Generating HT

Mister, you ain't no engineer. That much is clear from your other contributions.

Is there even such a thing as a snubber /diode/? And if there were, would you connect it between source and drain fof the /FET/?

... Phil

incorrect

NT

Either way would zap the bad guys. Most stun guns seem to end with a transformer.

I would have thought it would make more sense to feed the CW multiplier from a tranny, though. Doing it the other way around leads to a higher risk of flash-over within the tranny, no? All those windings crammed in so close together and whatnot dontcha know.

and it acts like a zener

-Lasse

That's the way colour TV EHT was normally done. But a Cockcroft Walton isn't ideal for generating a quick pulse of EHT.

NT

.... Phil

By limiting the reverse voltage level? (the "avalanche-rated" feature of some power MOSFETs). Just a guess...

According to this article, there's also a parasitic BJT in parallel with it!

You're right about that, John. I just got one for free. Works so well for the purpose I've had to use a 50ohm series resistor to limit its output energy. Thanks for a great suggestion!

Limit its output? That's no fun!

Right. Lots of mosfets will zener when the drain voltage exceeds the max rated Vds, typically by roughly 10-20%. They often have single-shot avalanche energy ratings, hundreds of mJ range. So an unclamped mosfet can drive an inductive load, like a solenoid or a print hammer.

Of course, avalanching generates more heat. The thermally-limited self-protecting mosfets have an interesting dilemma driving inductive loads: They shut off at the max safe Tj, then the avalanche energy spikes the temperature even higher.

In some cases, you can series a string of mosfets to high voltages without anything to equalize the off voltages; let'em avalanche as needed.

I guess a shorted-gate mosfet makes a cheap 600 volt zener diode. There must be a use for that.

"Avalanche voltage ratings on datasheets have long mystified designers."

Well, maybe they mystified him.

Avalanche also generates serious quantities of minority carriers. What they don't tell you is, the poor thing won't *turn off again* for some microseconds. Slow junk like solenoid coils, no worries; switching converter overshoot? Psst-BANG!

For the same reason (carrier lifetime), avalanche-rated parts have slower body diodes.

That can be a killer when your high voltage (600 or 1200V rating, say) application requires body diode action and they're drooling for almost a whole microsecond of reverse recovery.

That was one of the innovations that made power diodes reliable: the early ones in those charismatic packages (top-hats, studs, etc.) had problems with impurities in and around the junction (especially edge and surface defects), causing hot-spots and destructive breakdown even for very low currents. That's why all the old ARRL handbooks show chains of diodes with resistors and capacitors across 'em, to make sure no one went into the danger zone.

It's pretty safe to connect 1N400x in series, alone, though.

I wouldn't actually recommend connecting ultrafast diodes in series, because you're pushing them much harder. Not only do you get avalanche in the first diode to recover (and attendant overheating*), you get dynamic avalanche, where the breakdown voltage rises as charges clear from the junction. (When this is carefully controlled by doping, you get "soft recovery" diodes. It's not a bug, it's a feature!**)

**And as you well know, gulps of forward-bias charge can do interesting things during recovery. Like drifting and stepping.

Tim

Actually, there is. I've written about it here.

Good stuff to know.

Yeah, I connect diodes in series without thinking about it. Schottkies have a soft reverse current pattern, just leak more as the voltage goes up, so are generally self-equalizing too.

I built the controller for a very weird machine once, a pseudo-medical inversion table, and used a mosfet half-bridge to drive the DC motor. It blew out fet gates. Turns out the body diodes in the Motorola mosfets were excellent step-recovery diodes. When they snapped off, the huge voltage spike zapped the gates.

Here's the output pin voltage of an LM3102 synchronous switcher. The body diode has classic SRD behavior, with four distinct things going on in the first 40 ns. The chip is essentially useless.