Avalanche Pulse Generator: Chassified

schematic:

Pictures/layout:

It doesn't appear you're getting avalanche. How high is your 8ns pulse? General comment, all the wiring around the avalanche transistor needs to be ns-style RF wiring, e.g., short leads, with a ground plane, small enclosed area for all fast current loops, etc.

A 2N3904 is a fast epitaxial transistor, and in my limited experience, epitaxials tend to not make decent avalanche pulses. The best pulsers are older diffused parts with fairly low Ft's, things more in the

2N2000's. I think 2N2369 is a candidate, but even then you may have to select different parts, maybe from different manufacturers, to get a good one.

The Zetex avalanche transistors are dependable and make huge pulses, essentially shorting out like an SCR when they fire, but are rather slow, in the 2 ns rise sort of range. They're apparently made in Russia on an ancient fab process.

John

schematic:

Pictures/layout:

Right, it's likely not avalanching.

I find the best way to trigger them is through a small pulse transformer, maybe 6 turns of bifalar wire-wrap wire on a small ferrite toroid. That way, when it fires, you don't have all the base drive stuff loading the pulse.

He does have a nice blocking-oscillator inverter.

John

Yeah, but like I said, I tested these in a different environment and got a hearty 90V dropping to a couple V in under 4ns. (I was watching on the solderless breadboard with the 10x probe and 6" ground wire, so don't expect any remarkable figures.)

All the 2N3904's I have are from the same batch, or at least the same date (Fairchild logo, "2N3904", "331" = 2003, 31st week?). I think I have some in my assorted NPN bin though.

I've been here before: you may remember this, which I believe I posted here.

This was with a PH2369 from around +100V, which I supplied by an ad hoc inverter consisting of whatever inductor was handy, a transistor and signal generator.

Interesting, the waveform I observed from the '3904 was a steep fall for most of the voltage, followed by a flat region about 1us long, followed by an ~RC rise to ~Vcbo. It really did look like a fast SCR. It occasionally had short misfires of about 10V in height once reaching Vcbo.

More information on the project: I wend back down and played around with it some more. My best guess is the sloppy 8ns pulse is actually the sync amp being amplified (or more likely, followed, given where the emitter is) by the avalanche transistor, which isn't avalanching, even from +150V. It's acting more like a zener, obviously with some negative resistance (varying the voltage up and down shows a hysteretic offset to the output voltage), but not enough to break over seriously. But get this, it works perfectly

*with* a pulse line. So I get

schematic:

[...]

Tim, that's an awful lot of emitter degeneration you've got on the avalanche transistor, especially if you want short pulses. I'd wonder if that discourages the avalanche cascade. Might that be a factor?

I tested various transistors in avalanche some time back--I'll check my notes on the 2n3904.

Cheers, James Arthur

Couldn't find what I wanted, but did see that I used a PN2222A successfully, and the 2n2369a--the old standby--as well.

I never did use the avalanche stuff. I found I could make 500pS edges with logic gates into ordinary RF transistors, which was a lot more convenient.

HTH, James Arthur

schematic:

I got a 404 not found...

Martin