High voltage PSU Mosfet choice

[MarcusFunt]

Hi!
 
I am building a device for a national competition ("Unge forskere") That should be able to sterilize hands with nothing but electricity.

For this project, I need a High-Voltage power supply, capable of delivering 12(or so)kV at 30-100khz.
Therefore my head automatically went to the very popular "ZVS driver" because I want a very efficient operation without much heat as it will be placed in a closed box.
You can look at my current schematic in the attached picture.
I modeled the schematic in lTSPICE and found that 71v spikes occur at startup, and therefore can't use the 60v IRFZ44N I originally wanted to use.
Now the problem is that I don't know what other MOSFET I can use.
I don't care if it is SMD or THT.
I found a few FETs I think will work, but I'm not certain
https://www.mouser.dk/ProductDetail/onsemi/NTP011N15MC?qs=DPoM0jnrROX1k1aAlpufTw%3D
https://www.mouser.dk/ProductDetail/onsemi-Fairchild/FDP085N10A-F102?qs=0lQeLiL1qyaFktU8QMby%252BQ%3D%3D
https://www.mouser.dk/ProductDetail/Infineon-IR/IRLR3110ZTRPBF?qs=Z8%252BeY1k3TILcI4snfu078g%3D%3D

Any help, both with MOSFET choice and flaws in my design are greatly appreciated

[MarcusFunt]

My input voltage will be 12vdc from a commercial smps. but the voltage across the MOSFET is dependent on the resonant circuit formed by the primary coil and resonant capacitor. the resonant voltages and current are calculated by LTSPICE: 17A peak. 42vac peak. I will at max be drawing a few milliamps from the transformer.
The MOSFETs will however only pass around 10A max.
The gates will be getting 12v protected by Zener's and flyback diodes(probably not necessary)
the inductance of my primary is going to be around 3-6microheny(very vague I know)

[davekni]

y input voltage will be 12vdc from a commercial smps. but the voltage across the MOSFET is dependent on the resonant circuit formed by the primary coil and resonant capacitor. the resonant voltages and current are calculated by LTSPICE: 17A peak. 42vac peak.

42V sounds reasonable.  With a simple model of perfect switching, peak voltage would be PI * DC_input_voltage, 38V in your case.

BTW, great to see more use of analog simulation!  Especially with ZVS oscillators, many are puzzling about startup issues.  Simulation helps clarify such.

I modeled the schematic in lTSPICE and found that 71v spikes occur at startup, and therefore can't use the 60v IRFZ44N I originally wanted to use.

Do you have the option to turn on 12V supply to gate resistors (460 ohm in your schematic) before power to inductors?  That generally reduces startup spike, perhaps enough to use IRFZ44N.
460 ohms seems high for 12V input.  Is that copied from higher-voltage units?  At 12V, lower resistance will provide enough current for fast FET switching.  I'd try ~100 ohms.  Will still be under 1W each during operation.

If you center-tap the transformer primary, the two 50uH inductors can be replaced with one smaller inductor from center-tap to 12V supply.

[MarcusFunt]

Hi

If you center-tap the transformer primary, the two 50uH inductors can be replaced with one smaller inductor from center-tap to 12V supply.

I know, but it is simpler to measure and troubleshoot.

Do you have the option to turn on the 12V supply to gate resistors (460 ohms in your schematic) before power to inductors?

Yes, but it would have to be done automatically and quickly. Could an RC Time circuit and MOSFET get the job done?

460 ohms seems high for 12V input.  Is that copied from higher-voltage units?  At 12V, lower resistance will provide enough current for fast FET switching.  I'd try ~100 ohms.  Will still be under 1W each during operation.

I have used 460 ohms before, all the way from 16 to 36 volts input. I will probably change it to a lower value. 470 does seem like a lot.

Thanks

[davekni]

Could an RC Time circuit and MOSFET get the job done?

Probably could work.  Try it in simulation.

Might be useful to look at my ZVS plasma ball supply and the issue I had of transformer secondary winding capacitance forming dual-resonant system.  ZVS tended to lock to upper pole, with little output voltage.
https://highvoltageforum.net/index.php?topic=924.msg6187#msg6187

There's also a very long thread about ZVS oscillators that includes useful information about startup issues:
https://highvoltageforum.net/index.php?topic=530.msg3247#msg3247

[MarcusFunt]

Could an RC Time circuit and MOSFET get the job done?

Probably could work.  Try it in simulation.

I designed a NE555-based circuit that eliminates all the startup problems
Green=before
Blue=after

Something weird I noticed though. Why is the voltage of the blue waveform so much higher?

[davekni]

I designed a NE555-based circuit that eliminates all the startup problems

Sounds interesting.

Something weird I noticed though. Why is the voltage of the blue waveform so much higher?

No way to know without seeing circuit and a zoomed-in view of final higher-voltage oscillation plot.

[MarcusFunt]

I designed a NE555-based circuit that eliminates all the startup problems

Sounds interesting.

Something weird I noticed though. Why is the voltage of the blue waveform so much higher?

No way to know without seeing circuit and a zoomed-in view of final higher-voltage oscillation plot.

I thought the rise at the end was weird so I ran it again and got the exact same result but without the bump. It was just some weird anomaly. I am designing the final schematic tomorrow and will post it here. It would be really nice if you would take a look at it, and tell me if I did something weird/stupid .

[Twospoons]

Any time you are simulating resonant circuits make sure to include  ESR and winding resistance. Otherwise your resonant tanks can end up with ridiculously high and unrealistic Q.

[MarcusFunt]

I am done with the first version of my schematic. It would be great if someone would go through it and tell me if I did something wrong

[davekni]

A few recommendations:

ICs generally need bypass capacitors from power to ground.

Switching IC power directly from another IC output is problematic once bypass capacitors are added.  No need for such.  The reset pin is sufficient for your logic needs.

Not sure if there are any convenient NE555 models for LTSpice.  However, I suggest simulating at least the power switch (Q3) using a stepped voltage source to simulate output of U5.  Look at NE555 specifications to see the output Voh level, typically about 1.3V below VCC.  Use that voltage for stepped voltage source.

BTW, is there any reason for the ZVS oscillator to remain on after fan turns off?  I'm just wondering, depending on how your device is intended to be used.

[MarcusFunt]

A few recommendations:

ICs generally need bypass capacitors from power to ground.

Switching IC power directly from another IC output is problematic once bypass capacitors are added.  No need for such.  The reset pin is sufficient for your logic needs.

Not sure if there are any convenient NE555 models for LTSpice.  However, I suggest simulating at least the power switch (Q3) using a stepped voltage source to simulate output of U5.  Look at NE555 specifications to see the output Voh level, typically about 1.3V below VCC.  Use that voltage for stepped voltage source.

BTW, is there any reason for the ZVS oscillator to remain on after fan turns off?  I'm just wondering, depending on how your device is intended to be used.

I simulated both the entire zvs circuit and the fan controller in LTSpice, and everything worked as it should. the 555 can output 200ma, so it should be more than enough to drive both the other 555 and the MOSFET. I am considering adding a push-pull amp to Q3 though or maybe a dedicated driver ic and signaling into that via the 555.

[MarcusFunt]

BTW, is there any reason for the ZVS oscillator to remain on after fan turns off?  I'm just wondering, depending on how your device is intended to be used.

Yes. The fan should stay off for a few seconds after the zvs turns on, then stay on for a few seconds after the zvs are turned off(I don't the best way to do that is with 2 555's. If you know a better way to do it, Please educate me ) it is to allow the ozone to build up a bit first before it sends it through a filter

[davekni]

I am considering adding a push-pull amp to Q3 though or maybe a dedicated driver ic and signaling into that via the 555.

Yes, that was my concern, if output voltage (source) of Q3 was high enough when on.   A pull-up resistor on Q3 gate (U5 output) would work as well as a driver chip unless the driver chip can pull Q3 gate above 12V.  There are boot-strap etc. options if you need the full 12V on Q3 source (above 12V on Q3 gate).  Or, use a PFET for Q3 and a low-true signal to enable.

I simulated both the entire zvs circuit and the fan controller in LTSpice, and everything worked as it should.

Any need for bypass capacitors won't show up in simulation unless you also simulate wiring (ECB trace) inductance in power and ground connections, and have a 555 model that accurately simulates power supply current spikes.  You may be OK without bypass capacitors.  I think all 555 timer chips available today have much lower power pin current spikes during switching compared to the original 555 parts of 50+ years ago.

the 555 can output 200ma, so it should be more than enough to drive both the other 555 and the MOSFET.

Driving 555 supply pin won't be an issue unless the 555 being driven has a supply bypass capacitor.

Yes. The fan should stay off for a few seconds after the zvs turns on, then stay on for a few seconds after the zvs are turned off(I don't the best way to do that is with 2 555's.

If simulation shows it working as intended, then should be fine.  Looked to me like your circuit would turn fan on 1.5 seconds after ZVS, then turn fan off 3.5 seconds later, even if ZVS is still running.  That presumes J4 is on for more than 5 seconds.  If J4 switch is on for less than 5 seconds total, then it looks like fan turns off along with ZVS when J4 switch is turned off.  However, fan might stay on.  I'm not sure exactly how 555 timers will behave when power is removed and there is still charge on their timing capacitors (C2 and C3).  555 outputs may have residual voltage fed from timing caps.  Or, 555 output may be floating, leaving charge on Q4 gate due to gate capacitance, leaving fan on.