|
I've learned a lot in the last couple years. I cringe now when I see some details on the boards that I never noticed before. Mostly relating to loop areas. One problem with the board you have is the pwm signal from the control board is a VERY high impedance signal that's long and going through the air to the driver board. And I didn't even have the ground return wire close to it. It's like a big open circle of gaping noise collection. I added an extra ground wire to reduced the problem I think. Oh heck, I'm not even sure which one you have. haha.
The problem with the mosfet approach I used is there's such a small range of forgiveness for error. The mosfets are either 200v or 250v or something. With 600v igbts, it's idiot proof. And the layout I have worked out has a loop area that's very close to zero. And each of the 3 600v 600amp igbts has its own dedicated signal, and isolated supply. There's also the option to not even use hardware based pwm, and do it all in software, so I could allow for delays on the order of 0.00000005 second so as to make sure each igbt is turning on at essentially exactly the same time. There's always variation with the propagation delay from optocoupler to optocoupler, and I could eliminate that variation.
So to summarize, the new controller I want to do is a SBE ring capacitor, mounted upside down on top of three 600v 600amp igbts, all bolted to an aluminum plate. Each igbt driven by its own dedicated isolated supply/driver that can do 15amp peak. Each driver has a fault return signal that turns off everything if desaturation happens. The dimensions of the base plate would be around 12" x 15". Is that too big? I'm not sure how tall yet though. Maybe 4" or 4.5"?
-Paul
|