It works and I KNOW WHY!

[Oznog]

Stupid PIC MOSFET driver...

I finally found why this thing goes nuts. Even though I'm just driving a gate, no load on the FET, the pin and rail were seeing spikes to 6.5v or so and it latches up. It uses a reg and had a 0.1uF ceramic on the end of the DIP pkg. That 0.1uF is also the cap for the reg output which is only 0.1" away.

It turns out that either 0.1uF was not enough or it wasn't located close enough, the Vdd/Vss pins on the PIC18F1320 are in the middle, 0.5" from the end of the chip. Sticking another 0.1uF on the underside of the board directly on the pins makes everything run nice and smooth!

A real feel good moment.

 

[williB]

Whatcha makin ? 8)

 

[Oznog]

It's that strobe that I needed to use a fet to commutate the xenon flash. Actually now the HV transformer is being fed by a fet half-bridge. I didn't even power the drains to see the problem.

I saw the real problem- I can't believe I was this stupid! 0.1uF ceramic is not enough to stabilize any reg. I have a MIC2950 which is described as a nifty"bulletproof" reg with a low output capacitance requirement, but it requires "only" 1.5uF. It was bugging me as I went to sleep last night how driving a gate capacitance could produce >5v on the rail. I realized it was impossible and the reg stability had to be the problem. I don't know when I decided a reg could work on one of the cheap ceramics I have so many of, I need to check my other designs to make sure I didn't do this.

MIC2950- great micropower reg, handles voltage spikes over 60v, low quiescent current, low capacitance requirements. Most often comes in a TO-92 pkg. **broken link removed** carries them cheap.

 

[williB]

can i see a schematic , if possible..

 

[Oznog]

Darn it... it got up to 450v, the tube fired, and blew the bejeesus out of the whole thing.

The thing was built on a Radio Shack prototyping solder board. The power trace for the +450v arc'ed onto the ground trace. It didn't go when it reached 450v but as the tube fired (or maybe during the commutation) the high dV/dT rate broke it down. I think I saw flash start in the tube at least.

The trace is a bit warped now. At least the op amp that was buffering the voltage divider so the PIC's ADC could read it is toast. PIC may likely be toast too.

This is why I wear safety glasses on this particular circuit. Stuff blowing up wasn't entirely unexpected.

 

[Someone Electro]

LOL :lol:

Remids me of the resisistors pin shoring the cap bank on my coilgun.