arrgg - ES1J failed

Tim

That's kind of the obvious go-to, in my mind -- then again, I don't see "SiC" in the AoE3 index, so perhaps Win could use the education? :-)

I'm sad that they don't have any SM{A|B|C} style SMTs -- only DPAK and up. Also, that they don't have any small ones -- they start at 1.5 or 2A or thereabouts, and go from there. Maybe there just aren't that many applications for low power, high frequency, high voltage diodes?

Tim

That is likely, no market. Fluorescent backlights are going the way of the dinosaurs and those wouldn't have tolerated the high prices anyhow.

We've been down this road before. Your circuit has enough inductance in ser ies with the diode to avalanche it when it tries to abruptly cease conducti on at the end of reverse recovery, causing enough of a peak power dissipati on to eventually destroy the part.

. 20V ---####--+---|>|--+-- 350V . | | . +--, cap . D | . --22-- G body diode . S | . +--' . gnd

Wouldn't any inductor flyback be halted by the MOSFET's DS diode?

And by the fet turning on.

But that's easy to scope. Scoping fet current would be interesting, too.

The diode dissipation is almost certainly reverse recovery. Get a better diode!

You can do this with a standard dual-winding inductor:

which may need a little snubbing in discontinuous mode.

Not if the series inductance is on the anode side (output) of the circuit...

Correction- that would be the cathode side-

Less than 0.5-inch of PCB trace both sides.

So, the MOSFET-diode-capacitor loop is under 20nH, more or less? Should be pretty low.

It is noteworthy that diodes exhibit dynamic avalanche: soon after the junction "turns off", it's only "off" to a certain breakdown voltage. Effectively, the junction is thinned by the gulp of charge still diffusing/recombining away.

This isn't usually documented in diodes, but is sometimes documented in switching BJTs. Some discussion here:

They say the time dependency is cooked into the RBSOA conditions, but I've seen curves that plot time parameters as well. (I can't seem to find the transistor I saw that on.)

You can avalanche a switching BJT by using a very inductive load. The collector waveform is clamped by avalanche, but the clamping voltage rises as time goes on. It takes some microseconds before it will handle the full Vcbo (assuming you're in the negative base bias region).

Anyway, the same is true of diodes, they're just a bit faster thanks to not needing to be optimized for useful hFE. :)

Tim