View Single Post
Old 10-22-2015, 03:59 PM   #2224 (permalink)
e*clipse
Permanent Apprentice
 
Join Date: Jul 2010
Location: norcal oosae
Posts: 523
Thanks: 351
Thanked 314 Times in 215 Posts
Quote:
Originally Posted by MPaulHolmes View Post
I think I did that test of the turn-off spike, but now I don't remember the results. I could do it again. I know the "theoretical" turnoff spike is supposed to be:

V_turnoff_spike = L * di/dt. And the sbe ring cap is supposed to have about 5nH of stray inductance "in a suitable laminar bus structure" haha. So, if you are turning off in 1uS:

V_turnoff_spike = 5*10^-9 * 1000Amp/10^-6
= 5 volts, assuming 1000amp.
That's good news; great - in fact.

Just roughly speaking - no solid numbers - did the tests you did sort of match the theory?

So, lets say I really screw up the bus design and the inductance is thee times that. At the same time, my insanity w/ SiC drives me to decrease the turn off time to 28nS driving the Wolfspeed SiC switch as hard as possible. Fortunately, I'm running about 200A max (1/5 of the estimate).

With all that silliness:
V_turnoff_spike = 15*10^-9*200Amp/28^-9 = 107V

That sounds bad, except 200A is twice the current I will use AND I'm not likely to screw up the bus structure that bad. The main parameter I can screw up in the bus structure is the distance between the plates, and that's not a very sensitive number.

So, it sounds like a 700V max capacitor would be pushing the limits, but work. It seems there's a decent safety overkill in those capacitors. For example the 700V rated capacitor can handle 950V for two minutes. A 900V capacitor would be really safe. That's great news- there's lots of choice in the 8 1/2" diameter range - up to 1000uF.

Thanks a bunch!

- E*clipse
  Reply With Quote