ping Win, SiC fets

In Spice 1f == 1fF, 1u == 1us, etc, only the first letter is "seen". ...Jim Thompson

Sometimes between 1795 and 1960 at least as kilogram has been officially been abbreviated as kg.

Inofficially in computing uppercase K has been used to mean 1024. Thus

kV and ns are bad, but doable I suppose. But 140 amps and ns, with MOSFETs, that's another matter. 140A/ns creates a voltage V = L di/dt = 700V of across 5nH. Consider the FET's source lead inductance carrying this high dI/dt, and the safe, controlled 10V Vgs you're trying to impose from gate to source. Ouch.

I'm doing 10A/5ns and having no problem.

Yep.

Hey, triacs are for *electrical* engineering, and we're into *electronic* engineering. But, yes, this time there's some basic info and part numbers for interfacing to triac line switches.

Oohh. Controversial territory! :->

The Ixys (old DEI) flat fets might work. Money might not matter here.

Less than two orders of magnitude away! I can probably reconfigure some things, like driving 75 ohms and such, and get down to 70 amps peak, now a mere factor of 7 from your case.

Can you reveal what you're using for gate drivers?

Hah, I now have 6 different versions of the board, with six different gate-driver approaches, trying to reach my 10MHz repetition-rate goal. I'm not happy with any of them. My latest scheme gives up on TO-220 driver ICs, and uses a fast recent-vintage driver. I suffer its poor thermal properties, and add NPN and PNP emitter-followers to handle some of the C V^2 f power dissipation. Note that SiC MOSFETs have 1/4 the capacitance but require twice the gate voltage, hence 4x more power killing that advantage. The dissipation is now shared by two 1812 resistors, two sot-223 BJTs and the driver IC. The latter has a heatsink.

The Art of Jumper Cables.

Wow, I thought that war had been settled in the 1960's!

Jon

Correct. However, a capacitor marked as just 1 is interpreted as 1 farad.

And Jim is correct about meg or Meg being an appropriate multiplier.