 or 133.45pf now if I plug that into a LC calculator I get a resonant frequency of 24.2Mhz so how can I be resonating at 45.25Mhz.
Sorry so many things I find very confusing, give me logic any day :)
Maybe we need to keep things simple we could spend all year working these things through, I need a simple efficient way to tune the LC with my hot iron tube in place and get maximum energy transfer to the inductor.
I think I really need to build another PI network, use toroids for the inductance, use 1 pi network after my pre-amp feeding my 16 Watt amplifier and then the main PI filter feeding the LC coil, in fact because the last capacitor in the PI filter is across the final inductor, I would imagine I can do away with any parallel smaller value cap across the final inductor.
Hi Peter,
I agree to keep things simple. Working around 45 MHz is not easy as you find.
So to step forward, terminate the output of your Pi filter with a 22 Ohm, at least 1/2 W rated resistor. A normal metal film type is ok because its parasitics will be included in the Pi filter. The idea is to transform this 22 Ohm up to say around 60 Ohm as seen from the input of the Pi filter where the preamp output is connected, ok? So the preamp will drive the 22 Ohm via the Pi filter. There are two reasons for doing this, one is the needed tuning capacitor value across the filter output will be roughly in the 350-360 pF range (this helps utilize the gate-source capacitance of the MOSFET you indicated to use), and the other thing is the 22 Ohm is already a low enough value to keep input (gate-source) impedance of the MOSFET amplifier also low, to avoid self oscillations a linear or nearly linear amplifier is prone to. I know the transformed 60 Ohm is not 50 Ohm but the output of the preamp will equally be happy with it for sure.
See this link and use 60 Ohm for RL and 22 Ohm for RA, loaded Q=4 and frequency=45.5 MHz, you will get around 233 pF for C1 and 363 pF for C2. The coil would be about 78 nH, rather low, either use the formula you already used for the mechanical sizes or wait for the small RF toroidal cores.
http://people.physics.anu.edu.au/~dxt103/calculators/pi_tank.php
With this I also mean do not bother with testing what the output impedance of the preamp may be, it will certainly be good for it to "see" roughly the 60 Ohm transformed impedance. (Of course if you wish, you can still test it as I wrote above earlier.) And the input of the preamp will see the output impedance of the DDS what I assume has around 50 Ohm impedance, right? Do you know any spec on the DDS?
You have a nice LCR meter, thanks for the link to the manual. I think it measures one way only and the processor calculates from the measurement either the a series values or the equivalent parallel values, see page 21-22. (what it measures may also depend on the test frequency and on the lowest and the highest limits of the measurement ranges) For measuring 100 nH for instance I think the test frequency should be at 10 kHz. There is a calibration procedure to be performed as is written, page 20-21.
I understand your quest on driving a coil from the Pi filter and tune to about 45.5 MHz, this surely gives a feeling on such things. (The high peak to peak amplitudes come from the LC resonance as you know.)
When you have the iron powder in place with the coil around it and all this is embedded into the iron 'jaws', then you will see what inductance is going to come about and tuning method can be chosen only then, I think.
Gyula
Hi gyula
I must thank you for your patience and very important help ;)
I did power up the amp stage last weekend and it appears to work but didn't do much as the next step is to built another pi network to feed it, which I will do as per your above post, to make such large pf values up I will parallel a fixed value cap to each variable cap.
Now when i glanced over the dds board it does not appear to be 50 Ohm terminated, there appears to a multiple stage maybe pi network on the dds output and there is a unsoldered surface mount resistor with the resistor missing on the output of this network,so it would appear a resistor could be soldered across if needed, i do have 50 ohm smd resistor, the preamp seems to have a agc because when I adjust the output level on the DDS the output of the preamp is constant unless the level is virtually off.
The other bit of news is that there is a SMA input on the DDS board as well and i have a feeling this can be driven with an external signal to turn the output on and off,i will look into this at some point.
The cores have arrived, i could not find the sizes you stated on ebay, I ordered Amidon T37-2, i am now wondering if they would be too small, maybe not for the xtal and the preamp pi filter but maybe i should get bigger for the output pi filter stage that will feed the LC.
OK just looked at your Amidon calculator link, I should have looked at that previously, I now realize I have type 2 red, not the type 6 yellow, and red looks only good for 30Mhz, I will look again on ebay and re order.
OK just ordered a ciuple of Amidon T44-6 Yellow, these are in the UK so should be fairly fast
I added 2 more turns to the coil to see how it performed, it now only oscillates at twice 45.250Mhz and the tuning on the pi last cap is sharp to get the peak resonance, so it would appear I need to take a turn off and try again.
EDIT
OK that didnt work out, in the test above i was sniffing the signal with the scope probe earth connected to the coil only, it appears the scope is only able to sniff twice the resonant frequency, i cut 1 turn at a timne off the coil and sniffed and it would always pick up the double frequency, if i connected my 560 Ohm and cap and went across the coil i would pick up 45.250Mhz not twice the frequency, strangley the last cap in the pifilter always had to be fully engaged to get max pk-pk at 45.250Mhz, something strange going on, probably because my pi coil and first variable cap are incorrect, i will move to calculate these values and construct the final coil using calculators now.
In the picture attached with the pi filter the second cap is half adjusted and this was to get maxmium amplitude for the 90Mhz signal seen on the sniffer probe.
Sorry for my ignorance Peter, ultimately what are you trying to accomplish/create?
I am using a DDS to feed a 1.5 watt pre amplifier and this will then feed a home built 16 watt fet amplifier that will then feed a tuned LC.
I have a quartz glass tube filled with pure iron powder that will be heated above it's curie temperature, the tuned output inductor is wound over the quartz tube containing the hot iron, there is also a strong static magnetic field that aligns the irons atoms spin 90 degrees from the axis of the coil and tube.
I need to transfer as much power as possible into the iron powder at 45.25Mhz as possible.
I'm really waiting on a couple of cores at the moment to build the inductors in a couple of pi networks to impedance match the preamp output to the fet amp stage input and then another pi network to match the impedance of the output amp to the tuned LC.
because the scope probe has been loading the LC i thought i would try sniffing with the scope probe but this did not work.
In the above post the idea was to see what coil I could wind with the maximum turns I could use with the pi network to obtain the best resonance.
The device is a cross between Meyer/Mace and Gillespie/Coleman, Meyer used copper and iron and Coleman used Colbalt, both devices supposedly generate energy from the iron/copper/cobalt fuel by destabilizing the atom to release energy in the form of beta radiation, Meyers device was reported to generate 30 times the input power, I am heating the iron above it's curie temperature to help with skin depth and increase signal penetration.
I started out trying to build a xtal oscillator I bought on ebay which was 45.666Mhz and was hoping we could drop that down, but at the moment I have given up on that and reverted to my DDS generator.
I think if you use a 2 or 3 turn air core coil (OD say 2 cm) for sniffing and hook up such coil directly to the scope probe, then you could pick up a more stable sample from the coil to be 'sniffed' (mutual coupling). Move the sniffing coil close to the other coil and when you start seeing a waveform then adjust distance and tune the other coil circuit, in steps. Aim for as loose coupling as possible.
Consider that your series 560 Ohm may still not give enough isolation when you use it...
Gyula
What are you using for the 1.5W preamp?
A question for gyula...
Are RF transistors manufactured to be optimized for a certain frequency band? For eg. could a transistor that is supposedly optimized for 136MHz-174MHz operate just as well in a circuit designed for 45MHz?
I have access to a number of VHF (~ 155MHz) RF Power BJT transistors and could send them to Peter.
The pre amp was an ebay buy 20-500MHz 1.5 Watt
unfortunately I have no other information on it.
Here's a picture
I advise against buying amps that are not properly described, it usually means that the seller is not competent in the field and sells amps as he could sell lettuces or sofas.
This one (https://www.ebay.co.uk/itm/2-700M-3W-RF-FM-VHF-UHF-FM-Transmitter-Broadband-RF-Power-Amplifier-Shortwave/123515543735?hash=item1cc219a0b7:g:RwAAAOSw2XJbk5Yp:rk:9:pf:0) seems very acceptable, but perhaps not the same price.
Note that if this is not explicitly specified, in general the amp is not linear, i.e. the output power will not be exactly proportional to the input power. It's the case of this one in the link. It's only annoying for amplitude modulation, but here we work at a constant level.
Quote from: poynt99 on 2019.02.09, 03:31:01
A question for gyula...
Are RF transistors manufactured to be optimized for a certain frequency band? For eg. could a transistor that is supposedly optimized for 136MHz-174MHz operate just as well in a circuit designed for 45MHz?
I have access to a number of VHF (~ 155MHz) RF Power BJT transistors and could send them to Peter.
Well, my personal opinion is they can operate if the matching networks are redesigned or simply newly designed directly for the lower frequency (in this case for around 45 MHz). The gain of the transistor will be higher so the feedback network also needs special attention to tame it, avoiding unwanted oscillations.
Manufacturers specify their devices for certain frequency bands that suits for certain industry or military etc applications For those frequency bands the data sheet normally includes working conditions, input and output complex RF impedances, S parameters etc but outside that band the parameters are not always given, so one needs to extrapolate data.
The problem to be solved in Peter's case is to match an inductive load (a coil having iron core in powder form) to the output of a power amplifier at 45 MHz while the output power expected to be fed into the coil would need to be in the order of 15-20 W at least.
Gyula
Quote from: Peterae on 2019.02.09, 08:26:20
The pre amp was an ebay buy 20-500MHz 1.5 Watt
unfortunately I have no other information on it.
Here's a picture
Hi Peter,
By the look of the PCB, it has two stages, the input drives a low power wideband integrated circuit IC1 (like say a MAR-6 or ERA-6 for instance) and this drives the other amplifier, IC2, see attachment. These two integrated circuits are called MMIC (monolithic microwave integrated circuits) and usually they are designed for having nearly 50 to 70 Ohm input and output impedances for a specified wide frequency range. Data sheet for Mar-6 is here https://ww2.minicircuits.com/pdfs/MAR-6+.pdf (https://ww2.minicircuits.com/pdfs/MAR-6+.pdf) The other IC can also have various manufacturers, providing the 1.5W output across a 50 Ohm load, there are two SMD coupling capacitors from the IC pin towards the SMA power output pin. You may be able to see the label on the two MMIC chips by a magnifier, unless they are wiped away for good. IC2 has its own bias adjust circuit with the trimmer potmeter under IC2. IC1 drives IC2 via two series coupling capacitors.
The switch mode power supply circuit surely receives the DC input voltage range and provides a regulated supply voltage for the two amplifier ICs. In the picture below (taken from ebay) the IC type is LM2576HV if I see it correctly, data sheet: http://www.ti.com/lit/ds/symlink/lm2576.pdf (http://www.ti.com/lit/ds/symlink/lm2576.pdf)
This amplifier is still available at ebay:
https://www.ebay.com/itm/352486108062
What F6FLT is referring to is okay too for a good preamplifier (with 3W output) and here is the same amplifier for much cheaper if someone wishes to buy such:
https://www.ebay.com/itm/352486108062
Gyula
OK some updates.
The iron/resin core has been made, specs are 1.228kg 56% iron 44% Resin Approx. sprayed matt black, see picture, just printing the magnet holder.
Also the yellow 44-6 cores arrived, i put 4 turns on one and measured it, meter says 120nH calcs say 78nH, i made the new pi network, i had a 20Ohm load resistor and recalculated the values for the pi, i used 2 x 120pf variable caps, C1 was in parralel with a 220pf cap and C2 in parralel with a 330pf, it appears to give a good output across the 20 Ohm load at 5.8V RMS giving me a dissipated power of 1.682Watts.
One magnet is epoxied in (the easy bit) keeper is drying in foreground.
What do you make of this, I have mentioned this some time ago else where, it's also buried in one of Arie Degues patents.
https://www.electronicdesign.com/analog/resonant-circuit-generates-high-frequency-magnetic-field
It looks interesting Peter. Just hope the relatively high Q doesn't make it difficult to realize.
Quote from: Peterae on 2019.02.09, 18:37:51
What do you make of this, I have mentioned this some time ago else where, it's also buried in one of Arie Degues patents.
https://www.electronicdesign.com/analog/resonant-circuit-generates-high-frequency-magnetic-field
It is a good idea,
However, I run into a problem with it in pulsed NMR, because when the Tx LC network rings at the target frequency, then it is impossible to distinguish the LC oscillation from the NMR signal later, if an RF blanking is not used (e.g. a MOSFET which shorts or opens the LC circuit so it stops ringing immediately or a PIN diode which purposely detunes the LC circuit).
Also to tune the Tx LC network well, I had to sense the generated RF magnetic field, ...so I used another coil as a sensor.
Unfortunately, the sensor coil had its own self-resonances and a non-constant frequency characteristic, that confused my measurements and tuning process. The solution turned out to be a GMR head from a broken hard drive (the only difficulty was learning how to power its integrated head amplifier). Otherwise, it is a very good linear RF magnetic field sensor.
Quote from: Peterae on 2019.02.09, 22:03:09
Hi verpies
I don't believe I need to sense the NMR response, I will just scan around 45.525Mhz and try to detect beta radiation emission.
Ah, but you should sense the response.
This is because without sensing you will not know the width of the 90º pulse. The width of this pulse depends on the amplitude of the RF field penetrating the sample. The higher the amplitude the narrower the pulse.
If your pulse is too wide, then the spin axes will flip more than 90º, for example 110º ...or 180º ...or 360º which will amount to nothing.
If you want to scan blindly, you'd need to scan BOTH the frequency and the RF amplitude or RF burst width. That is a lot of work...and while you are at it, you are really blind to what's happening inside.
Hi Peter,
Glad you have achieved that nice sine wave across the 20 Ohm resistor, this will drive the linear MOSFET power amplifier. Of course, you will need to remove C2, 330 pF and its trimmer from the filter when you attach the Pi filter output (that includes the 20 Ohm resistor) to the MOSFET gate-source input. The trimmer cap across C1 (220 pF) can remain in place should any retuning due to the MOSFET is needed or not.
Regarding your coil calculation, the data for 0.281 uH from the formulas could likely be okay for the inductance but if you place the iron powder as the 'core' for this air core coil, then the inductance will very likely be increased, no? Or you started out with the 0.281 uH inductance value (which resonates with 22 pF at the needed higher frequency 63.996 MHz) that already has the iron powder core?
Here I assume you calculated this coil for doing the excitation for the iron powder.
Gyula
verpies
OK i see, interesting that Meyer and Coleman did not have a DC bias field or spin detector, but instead both used the conduction of the iron for power output, i wonder what happens to current flow in the iron when the atoms flip, what i am saying is that i wonder if the flip manifests itself linearly through the iron as a resistance change or a change in the current flowing through the iron.
gyula
Thanks, indeed we are ready to connect the amplifier input.
Ah i forgot about the iron, well i maybe able to compare the inductance measurement before and after inserting the iron to get an idea whats happening.
Interestingly the above article states that parralel LC is not best for circulating current,
'The coil current is generally small in parallel resonance'
and suggests that normally series would be used
'Up to now, the most practical and efficient way to drive high current through a magnetic coil is to use a series resonant circuit'
I think it would be good if I drop a low ohm resistor in series with the inductor and see if I can measure the current in the LC
Quote from: Peterae on 2019.02.10, 08:37:56
verpies
OK i see, interesting that Meyer and Coleman did not have a DC bias field or spin detector,
It is not surprising, because the magnetic domains in iron provide their own 33T DC bias field.
It helps to align the domains in one direction, but even without such forced domain alignment, the ferromagnetic remanence keeps their directions from being totally randomized and from cancelling out.
Quote from: Peterae on 2019.02.10, 08:37:56
what i am saying is that i wonder if the flip manifests itself linearly through the iron as a resistance change or a change in the current flowing through the iron.
There could be some relationship. I did not check it yet.
Quote from: Peterae on 2019.02.10, 08:37:56
I think it would be good if I drop a low ohm resistor in series with the inductor and see if I can measure the current in the LC
Instead if detuning the LC circuit with such measurement, use a GMR head from a broken hard drive to sense the magnitude of the MMF generated by the coil, which is directly proportional to the current flowing in it.
This is minimally invasive and reliable.
I got the second magnet mounted, I thought I was going to have problems due to the North / South pulling at each other, but it turned out the pull to the iron core was greater :)
Not yet sure what sort of magnetic field strength I have, I do have some Ratio metric Hall sensors but these only go to 900G, if anyone knows of one that goes up to the required 6000 gauss let me know please.
I did connect up the preamp, input pi filter, amp and tried a LC using calculations for 45.250Mhz, I placed 4*1 Ohm resistors in parallel to give 0.25 Ohm in series with the final inductor so I could monitor current. I did not manage to capture waveforms, by scoping across the final LC I did manage to get 191V pk-pk nice sinewave at 45.250Mhz but the fet overheated and burnt.
Things I need to do and solve, the little brown 120pf variable caps on my preamp pi filter stage fell apart, not sure if it melted but I need to find better variable caps or maybe switch to the 45pf ceramic caps I have, i need to wind the 2 inductors on the amplifier on toroids, i presume i will need bigger cores for this due to saturation currents.
I may need to add a pot to attenuate the drive signal to the amp.
The heatsink on my fet is way too small to dissipate the heat, i will rebuild next time with larger heatsink.
Hi Peter,
Please provide: what was the DC supply voltage to the MOSFET? What is the exact type of the MOSFET?
The MOSFET is supposed to operate in Class A mode (maybe Class AB) and an idle drain current should be adjusted first with the gate bias potmeter. It may be around not more than perhaps 50-100 mA DC when there is no any RF drive (of course the 20 Ohm should be there).
The coil in the drain circuit should be a choke type, maybe you can scavenge such from mains input of PC supplies or other appliances.
There should a kind of matching circuit between the drain output and the RF coil that would include the iron powder, especially if you wish to drive it with the series resonance method you referred to, to use the double current advantage via the coil. The impedance of such network will be in the some Ohm or less range, this network is almost a short circuit load for the drain output, this is why a matching circuit would be needed. Perhaps, just perhaps the output Pi filter could do this matching, this involves a second coil for the Pi filter. Probably other matching circuits could be devised.
The smaller type trimmer caps do not 'bare' too much RF current or voltage, unfortunately.
Gyula
Hi Gyula
Yep I have some chokes somewhere.
The fet as per the circuit IRF610 which has a VDS of 200V, I wonder if my drain was seeing the pk-pk voltage i was seeing across the LC causing it to blow.
The circuit stated 50V supply at 0.5A, i did try 50V and adjusted the pot to bring the current up, i cannot remember the settings when it blew, dont get too hung up, i was jumping the gun somewhat, i built this amp board a couple of years ago when i was following a similar path, i will rebuild it from the things i have learnt so far, bigger heatsink and better quality caps.
Now what i really need is a setup process, i think i am going to need to set the input amplitude and i think a small pot will not be high enough wattage, so i will probably use 2 fixed resistors to form a divider chain using metal film resistors I can always change the values to get the drive I need, where is the best place for the resistor divider, does it go straight after the pi filter but before the 10nf cap feeding the fet gate or after the 10nf cap feeding the fet gate.?
Also to prove i have perfect drive i need a way to power the amp up and scope the drain, preferably without a pi or load to start, will it damage the fet if i have nothing connected to the drain or should i connect a resistor as a load maybe.?
Once i know i have good fet drive i then connect the output pi filter and drive the 50Ohm dummy load resistor
Thanks
Hi Peter,
Indeed the 50V supply voltage must have been too high as a start for the amplifier if there was no any previous matching / tuning attempt at a lower supply voltage.
Regarding a setup process, here is a possibility. This text is going to be long, sorry, I think it is needed.
See a modified input circuit for the amplifier attached, I added a resistive divider ("a resistor Pi") ahead of the input Pi filter. The 470 Ohm potmeter can be any normal one, not a wire wound type, to be able to reduce the > 1 W RF power from the output of your preamp to 0.2-0.3 W level, not to overdrive the linear amplifier.
Next, remove the MOSFET, put a 360 pF across the 22 Ohm resistor and drive the input of the resistive diivider from the preamp. Adjust C1 of the Pi filter for max sine wave amplitude around 45 MHz, measure across the 22 Ohm resistor with the scope probe. Check by turning the 470 Ohm potmeter between what minimum and maximum range it can vary the RF voltage amplitude across the 22 Ohm. Leave the wiper at say 2-3 V peak to peak across the 22 Ohm, if this low amplitude is not possible (i.e. at the full 470 Ohm wiper position the voltage is still higher than that), then place say a 100 or 150 Ohm resistor in series with the potmeter and repeat this adjustment.
Now remove the preamplifier, i.e. no RF drive, and also remove the 360 pF cap which so far substituted the gate-source capacitance and put the MOSFET back to the circuit. Also, remove the choke from the drain and put a 50 Ohm dummy load instead, capable of the expected 15-20 W dissipation. It is ok you wish to see and check the signal at the drain but you have to have a load there too for feeding the MOSFET with the supply voltage.
You may wish to start with much lower than 50 V like say 10-12 V DC input only. First the DC operating point is to be adjusted with the Set Bias potmeter: you need to measure the DC drain current and set it first say to 200 mA only at the 10-12V supply voltage and no any RF drive yet. The wiper of the bias potmeter should be set in advance to start from about 2V because the minimum threshold voltage for this MOSFET is 2V. But your MOSFET may have say 3.2V threshold voltage from where drain current can start flowing at all, so slowly turn the bias pot up from the 2 V and watch the drain current rising to about 200 mA.
Now if you increase the supply voltage towards 40 or 50 V, still monitor the DC drain current, it will increase above the previous 200 mA of course (without touching the already set bias potmeter) and the heat dissipation at 50 V will increase beyond 10 - 12 W, a big heat sink is needed for the MOSFET of course. Naturally the same drain current will dissipate heat in the dummy load too, it should be able to dissipate it.
I suggest to run the MOSFET at this operating point for at least several minutes to check temperature and drain current stability (with still no RF drive input yet).
If all seem ok, then increase drain current up to 400-450 mA, supply voltage is at 50 V and see how heat sink temperature increases, hopefully a small ventilator would not be needed to cool it. Run the setup for several minutes at least.
Then, if you think, apply the smallest RF drive the 470 Ohm pot let through (as previously set above) and check the RF amplitude across the dummy load or across the drain and the negative rail, ideally they have the same voltage across them. Then you could increase RF drive by turning the 470 Ohm pot and monitor the drain current and the RF amplitude at the drain or the dummy load.
It is possible you need to retune a little the C1 cap in the input Pi filter when the supply voltage for the MOSFET is in the 40-50 V range because the gate-source capacitance changes to a lower value and this may affect input Pi filter matching.
It is possible the RF input drive to the MOSFET (depending on where the 470 Ohm pot is set) will increase drain current beyond the DC current previously set by the bias pot. Well, a small overdrive is not yet a problem but you can see this as a starting clipping (limiting) at the peaks of the drain-source voltage wave form. It is also possible you adjust DC drain current to say 600 mA to increase output power, this would allow an increased input drive too to achieve higher output, only the MOSFET dissipation is the limit.
If all seems ok, then so far the 0.5 A drain current at 50V supply voltage will provide roughly 12.5 W RF power in the 50 Ohm dummy load (the other 12.5 W is dissipated in the MOSFET). Such is the case for an ideal Class A power amplifier.
Next step is to agree on how you try to match the coil with the iron powder core to the drain of the MOSFET. The Pi filter at the output first sounds good but your coil will represent either a much higher than 50 Ohm impedance load when tuned to be a paralell resonant LC tank at 45 MHz or will represent a much lower than 50 Ohm load when tuned to be a series resonant LC circuit, so the output Pi filter should be thought over.
Gyula
OMG Gyula!, you have a really good understanding of RF amplifiers!
Gyula
Thank you very much for the time this has taken you, fantastic and will keep me on track with a good chance of succeeding. ;)
I have been thinking about how to carry forward, before i complete the amp, i need to study Parralel LC, series LC and the C-LC article above, monitor current flowing in the L component and try to work out the impedance of all 3 methods , then add the iron tube and heat to 600 Deg C and try again the 3 methods above.
Lots to do but it should be interesting.
I've been trying out the L-LC tuned circuit from that article, driving from my 1.5 Watt pre amp through the pi network, feeding a capacitor and then feeding the parallel LC, I used the circuit from figure 11 and 1 ohm resistors to monitor current, when i tune the LC current for max level and then disconnect and use a GDO to drive it loosely coupled, max current appears at about 33Mhz which is indeed 0.7071 of C-LC driven circuit at 45.250Mhz, this appears to prove the math of the article.
https://www.accelinstruments.com/Applications/WaveformAmp/Magnetic-Field-Generator.html
I am have trouble with current measurements across my 1 Ohm resistor see waveform at 4.95V RMS, which would mean i am dissipating 4.95 Watts, which is pretty clever seen as i am driving it with 1.6Watts, and as the resistor is not getting really hot (31 Deg C in 15 Deg C ambiant) i must assume these resistors do indeed have inductance.
So i will try next a film resistor to see if that cures my problem, until i get an accurate voltage i cannot work out the current or the magnetic field strength, i will say that there is a strong magnetic field around the coil because it's the first time i can easily sniff 45.250Mhz sine around the area with my scope probe with a couple of turns (35V pk-pk) or even with the earth clip on the probe end.
The impedance of the C-LC circuit is 4R or 4Ohm in the above case
Quote from: Peterae on 2019.02.15, 19:20:47
...
I am have trouble with current measurements across my 1 Ohm resistor see waveform at 4.95V RMS, which would mean i am dissipating 4.95 Watts, which is pretty clever seen as i am driving it with 1.6Watts, and as the resistor is not getting really hot (31 Deg C in 15 Deg C ambiant) i must assume these resistors do indeed have inductance.
...
It is only recently that I realized how difficult it is to measure high frequency signals with a scope.
The end of the probes is not coaxial, it acts as an antenna or is capacitively coupled. Then the ground wire creates a loop, in which we have induced currents.
The probes must be eliminated and the scope connected directly to the resistance by a coaxial cable designed for high frequencies (very well shielded).
In addition, the cable length must be small in comparison to the shortest wavelength of the signal, otherwise there are phase shifts or resonant line effects, because the input impedance of the scope is unrelated to the output impedance of the setup.
Certainly would explain the problems i am having, i am now thinking that the easiest way to do power measurements is to measure the heat produced by the resistance load. :(
With a series circuit composed of a germanium or schottky diode and a capacitor, in parallel on the resistance, the peak HF voltage can be measured as a DC across the capacitor, provided that the signal exceeds the diode threshold (about 0.3v, to be added to the measured voltage). It's a basic method but I find it much more accurate than the scope.
I found a pro's build of my amplifier for sale, it's interesting seeing what he has done.
https://www.ebay.co.uk/itm/151274234687
From today, I tried a parallel LC, for some reason I can only sniff 2nd harmonic with a coil.
I tried C-LC and sniffing this is comparable in amplitude to the sniffing I did with series LC.
I think for ease, Series LC is the way to go, but how to deal with a very low impedance that the amp will see, maybe a series resistor could be used, we know the current the inductor see's is common for the resistor, capacitor and inductor.
So as follows
Inductor is 12 Turns of 1.5mm wire wound to a length of 50mm inductance calculates to 559nH
for series resonance with a 559nH inductor I need a capacitance value of 22.13pF
if we use a 5 Ohm Series resistor and we manage a 30pk-pk drive voltage then we get
Xl = 2pifl = 158.93Ohms Inductive Reactance.
Xc= 1/2pifc = 158.93Ohms Capacitive Reactance.
Circuit Impedance Z = sqr(R2+(Xl-Xc)2) = 5 Ohm R is effectively the load resistance.
I = Vs / Z = 30/5 = 6 amps
Vr = I * R = 30 Volts
Vl = I * Xl = 953.58 Volts
Vc = I * Xc = 953.58 Volts
Magnetic field flux = 18.0956 Gauss
Quote from: Peterae on 2019.02.16, 18:06:52
....
I think for ease, Series LC is the way to go, but how to deal with a very low impedance that the amp will see, maybe a series resistor could be used, we know the current the inductor see's is common for the resistor, capacitor and inductor.
...
Hi Peter,
No need for using a series resistor if you meant it for easing the matching of the very low impedance, it can be transformed up by a matching network.
I attached the schema again with the addition of a possible matching network to transform a small (2-3 Ohm at resonance, indicated in the blue block) real impedance to the drain output circuit of the MOSFET. It is called an L matching network and needs a coil (L) and a capacitor (C) only. The C capacitor should include the drain-source output capacitance of the MOSFET plus a trimmer to make up for the needed value. For the coil a toroidal winding would be the best with ample core cross section but an air core one could also be used. See this link on calculating the L and C values: https://home.sandiego.edu/~ekim/e194rfs01/jwmatcher/matcher2.html (https://home.sandiego.edu/~ekim/e194rfs01/jwmatcher/matcher2.html)
Source resistance 50 Ohm (this would be the up-transformed impedance the drain circuit will see as a load)
Load resistance 2 Ohm (this is either your "current amplified resonant" or a simple series LC resonant circuit)
For the Source and Load Reactances use zero
For the Desired Q use say 5
For Frequency use 45.25e6
The L and C values you need will appear under the
LOWPASS Hi-Low MATCHING NETWORK circuit schematic (first row on the right), I received L=34.46 nH and C=344.6 pF.
You can use a normal series LC resonant circuit within the blue block too, for simplicity, if you wish, instead of the current amplified resonant circuit.
The tuning procedure (to use first a normal series resonant LC as the load for the power amplifier output) would be advisable like this (fully separate the coil mentioned below from the circuit) :
1) fill in the iron powder for the coil assigned for exciting it. (you have already that plastic coil holder for this right?)
2) try to measure the inductance of this coil with the L meter. perhaps first without the iron powder, then with it.
3) then try to find a capacitor which when connected in parallel with this coil (that filled with the iron powder) gives a parallel resonance around 45 MHz, checked with your grid dip meter. IF the capacitor needed for this comes to be a very small value like under 10 pF, then reduce the number of turns of the coil. Beware: if the iron powder can move inside the coil, as you handle the coil holder with your hand, its inductance will surely change!
4) aim for a capacitor of at least 15-20 pF to give resonance around 45 MHz with the coil.
5) when done, you now have a simple series resonant LC circuit if you connect the capacitor in series with the coil, giving a very low value real impedance (any value like 1-3 Ohm) which will be stepped up towards 50 Ohm by the LC matching network.
Before you build and switch on this total circuit, I advise to go through the procedure first with the 50 Ohm dummy load connected into the drain circuit as I wrote in the previous post.
When that is done and seems ok, there may come placing the choke coil into the drain with the matching LC circuit as calculated from the link and the L coil is wound and the C cap value is considered with the output cap of the MOSFET (C
DSS=55 pF at 25 V supply), and also the pretuned output series resonant circuit could be connected. Start with 12 V again first, only then raise the supply voltage higher and higher to 50 V. The main tuning may involve the trimmer cap adjustment across the drain-source for voltage maximum indicated by the scope probe with a sniffer coil. The L coil may also be tuned by pulling away the turns or pushing the turns closer slowly and carefully, it can greatly influence the up-transformed impedance towards the drain, hence the output power too. Notice if you increase this coil value, then the C cap would need to be decreased to keep the network at resonance (but the loaded Q is around 5 only, so bandwidth will surely be wide).
Gyula
ADDITION: here is shorter link to the ebay offer, https://www.ebay.co.uk/itm/151274234687 (https://www.ebay.co.uk/itm/151274234687) you may wish to use it in your above post, so the horizontal width of the page will be restored to normal (for those who do not use wide screen).
By the way, the
same FM Linear Amplifier is available as a DIY Kit here:
https://dutchrfshop.nl/en/diy-kits-pcb-s/589-diy-kit-15-watt-amplifier-87-108mhz-rd01mus-rd15hvf1.html (https://dutchrfshop.nl/en/diy-kits-pcb-s/589-diy-kit-15-watt-amplifier-87-108mhz-rd01mus-rd15hvf1.html)
The schematics, the Bill of Materials are also included to see. All the L and C values used for matching and in the filters are valid for the FM band which is roughly twice as high as the 45 MHz needed here. Also, the output impedance is surely matched to a 50 Ohm load, as usual.
The kit does not include the coils but the wire for winding them... :D By mentioning this kit, I do not hint to buy it... you decide. :)
Hi Gyula
Thats a nice amp kit, very tempting, i will give the home build a go first ;)
I'm going to need some high voltage caps, looking at 10Kv 22pf and adjust the inductor length for fine tune.
I have enough to start the build now.
thankyou very much for your help, very much appreciated, you are an incredible engineer. ;)
Just ordered a radiation detector https://www.ebay.co.uk/itm/143006383989
and 2 of these caps
https://www.ebay.co.uk/itm/273236302508
I am still working on this in the background, still plenty to do.
I made the framework for the fuel rod out of 3d printed hollow sections and poured resin into it, this is to make it more durable from the harsh conditions it will operate under and elevated temperatures which plastic alone is not good at handling.
The fuel rod was made of 2 round printed formers filled with 1200 Deg C compound to glue the quartz tube in place, each end has a copper wire coiled inside the tube to make electrical contact with the iron powder filling.
At first the iron did not conduct very well over 10 MOhm, but after heating with a blow torch it now measure about 1 Ohm and heats nicely with a dc supply connected, I will need chokes at each end to isolate the circuit from the 45MHz oscillations.
2 pictures below have the magnet assembly fitted as well.
I forgot to add the 45mhz coil onto the quartz tube before gluing the end caps on, OOPS the coil is small diameter than the end caps, so i will need to devise a way to wind the coil over the quartz tube without breaking anything, i have some kilm paper which i will wrapp around the quartz tube to insulate it and stop the heat affecting the Inductor coil.
Hi Peter,
Nice progress and build of the magnet assembly. Is the quartz tube fully packed with iron powder filling? (so that the powder should not move freely within the tube?)
Well, on the 45 MHz coil winding: probably only a few turns is needed to have the max some hundred nH inductance, so making that around the quartz tube may not cause much problem versus say making 20 turns.
The iron powder will certainly increase inductance and this is why I think of a few turns (4-5 maximum) only. This seems to be a limitation because the process or effect may perhaps happen more readily when the number of turns is higher excitation_wise, while a higher number of turns gives higher inductance at 45 MHz which then involves a higher inductive impedance hence less current for excitation, unless the excitation power is increased and increased. But the supply voltage to the 45 MHz power amplifier can be increased if needed, maybe at a price of using a better MOSFET.
Gyula
Hi Gyula
Thanks
Yes the iron powder is packed in the tube, I first glued one end and let that dry, I then heated the quartz tube slightly with a blow torch to get any possible moisture out and once cooled slightly I place the wire in the end and then added a little more iron powder to burry the electrode and then packed glue into the quartz tube end, once that was dry I added the printed end shell and packed that with high temp glue.
I believe it maybe possible to use 2 fets in parallel to double the power, I will see how many turns the inductor requires with the iron in place, although the objective will be to heat the iron above it's curie point after initial tests
Peter
Quote from: gyula on 2019.03.09, 20:58:08
Well, on the 45 MHz coil winding: probably only a few turns is needed to have the max some hundred nH inductance,
...yes and you really should use fine Litz wire for all HF windings.
Quote from: gyula on 2019.02.14, 17:11:19
Hi Peter,
Indeed the 50V supply voltage must have been too high as a start for the amplifier if there was no any previous matching / tuning attempt at a lower supply voltage.
Regarding a setup process, here is a possibility. This text is going to be long, sorry, I think it is needed.
See a modified input circuit for the amplifier attached, I added a resistive divider ("a resistor Pi") ahead of the input Pi filter. The 470 Ohm potmeter can be any normal one, not a wire wound type, to be able to reduce the > 1 W RF power from the output of your preamp to 0.2-0.3 W level, not to overdrive the linear amplifier.
Next, remove the MOSFET, put a 360 pF across the 22 Ohm resistor and drive the input of the resistive diivider from the preamp. Adjust C1 of the Pi filter for max sine wave amplitude around 45 MHz, measure across the 22 Ohm resistor with the scope probe. Check by turning the 470 Ohm potmeter between what minimum and maximum range it can vary the RF voltage amplitude across the 22 Ohm. Leave the wiper at say 2-3 V peak to peak across the 22 Ohm, if this low amplitude is not possible (i.e. at the full 470 Ohm wiper position the voltage is still higher than that), then place say a 100 or 150 Ohm resistor in series with the potmeter and repeat this adjustment.
Now remove the preamplifier, i.e. no RF drive, and also remove the 360 pF cap which so far substituted the gate-source capacitance and put the MOSFET back to the circuit. Also, remove the choke from the drain and put a 50 Ohm dummy load instead, capable of the expected 15-20 W dissipation. It is ok you wish to see and check the signal at the drain but you have to have a load there too for feeding the MOSFET with the supply voltage.
You may wish to start with much lower than 50 V like say 10-12 V DC input only. First the DC operating point is to be adjusted with the Set Bias potmeter: you need to measure the DC drain current and set it first say to 200 mA only at the 10-12V supply voltage and no any RF drive yet. The wiper of the bias potmeter should be set in advance to start from about 2V because the minimum threshold voltage for this MOSFET is 2V. But your MOSFET may have say 3.2V threshold voltage from where drain current can start flowing at all, so slowly turn the bias pot up from the 2 V and watch the drain current rising to about 200 mA.
Now if you increase the supply voltage towards 40 or 50 V, still monitor the DC drain current, it will increase above the previous 200 mA of course (without touching the already set bias potmeter) and the heat dissipation at 50 V will increase beyond 10 - 12 W, a big heat sink is needed for the MOSFET of course. Naturally the same drain current will dissipate heat in the dummy load too, it should be able to dissipate it.
I suggest to run the MOSFET at this operating point for at least several minutes to check temperature and drain current stability (with still no RF drive input yet).
If all seem ok, then increase drain current up to 400-450 mA, supply voltage is at 50 V and see how heat sink temperature increases, hopefully a small ventilator would not be needed to cool it. Run the setup for several minutes at least.
Then, if you think, apply the smallest RF drive the 470 Ohm pot let through (as previously set above) and check the RF amplitude across the dummy load or across the drain and the negative rail, ideally they have the same voltage across them. Then you could increase RF drive by turning the 470 Ohm pot and monitor the drain current and the RF amplitude at the drain or the dummy load.
It is possible you need to retune a little the C1 cap in the input Pi filter when the supply voltage for the MOSFET is in the 40-50 V range because the gate-source capacitance changes to a lower value and this may affect input Pi filter matching.
It is possible the RF input drive to the MOSFET (depending on where the 470 Ohm pot is set) will increase drain current beyond the DC current previously set by the bias pot. Well, a small overdrive is not yet a problem but you can see this as a starting clipping (limiting) at the peaks of the drain-source voltage wave form. It is also possible you adjust DC drain current to say 600 mA to increase output power, this would allow an increased input drive too to achieve higher output, only the MOSFET dissipation is the limit.
If all seems ok, then so far the 0.5 A drain current at 50V supply voltage will provide roughly 12.5 W RF power in the 50 Ohm dummy load (the other 12.5 W is dissipated in the MOSFET). Such is the case for an ideal Class A power amplifier.
Next step is to agree on how you try to match the coil with the iron powder core to the drain of the MOSFET. The Pi filter at the output first sounds good but your coil will represent either a much higher than 50 Ohm impedance load when tuned to be a paralell resonant LC tank at 45 MHz or will represent a much lower than 50 Ohm load when tuned to be a series resonant LC circuit, so the output Pi filter should be thought over.
Gyula
So starting the amp again and going through Gyula's post, I am up to adding the 360pf cap in place of the fet, see shots below, waveform at the fet gate position is varying up to 3.8v pk-pk, currently set at 2v pk-pk.
I'm using blue/yellow toroids size 55, for the output inductor i am wondering if this will be big enough, is there any problem stacking 2 cores and winding the stacked cores as one core?
Placing my hand on the ground plane of the circuit board makes the sine look clean, instead to get the below scope shot I had to alter the first cap of the PI network to a lower value, it's in parallel with a 0-75pf variable cap and moved to a 100pf instead of a 220pf, infact because in the picture the variable cap was open, I have now moved even lower to 47pf.
if I hold the ground plane then the calculated value 220pf+var cap works and gives a clean sine.
Anyway time to move on and add the bias circuit and fet with load.
Hi Peter,
Normally there is no problem with the stacking of toroid cores and it is okay as you described their 'how to' winding, but then the inductance may increase too high for even 2 turns only at 45 MHz and you would need to retune the filter of course and even change a trimmer cap if it goes out of range. But you know this, you nicely described how you went about it.
You wrote toroid size 55, did you mean 44 instead? Sizes are 30, 37, 44, 50, 68 etc.
Placing your hand on the ground plane of the PCB: well, hard to tell what the body's capacity picks up at such a high frequency which then goes to the ground. Perhaps the ground clip of the scope probe influences this too, try to clip it to other ground area, I am not sure.
Gyula
OK finally got the 50 Ohm load resistor as the last one went missing.
Quote
Now remove the preamplifier, i.e. no RF drive, and also remove the 360 pF cap which so far substituted the gate-source capacitance and put the MOSFET back to the circuit. Also, remove the choke from the drain and put a 50 Ohm dummy load instead, capable of the expected 15-20 W dissipation. It is ok you wish to see and check the signal at the drain but you have to have a load there too for feeding the MOSFET with the supply voltage.
You may wish to start with much lower than 50 V like say 10-12 V DC input only. First the DC operating point is to be adjusted with the Set Bias potmeter: you need to measure the DC drain current and set it first say to 200 mA only at the 10-12V supply voltage and no any RF drive yet. The wiper of the bias potmeter should be set in advance to start from about 2V because the minimum threshold voltage for this MOSFET is 2V. But your MOSFET may have say 3.2V threshold voltage from where drain current can start flowing at all, so slowly turn the bias pot up from the 2 V and watch the drain current rising to about 200 mA.
Now if you increase the supply voltage towards 40 or 50 V, still monitor the DC drain current, it will increase above the previous 200 mA of course (without touching the already set bias potmeter) and the heat dissipation at 50 V will increase beyond 10 - 12 W, a big heat sink is needed for the MOSFET of course. Naturally the same drain current will dissipate heat in the dummy load too, it should be able to dissipate it.
I suggest to run the MOSFET at this operating point for at least several minutes to check temperature and drain current stability (with still no RF drive input yet).
If all seem ok, then increase drain current up to 400-450 mA, supply voltage is at 50 V and see how heat sink temperature increases, hopefully a small ventilator would not be needed to cool it. Run the setup for several minutes at least.
I scorred a super large heatsink with holes drilled and heat washers with screws from a discarded power supply.
So i am running at about 520mA @ 50V without any input for over half hour,520mA as this seems to give me an equal voltage across load resistance and fet, the heatsink sits at 44Deg C, 23Deg C above ambiant, the front of the fet is at 65Deg C and the resistor is only about a degree above the heatsink temperature.
I need a longer sma lead to reach the board from the DDS now :(
Good progress so far.
Good progress Peter, and following with interest.
Hi Peter,
You have a nice Class-A amplifier now with the resistor in the drain, so dissipation in the MOSFET should not go higher than 25-26W when driven by. And when you have an AC impedance transformed by the matching circuit back to the drain to a similar value to that of the present resistor, a similar dissipation could be expected, except that choke coil DC resistance will be much less than the present load resistance of 48 Ohm, the drain current will go up by some ten mA (drain voltage versus drain current characteristc curves are not horizontal but slightly rising), so you will have to readjust the bias potmeter to have the 510-520 mA drain current then too.
Good progress.
Gyula
OK got everything bolted down and wired 3 psu's, 1 for 5v bias supply, 1 for DDS & CPU control, and 1 for 50V supply, these are all buck converters fed from 24V.
I am wondering if the operating frequency of these is affecting operation, unfortunately I ran out of time to run many tests.
but here's a scope shot with 50Ohm load and inductor in the 50V supply feed.
The yellow trace is ground to drain, the cyan is ground to inductor side of 50Ohm load.
Edit I need to look at the inductor i placed in the supply, it is heavy guage inductor, i did measure it before but cannot remember the value but it was over 100uH, maybe i'm operating above it's SRF.
I did try adjusting the PI filter but this did not seem to clean up the sine, i suppose it could be my scope probes giving false signals.
Ah Maybe I should not have the supply inductor in place while testing with a 50 ohm resistor in the drain?
Quote from: Peterae on 2019.05.04, 20:10:18
....
Ah Maybe I should not have the supply inductor in place while testing with a 50 ohm resistor in the drain?
Okay, just noticed your above addition and yes, that is what I asked when was preparing an answer:
Hi Peter,
What would you like to test with putting the 50 Ohm load in series with the choke (inductor) in the supply feed?
If you wanted to do the next step after the undriven DC 'soak' test you showed in Reply #88, then place either the choke alone in the drain or the 50 Ohm alone in the drain but not both. AND if you decide say to use the choke in the drain to feed in the 50 VDC, then you would need a 50 Ohm load too for the drain because the choke has many kOHm inductive impedance at 45 MHz. The 50 Ohm load could be connected between the drain and ground via a DC blocking capacitor to imitate the output load, to be transformed later by the output Pi filter when the fuel road etc assembly is connected.
The 'unclear' waveforms may come from the huge near field the choke coil radiates around itself hence sprays on quasi any nearby things.
Gyula
Thanks Gyula
I have now removed the inductor from the drain, i just wanted to see the sinewave on the drain so i am scoping across my drain to ground and 50v supply to ground, yellow trace is now on 50v and cyan on drain side of 50 ohm load, this is so i can do a math subtraction to see whats across my 50 ohm resistor, see below, purple is the math trace A-B.
first thing i dont get enough drive and could do with more, in the scope shot i fully turn the amplitude pot up to max.
There does seem to be a low frequency component the 45.5mhz sine rides on, i am not sure if thats a problem, seen as we are driving a resonant lc i imagine it should not be an issue.
I also have some ripple from the DC-DC converter on the 50V supply, not sure if thats an issue, maybe i could add more capacitance on the supply, maybe this ripple is causing the low frequency component.
I scope shot i zoomed out on the timebase to capture the low frequency noise.
When i comes to the drain inductor i will use 2 of my cores stacked to constrain the mag field. ;)
Next I need to see if it's possible to heat my iron tube to above the currie temperature and control that point by monitoring it's resistance, ie a resistance controlled temperature control of some description.
Hi Peter,
Try to check the ripple riding on the 50 V DC voltage by using AC coupling temporarily for CH1 yellow channel, this way you can set the 10 V/DIV to a more sensitive range and see the AC amplitude better (though this not really needed because its amplitude 3.6 V is small wrt 50 V).
By the look of it, it also seems a 45.5 MHz waveform riding on the 50 VDC rail. Very likely you can reduce its amplitude by using the choke coil (you used yesterday) inserting it into the 50 V wire coming from the DC-DC converter and is connected to the top end of the 50 Ohm load.
And use a filter capacitor to the negative ground from the common point of this choke and the top end of the 50 Ohm. The choke and this capacitor will form a low pass filter inserted into the positive 50 V rail. You can use another filter cap to the ground from the other leg of the choke coil which receives the direct 50 V from the DC-DC converter. These filter caps could be some hundred nF poli or ceramic type (goal is low loss at 45 MHz), if you do not happen to have such, then use some uF electrolytic type, 63 VDC.
I assume when you turned the amplitude pot to max drive, you checked the settings of the Pi filter trimmer caps too whether they need some retuning?
Another possibility to have more drive is to omit the 62 Ohm from the left side of the 470 Ohm amplitude pot because now that the pot practically has near zero resistance, the two 62 Ohms are directly in parallel and may cause too much load i.e. loss in the input signal. I refer to the schematic attached to reply #80 above, I assume you use that.
Gyula
Thanks
Yes tried retuning the pi input filter, the sinewave get a little noisier like yesterday with the choke in place but not as bad.
I will try removing the 1st resistor left of the input pot, and try a filter on the 50v line as suggested. ;)
Tried the CLC 50V filter, the only thing that makes a little difference is moving the scope probe away from the 50R load resistor and placing directly on the filter, but it's still there,the inductor measure 57uH so I was wrong with my previous guess, strangely the DC-DC converters current meter reads zero at 45.5Mhz and the sinewave seems to have a little noise on it, I tried 40Mhz and 50Mhz and these are nice and clean, maybe there's iron nearby that's interfering :)
I did remove the left resistor of the pot and that has increased my drive amplitude, it could still do with a little more, but I am not going to get hung up on this as it is only 5V each side of the sinewave, it maybe because I lowered the PI cap value during tuning, I might try increasing that cap back up again.
I am very happy with it and it sits there quite cool running.
I am running at 55V.
I have some scope shots 40Mhz, 45.5Mhz & 50Mhz.
and a couple of pictures.
Thanks for all your help Gyula ;) It's been fun, theres still a lot to do yet.
Hi Peter,
Okay, good progress again. Yes, the scope probe placement surely has its 'sweet spot' to place, even small distances may influence the waveforms (also the length of the probe ground wire to the croco clip) and there may be other inherent yet unknown issues, these are always hard to rectify when high level, high frequency is present. I do not think either that you would need to bother with further clean up of the 50 (or 55) VDC supply.
When you have the L matching network (L, C in the schema above) connected to the drain of the MOSFET, driving the 'current amplified resonant circuit' (Cs, Cp and the coil with the iron powder in it), then the power amplifier output will have more selectivity and things may change again... so even more fun is guaranteed. :) O0
Gyula
Hey Peter,
Out of curiousity, what are you going to be investagating with your frequencies in this range? Does it have anything to do with the magnetic precessional rates of 42.4923 MHz that Ken Wheeler suggested? I noticed you put a little smiley emoji when it was suggested that Iron was distorting your signal..
Thanks,
Dave
Thanks Gyula ;)
Next I need to heat up the iron until an insulated magnet drops of to indicate we reached our Currie temperature, make note of the voltage and current at this point, work out it's resistance and then build a circuit that on/off controls that temperature every second by locking onto the known resistance.
Hi Dave
It's a sort of version of a Michel Meyer device, we also know Coleman guilespie did similar using cobalt metal.
Some time ago we discovered a paper which said the NMR frequency for iron is different where the NMR frequency is stable at 45.5Mhz in varying magnetic field strengths, which gives us the chance to align the iron atoms spin and resonate it at it's NMR frequency, with the hope of destabilizing the atom and releasing some energy in the form of beta radiation and dropping to a lighter isotope.
With iron we have a penetration problem due to skin depth, I hope to alleviate this by passing a current through the pure iron powder to heat it above it's Currie point.
Quote from: Peterae on 2019.05.07, 18:47:29
Some time ago we discovered a paper which said the NMR frequency for iron is different where the NMR frequency is stable at 45.5Mhz in varying magnetic field strengths, which gives us the chance to align the iron atoms spin and resonate it at it's NMR frequency, with the hope of destabilizing the atom and releasing some energy in the form of beta radiation and dropping to a lighter isotope.
With iron we have a penetration problem due to skin depth, I hope to alleviate this by passing a current through the pure iron powder to heat it above it's Currie point.
Many references state that
"The absolute frequency of iron is 3.237778 MHz with respect to 100.00 MHz for TMS".
That refers to the NMR frequency of
individual iron nuclei subjected to an external magnetic flux of such density that it causes the single protons of the Tetramethylsilane to precess at 100MHz. This happens at the magnetic flux density of 2.34 Tesla.
At a lower flux density, e.g. at 0.5T, the
individual Iron nuclei resonate at 690kHz.
The dependence of the NMR frequency on the magnetic field is better captured by the Gyromagnetic Ratio which precisely relates how the NMR frequency of
individual nuclei varies in response to the density of the externally applied magnetic flux.
For
individual iron nuclei, this Gyromagnetic Ratio is 1381.56Hz/mT. You can see the values for other metals here (https://www.overunityresearch.com/index.php?action=dlattach;topic=4525.0;attach=49514).
However, for the ferromagnetic bulk metallic iron, the effective Gyromagnetic Ratio and resulting NMR frequencies are
wildly different because of its huge negative internal hyperfine field which affects the iron nuclei in addition to the externally applied magnetic field.
The authors of this paper (https://www.overunityresearch.com/index.php?action=dlattach;topic=4653.0;attach=51014) experimentally measured metallic Iron's NMR frequency as 45.525MHz. ( half of it is 22.763MHz ) in the absence of an external magnetic field (and in presence of its -33.02T internal hyperfine field).
Unlike non-ferromagnetic compounds, the metallic Iron's nuclear resonance frequency is relatively independent of external magnetic fields because the internal -33.02 Tesla hyperfine field swamps any externally applied fields.
An external magnetic field >0.75T saturates the bulk metallic Iron (i.e. coalesces and orients all its magnetic domains in one direction) and the flux density of this field does not need to be precisely controlled/correlated with the nuclear resonance frequency (unlike with non-ferromagnetic materials) because the internal -33T hyperfine field of Iron affects its resonance frequency much more than any external field, which we mere mortals could apply.
For example, the application of a 1T external magnetic field decreases the total magnetic field to which the Iron nuclei are subjected to, to -32.7T
* which decreases the Iron's nuclear resonance frequency by only ~350kHz, and for external fields well below the Iron's saturation level (< 0.6T) that frequency changes negligibly (by only -0.033%). See the hollow squares graph line of
f vs.
BEXT below:
(https://www.overunityresearch.com/index.php?action=dlattach;topic=3107.0;attach=19130)
It is important to remember that even when an external magnetic field is not applied, increasing the temperature of the Iron metal significantly decreases its nuclear resonance frequency, so if the oscillator does not track the temperature then periodic cool-downs are required.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4525.0;attach=51374)
The Curie temperature (TC) of pure iron is 1040°K* The ferromagnetic magnetization (domain rotation and coalescence) takes 0.75T to happen. After that, the remaining external flux density directly subtracts from the internal hyperfine field.
Quote from: Verpies on 2026.03.25, 06:38:26
...
Unlike non-ferromagnetic compounds, the metallic Iron's nuclear resonance frequency is relatively independent of external magnetic fields because the internal -33.02 Tesla hyperfine field swamps any externally applied fields.
An external magnetic field >0.75T saturates the bulk metallic Iron (i.e. coalesces and orients all its magnetic domains in one direction) and the flux density of this field does not need to be precisely controlled/correlated with the nuclear resonance
...
That's very interesting to know. A lower dependence of the frequency on the external field, combined with a strong static internal field, should make NMR spectroscopy of ferromagnetic materials much easier. Yet I don't think that's the case. So where's the catch? I guess it lies in the extreme dispersion of the internal fields, which broaden the frequency range, and in the difficulty of allowing the variable external field to penetrate, particularly at high frequencies.
Quote from: F6FLT on 2026.03.27, 21:35:56
That's very interesting to know. A lower dependence of the frequency on the external field, combined with a strong static internal field, should make NMR spectroscopy of ferromagnetic materials much easier. Yet I don't think that's the case. So where's the catch? I guess it lies in the extreme dispersion of the internal fields, which broaden the frequency range, and in the difficulty of allowing the variable external field to penetrate, particularly at high frequencies.
Actually the broadening is minimal in pure annealed iron but it increases with contamination.
(https://www.overunityresearch.com/index.php?action=dlattach;topic=4525.0;attach=49528)
The NMR spectrum of pure annealed iron (solid square)
and natural iron (open circle) at room temperature.
Both of natural isotopic proportions.Isotopically pure
57Fe exhibits further narrowing and 1852x higher NMR peak amplitude than natural iron (the natural abundance of
57Fe is 2.1%).
As you have rightly noticed, the skin depth / penetration of alternating magnetic fields into bulk metallic iron is very small and this constitutes the major problem.
This can be somewhat mitigated by finely dividing the metal (powders, thin films) and in iron-rich ferrites of negligible conductivity.
High temperature is another way because it lowers the permeability and conductivity of metallic iron but it also lowers its Larmor frequency.
The ultimate (100%) bulk metal penetration is achieved by ultrasonic stimulation, also known as nuclear acoustic resonance (NAR) but ultrasonic transducers which are able to handle 45.5MHz are not ubiquitous and hard to drive.
Some iron-rich ferrites exhibit magnetostriction and low conductivity. Such ferrites are especially susceptible to NAR and NMR but driving the windings wound over them at such high frequencies poses an engineering problem.