Trying to learn a lesson from failure (IGBT bridge)

As most of you know, there were a couple of days of having a working IGBT bridge, and running it under full 200 amps.

After that, I installed some snubber boards that were on this bridge. I also messed up my original snubber and installed the diode backwards.

That created oscillations that would go below zero that were visible on the scope, but I ignored them and fried my bridge and drivers.

My tentative conclusion was that it was not overvoltage as such, but negative voltage (negative on the DC+ rail and positive on the DC- rail), that fried my circuit.

The piece that burned on one of the cards, in fact, is the low side FET driver chip.

Would you say that this conclusion is warranted by facts?

If so, a simple way to prevent this would be to install a diode such that if the DC+ voltage drops below DC- voltage, it would instantly start conducting. Is that at all sensible?

Or am I on a wrong track?

So let me get this straight: when the bridge switches state, it creates a [dV/dt or dI/dt?] that causes the power rail to swing below 0V?

With no potentials inherently below 0V, this requires a combination of L and C, probably activated by the transistors (that is, the L and C are connected together in such a way that voltage will swing under 0V, rather than already being connected and the change in voltage/current exciting the resonance), so that parasitic (or intentional!) stored energy is able to invert the voltage like that.

Once the voltage is negative, who knows... if your IGBTs don't have internal diodes (and/or the snubber boards don't have diodes), you could get all sorts of troubles. Transistors generally don't appreciate negative collector voltage, and I have no idea what that would do to an IGBT in particular. (An NPN transistor would pull the base down (until the E-B junction fails) because the B-C junction conducts in that direction, but IGBTs are a bit more than a BJT.)

Tim

-- Deep Fryer: a very philosophical monk. Website:

formatting link