Marx generator power supply protection

[MRMILSTAR]

I built a 20-stage Marx generator. The power supply is a flyback transformer driven by a ZVS driver. The ZVS driver is powered by two 12-volt batteries in series for a total of 24 volts. The Marx generator works as expected with 6 inch sparks, or greater if I want to push it. The problem I am having is how to protect the HV power supply, in particular the ZVS driver. The ZVS driver is one of those Chinese $10 units which uses a standard circuit. The problem I have had is that some of the HV discharge kicks back into the flyback secondary. This then appears to induce a voltage into the flyback primary (center-tapped coil with dual 4-turn coils). The induced voltage then kicks back into the MOSFETs of the ZVS driver. Sometimes this voltage exceeds the source-to-drain voltage rating (250 volts) of the MOSFETs and blows them out.

One idea that I had was to connect varistors across the ZVS driver output. One varistor across one 4-turn coil of the flyback primary. Another varistor across the other 4-turn coil of the flyback primary. The 3rd varistor would then connect across the outer turns of the two coils. I know that varistors have a limited lifetime but the Marx generator only fires about every 3 seconds so I think the varistors would last a reasonable amount of time. They are inexpensive and easy to change.

If this seems like a viable approach, what peak voltage rating for the varistors should I use? If this is not a viable idea, does anyone have any other method for protecting the HV power supply?

[T3sl4co1l]

What transistors does it use?

Tim

[MRMILSTAR]

What transistors does it use?

Tim

IRFP250 MOSFETs

[T3sl4co1l]

I would guess, use a ~140V TVS or ~100V MOV, tied as close as possible (shortest wiring distance) to source-drain, and also a 15V TVS from source to gate.

Tim

[coppercone2]

I have a HV pulse circuit thats setup to be triggered with IR that I bought from some national defense lab or something. It discharges a special tube through a transformer, but I noticed something strange about it, the HV supply was connected to the rest of the circuit by a very large choke, I think like 5Henry. I could not figure out for sure what it was, but it isolated the capacitor from the solid state DC supply. I think it might be protection. The supply is too weak to to require current limiting to the capacitor ESR.

[MRMILSTAR]

Since it is the easiest, my initial attempt at protection will be to add several large ferrite cores which I have to the HV lead coming from the flyback transformer. If that doesn't work I will proceed to the MOVs.

[T3sl4co1l]

Ferrite beads are actually rather worthless here: the impedance per core is maybe a hundred ohms, so even with a hundred of them stacked (or tens of turns, but that has the problem of capacitance between turns), a high voltage pulse of, say, 10kV or more, still draws amperes.

And that's if it's a linear system, which it isn't: a ferrite might handle, say, 100µVs of flux, or with tens of turns, thousands of µVs; but at 10kV, that's saturated in a mere 100ns.  You can take some of the edge off, but you can't absorb it this way.

(Flux is just what its units sound like: the voltage across a component, multiplied by time.  If the voltage is constant, flux goes up linearly with time.  If voltage isn't constant, then more generally, it's the integral over time.  Basically, while voltage is positive, flux is counting up; while voltage is negative, flux is counting down.)
(Saturation means the ferrite momentarily loses its inductive property, so the impedance drops and it looks more like a short circuit, or air-cored coil.  Saturation isn't stateful, as soon as the current goes down the impedance goes back up.)

Tim

[MRMILSTAR]

I wasn't so much trying to absorb the kickback pulse, just slow down its rise time. A slower rise time should induce less kickback voltage in the flyback between the primary and secondary.