[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [TCML] The Hammertone Predikter



Hi Finn,

This sounds perfect!!!

Nice to see optical gate drive too. Gets ride of those heavy old GDTs. :D With the Predikter, the delay is no problem now ;) I was wondering about diode pumping the Opto driver LED with a capacitor like I did (C4):

http://drsstc.com/~terrell/schematics/OpticalGateDrive-1.0.gif

That improves speed, but it also is sort of scary to drive a lot of current into that diode. But your Predikter solves all that anyway ;)

Current source gate drive at 10 amps is perfect! I use FETs with 10300 pF of input capacitance for another thing and I can pump about 500 peak amps into the gates directly. That turns them on really fast!! But it only works once. ;) But that molten copper sure does reduce the RDS(on) :D

So I can't think of anyway to make it even better. It seems to be perfect now! Excellent work!

Terry



Finn Hammer wrote:
piranha skrev:
Very nice Finn!

I had two questions.

1. What gate voltage are you guys running these days? I think the Powerex bricks are "rated" at 15V but I was just wondering if you still over drive and by how much.

2. You say there are no gate resistors. Is there a story behind that? I would worry about the high dI/dT damaging the gates due to very high gate currents (max Ig) during switching. But maybe the gate driver is not enough to hurt the big CM600s. Of course, switching noise is not a problem with good zero crossing and then snubbers have no use.

Terry

Thanks,

Gate voltage is -10V +20V and the gate driver is a n/p channel totem pole driver, which has very low shoot through. At some point in the development phase, I got obsessed with creating a totem pole without the usual intermediate resistor, to diminish the inherent shoot trough of the combination. The anti shoot through network on the gates of the N/P pair creates a controlled current output of 10A, so the gate voltage swooshes up in 200nS.
http://www.hammertone.com/temp/totempole.jpg
The 2 diodes closest to the gates *really* are 10V zeners, but the cad package didn´t have them in surface mount, so..... It is pretty slick: You cannot see the Miller threshold as is usually visible on gate voltage scope traces.

In comparison, the CM600 datasheet assumes a 2.1ohm gate resistor, which would result in a 15 amp charging current given +-15V drive. I did this to demonstrate that it *could be done* with the predikter, without creating overshoot problems.

Do you think I am on safe ground with this?

Cheers, Finn Hammer

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla



_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla