I blew up some IGBTs - KABOOM!!!

Yesterday, I was working on a PWM drive that uses 4 IGBTs in a H-bridge configuration. It had been standard practice in the past to disable this drive by unplugging the ribbon cable that feeds the gating circuitry mounted atop each transistor; this circuitry also performs monitors the IGBTs "health" and is used to generate alarms. When this ribbon cable is unplugged, all power and ground is removed from the boards, but the DC bus voltage is still on the transistors. Again, this has always worked.

Now for the kaboom part. We removed the boards and ribbon cable from the drive to examine them. I had the technicians power up the drive. Everything was normal at first, then there was a loud KABOOM! as four transistors exploded. Fortunately the transistors just craked open and debris did not fly out from the drive nor destroy anything adjacent to them. The transistors are rated for 1000A and are fed by a rather large bank of capacitors.

Why did the transistors explode? My theory is that without anything attached to the gate leads, that the transistors went into full conduction, thus putting a dead short across the dc bus. Yet with all my Googling I can't find any info that says that IGBTs will trigger if the gate circuit is open. Evidently, even when the firing boards are attached, even though they are powerless, they somehow prevent the IGBTs form firing.

BTW, this is an off-brand drive - the manufacturer will remain nameless.

Anything else connected to the gates? Cap, resistor, etc.
IGBT shouldn't turn on with a floating gate, but this assumes all devices are perfect. What if one had slight leakage on it's gate?

Electrically_Bound said:

...standard practice in the past to disable this drive by unplugging the ribbon cable that feeds the gating circuitry mounted atop each transistor...
Again, this has always worked.

Click to expand...

It doesn't matter how many times you say it, it's still wrong. The gates should be grounded during this type of test. Direct connection, or via a resistor.
It's difficult to speculate as to how it happened. Somehow you managed to turn on all 4 gates at powerup.
The Gate is capacitive. A discharged capacitor is a conductor until it charges. I suspect something along these lines.