Tetra Replication

Author Topic: Tetra Replication (Read 16356 times)

Topic Options

bte

Tetra Replication « on: 2024-04-11, 08:21:14 »

Group: Moderator
Sr. Member

Posts: 308

Placeholder. Current breadboard setup.

Mosfets on breadboard unstable above 50 to 75 volts. Running higher causes gate instability. Bought isolated driver boards. Rc snubbers increase load on hv supply too much. Output coil picks up ringing from pulses, dampens quickly. More to follow.

------------------------

Tetra Replication

IMG_20240410_104916768.jpg (2928.94 kB, 4160x3120 - viewed 532 times.)

bte	Re: Tetra Replication « Reply #1 on: 2024-04-11, 09:23:54 »
Group: Moderator Sr. Member Posts: 308	Schematic ------------------------ Tetra Replication IMG_20240411_042210778.jpg (1920.36 kB, 4160x3120 - viewed 437 times.)

lfarrand

Re: Tetra Replication « Reply #2 on: 2024-04-11, 09:24:55 »

Sr. Member

Posts: 280

I think you will probably need to break free from the breadboard if you want to achieve good results. There is plenty of stray inductance and capacitance in a breadboard setup. Every wire will be acting as a small antenna.

I designed and built a MOSFET driver board PCB with galvanically isolated gate driver. I paid special attention to the MOSFET gate to reduce stray inductance by using a copper pour instead of traces.

I've attached the Gerber files that you can provide to a PCB manufacturer to have them build some for you. I've also attached the KiCad 8 files so you can customize it if you so wish.

I use 4-pin TO-247-4L MOSFETs with the Kelvin connection pin. You could easily modify the footprint to use 3-pin MOSFETs instead.

Here's a photo showing me testing the first revision 4 board.

I made some changes to decrease the size from revision 3 to revision 4.

------------------------

KiCad.zip (195.31 kB - downloaded 46 times.)

MOSFET_IGBT_Driver_Board_4_Gerber.zip (148.12 kB - downloaded 48 times.)

bte	Re: Tetra Replication « Reply #3 on: 2024-04-11, 10:05:20 »
Group: Moderator Sr. Member Posts: 308	Thank you for your feedback. Yes, I do believe any serious attempt will need isolation. I bought 4 isolated driver boards. I just need to wire the rcd snubber up to it. Otherwise, the voltage slowly rises back up to the rail voltage over a couple uS, which is pretty bad. I'll try with the driver boards tonight! I also think I'll need a bigger output coil. I only have a few turns on it atm.

bte	Re: Tetra Replication « Reply #4 on: 2024-04-11, 15:22:20 »
Group: Moderator Sr. Member Posts: 308	I think the noise I see on the control circuitry is just the probe picking up emi. I tested with a snubber on my isolated drivers and had no problems so far. I need to sort out my 300V supply issues and wire up the drivers to the breadboard.

bte	Re: Tetra Replication « Reply #5 on: 2024-04-12, 15:37:15 »
Group: Moderator Sr. Member Posts: 308	This lmg3522r050 looks like it would fit the bill, integrated and everything. A bit complex to route though. https://www.ti.com/lit/ds/symlink/lmg3522r050.pdf?ts=1712930747939&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FLMG3522R050

bte	Re: Tetra Replication « Reply #6 on: 2024-04-12, 16:56:11 »
Group: Moderator Sr. Member Posts: 308	I think I found my candidate (>10Mhz switching): https://gansystems.com/wp-content/uploads/2020/04/GS66502B-DS-Rev-200402.pdf Very fast. If these can't get an output out of the tetra, I don't know what can.

bte	Re: Tetra Replication « Reply #7 on: 2024-04-16, 07:01:33 »
Group: Moderator Sr. Member Posts: 308	I wonder if it really needs 60mhz. That's nearly impossible

sergh

Re: Tetra Replication « Reply #8 on: 2024-04-16, 11:04:32 »

Jr. Member

Posts: 81

ARF463AG RF MOSFET

100 MHz
500 V
100 W
TO-247

https://eu.mouser.com/ProductDetail/Microchip-Technology/ARF463AG?qs=6jAvgNDGUA97NlfR32pAKA%3D%3D

These or similar transistors are used in MRI machines to generate radio frequency probe pulses.
If you have access to a medical equipment repair service scrapyard, you can find these or similar.

lfarrand	Re: Tetra Replication « Reply #9 on: 2024-04-16, 11:15:05 »
Sr. Member Posts: 280	Anything beyond the low-MHz frequencies is going to be very difficult to tame. As frequencies increase then PCB traces and wires start to act as antennas for harmonics of your operating frequency and you have all sorts of problems with stray inductance and capacitance. Unless you have RF expertise then I'd say avoid this if you can!

sergh	Re: Tetra Replication « Reply #10 on: 2024-04-16, 13:15:10 »
Jr. Member Posts: 81	Up to 30 - 50 MHz there will be no big problems if you make the wires as short as possible and use capacitors and chokes for the power supply. A metal case is also necessary.

bte	Re: Tetra Replication « Reply #11 on: 2024-04-16, 13:49:56 »
Group: Moderator Sr. Member Posts: 308	The issue is the mosfets overheating potentially. I have here a schematic for the drive circuitry. Isolated, driving a GaNFET rated for >10Mhz hard switching. ------------------------ Tetra Replication ckt.png (40.84 kB, 1227x695 - viewed 301 times.)

bte	Re: Tetra Replication « Reply #12 on: 2024-04-16, 13:52:37 »
Group: Moderator Sr. Member Posts: 308	Wow! That's a heavy duty FET. Do you think it could hard switch at Mhz frequencies? Quote from: sergh on 2024-04-16, 11:04:32 ARF463AG RF MOSFET 100 MHz 500 V 100 W TO-247 https://eu.mouser.com/ProductDetail/Microchip-Technology/ARF463AG?qs=6jAvgNDGUA97NlfR32pAKA%3D%3D These or similar transistors are used in MRI machines to generate radio frequency probe pulses. If you have access to a medical equipment repair service scrapyard, you can find these or similar.

lfarrand	Re: Tetra Replication « Reply #13 on: 2024-04-16, 19:15:14 »
Sr. Member Posts: 280	Some attributes of a MOSFET to look for are: 1. Gate charge - this should be low so that it can be charged and discharged quickly. The gate is a capacitor, so t=RC applies. 2. Rise & fall times - these should be slow so that the FET spends less time in the region inbetween switched off and switched on. The time spent in this region results in heat due to resistance. 3. RDS(on) - this should be low (milliohms) in order to reduce the loss due to I^2*R losses.

sergh	Re: Tetra Replication « Reply #14 on: 2024-04-17, 09:45:20 »
Jr. Member Posts: 81	There's a problem here. You can find many mosfets on the market that are specifically designed for RF. If you use conventional power mosfets, even gallium nitride, you will not be able to achieve high efficiency at frequencies above 20 MHz and higher. High efficiency means efficiency of more than 0.5 - 0.6 https://aliexpress.com/wholesale?SearchText=rf+mosfet https://aliexpress.com/wholesale?SearchText=ldmos https://aliexpress.com/wholesale?SearchText=MRF186 https://aliexpress.com/wholesale?SearchText=BLF188XR If you are planning experiments in the range of 20 - 50 MHz, I do not recommend buying too high-frequency transistors. at a frequency of 900 - 1000 - 2000 MHz, they are prone to self-oscillation at microwave frequencies. There are a lot of non-original RF transistors on AliExpress, so it is better to buy used transistors with solder residues on the terminals. Plan that the first experiments will lead to damage to these transistors. Take a close look at high frequency circuitry and design. There are many features, failure to comply with which will lead to damage to transistors or emission of interference and fines from regulatory authorities.

sergh	Re: Tetra Replication « Reply #15 on: 2024-04-17, 09:57:52 »
Jr. Member Posts: 81	Quote from: bte on 2024-04-16, 13:52:37 ARF463AG RF MOSFET Wow! That's a heavy duty FET. Do you think it could hard switch at Mhz frequencies? This type of mosfet is supposedly used in some older MRI machines. You can look for other similar transistors of this type, their type starts with ARF4 In MRI machines, radio frequency mosfets are used in an powerfull RF amplifier, the output of which is connected to an inductive resonant load. Accordingly, switching occurs at zero current. This is most likely closer to class D. You can explore about MRI and the circuit design of a radio frequency amplifier for probing pulses. In addition to this, MRI also has other coils, gradient coils, which operate at very low frequencies and very high currents. And there is also a very powerful electromagnet based on superconductors, which works continuously.

bte	Re: Tetra Replication « Reply #16 on: 2024-04-17, 11:35:54 »
Group: Moderator Sr. Member Posts: 308	I'm going to round robin the mosfets so any one mosfets won't have to pulse too fast. 5x round robin means repetition rate divided by 5.

sergh	Re: Tetra Replication « Reply #17 on: 2024-04-17, 12:39:19 »
Jr. Member Posts: 81	Power mosfets are almost always the same in terms of high frequencies. The most optimal ones seem to be IRF510S. S - reduced gate capacitance. I experimented with conventional IRFs without s, in the construction of a radio frequency power amplifier for frequencies of 27 - 50 MHz. It works, but it works with low efficiency. I did not like. Works much better with specialized RF mosfets LDMOS. https://en.wikipedia.org/wiki/LDMOS https://frenning.dk/OZ1PIF_HOMEPAGE/50MHz_IRF510.htm This is a good circuit based on conventional power mosfet, you can use fewer transistors, and of course the power will not be so high. Please note that at the output of the amplifier there is a resonant coil and a capacitor. This amplifier only operates at a fixed frequency unless the output resonant circuit is rebuilt. This amplifier is not wideband. From a pair of ordinary cheap IRF510 (without -S) mosfets you will get 15 - 20 watts 50 MHz of output. The power required was 80 watts. The input needs to be 2-3 watts. Efficiency is very low, less than 20%

bte	Re: Tetra Replication « Reply #18 on: 2024-04-17, 12:51:56 »
Group: Moderator Sr. Member Posts: 308	I don't think running them in the linear region is possible if I just want to switch to saturation and off again? Is there a way to run in linear region and get a 0-300v pulse? Quote from: sergh link=topic=4642.msg111709#msg111709 date=1713353959 Power mosfets are almost always the same in terms of high frequencies. The most optimal ones seem to be IRF510S. S - reduced gate capacitance. I experimented with conventional IRFs without s, in the construction of a radio frequency power amplifier for frequencies of 27 - 50 MHz. It works, but it works with low efficiency. I did not like. Works much better with specialized RF mosfets LDMOS. https://en.wikipedia.org/wiki/LDMOS https://frenning.dk/OZ1PIF_HOMEPAGE/50MHz_IRF510.htm This is a good circuit based on conventional power mosfet, you can use fewer transistors, and of course the power will not be so high. Please note that at the output of the amplifier there is a resonant coil and a capacitor. This amplifier only operates at a fixed frequency unless the output resonant circuit is rebuilt. This amplifier is not wideband. From a pair of ordinary cheap IRF510 (without -S) mosfets you will get 15 - 20 watts 50 MHz of output. The power required was 80 watts. The input needs to be 2-3 watts. Efficiency is very low, less than 20%

sergh	Re: Tetra Replication « Reply #19 on: 2024-04-17, 15:26:04 »
Jr. Member Posts: 81	Quote from: bte on 2024-04-17, 12:51:56 I don't think running them in the linear region no, this is another. Class D with resonance For sample: https://www.google.com/search?q=A+compact+Class+D+RF+power+amplifier+for+mobile+nuclear+magnetic+resonance+systems

bte	Re: Tetra Replication « Reply #20 on: 2024-04-17, 15:34:07 »
Group: Moderator Sr. Member Posts: 308	I think I have a theory. I watched a video where Peterae pulsed a bifilar. He got pulses and RF noise at around 220ns. Spherics said that the COMP field isn't affected until the release of one pulse is affected by the next pulse. ie. if we have ~5Mhz top coil, then that's around 200ns, the device should have a frequency of around 5Mhz with no magnetic field coil if I'm correct! Quote "Only at high megahertz does the pattern caused by release combine with the next pulse to effect the COMP field."

bte	Re: Tetra Replication « Reply #21 on: 2024-04-17, 17:06:47 »
Group: Moderator Sr. Member Posts: 308	Not sure of the target frequency for a resonant effect, but looking into it. Quote from: sergh on 2024-04-17, 15:26:04 no, this is another. Class D with resonance For sample: https://www.google.com/search?q=A+compact+Class+D+RF+power+amplifier+for+mobile+nuclear+magnetic+resonance+systems

bte	Re: Tetra Replication « Reply #22 on: 2024-04-17, 18:52:53 »
Group: Moderator Sr. Member Posts: 308	Quote from: sergh on 2024-04-17, 09:45:20 There's a problem here. You can find many mosfets on the market that are specifically designed for RF. If you use conventional power mosfets, even gallium nitride, you will not be able to achieve high efficiency at frequencies above 20 MHz and higher. High efficiency means efficiency of more than 0.5 - 0.6 https://aliexpress.com/wholesale?SearchText=rf+mosfet https://aliexpress.com/wholesale?SearchText=ldmos https://aliexpress.com/wholesale?SearchText=MRF186 https://aliexpress.com/wholesale?SearchText=BLF188XR If you are planning experiments in the range of 20 - 50 MHz, I do not recommend buying too high-frequency transistors. at a frequency of 900 - 1000 - 2000 MHz, they are prone to self-oscillation at microwave frequencies. There are a lot of non-original RF transistors on AliExpress, so it is better to buy used transistors with solder residues on the terminals. Plan that the first experiments will lead to damage to these transistors. Take a close look at high frequency circuitry and design. There are many features, failure to comply with which will lead to damage to transistors or emission of interference and fines from regulatory authorities. I am considering a more traditional RF amplifier approach, would really simplify component count but pcb design needs to be rock solid, ground plane, impedance matching... I can make logic pulses into the amplifier, then output of amplifier will be through a cap, to the coil which should be the same result, a sharp impulse. ~10ns rise, then immediate fall. I can operate it in the linear region. A negative pulse to the coil should be the same effect as a positive pulse so I'm thinking this will work.

sergh	Re: Tetra Replication « Reply #23 on: 2024-04-18, 07:13:13 »
Jr. Member Posts: 81	At 50 megahertz, there are 2 rises and 2 falls per cycle. The cycle time is 20 nanoseconds at 50 MHz. As a result, we get a rise time of 5 nanoseconds. What impulses do you consider powerful? Pulses with a rise time of 10 nanoseconds and the same fall time, with an amplitude of 1000 volts into a 50 ohm load. This is 20 ampere current. Okay, the pulse power is 20 kilowatts, the average pulse duration is 10 nanoseconds. The energy of one pulse will be 0.0002 Joule. At a repetition rate of 10 kilohertz per second, the load will ideally release 2 joules. It's not that much.

bte	Re: Tetra Replication « Reply #24 on: 2024-04-18, 07:50:48 »
Group: Moderator Sr. Member Posts: 308	Quote from: sergh on 2024-04-18, 07:13:13 At 50 megahertz, there are 2 rises and 2 falls per cycle. The cycle time is 20 nanoseconds at 50 MHz. As a result, we get a rise time of 5 nanoseconds. What impulses do you consider powerful? Pulses with a rise time of 10 nanoseconds and the same fall time, with an amplitude of 1000 volts into a 50 ohm load. This is 20 ampere current. Okay, the pulse power is 20 kilowatts, the average pulse duration is 10 nanoseconds. The energy of one pulse will be 0.0002 Joule. At a repetition rate of 10 kilohertz per second, the load will ideally release 2 joules. It's not that much. Load is a 176mH coil, 144 ohms resistance. No idea what the impedance is at whatever frequency is needed. Pulse power is 1-2 watts max, very little per pulse because of the inductor. All voltage no current. The coil will begin to sink too much current if I don't keep the duty cycle super low. 50ns pulse at 10Mhz means 50% duty cycle, which means 150V equivalent across the coil which means it overheats. So if I want to target 5Mhz, double it to 10Mhz for headroom. 10ns pulse is adequate I think. Means 10% duty cycle at 10Mhz, 30V avg across 144 ohms should be fine. AWG 32 wire for coil, max current 0.09A, not much. All 4 coils in my setup are matched together as best as I could. ~175mH 145 ohm. Once I find the frequency, then I can design a dedicated circuit with one frequency required, minimal components hopefully! Thanks for all your help.