IGBT failure and circuit explosion

Question

I have a circuit with IGBTs which charges and discharges two different capacitor lines as a function of the voltage. i.e. when a voltage is detected below 160V, Q1 charges C1 and when a voltage over 160V and below 500V is detected, Q2 charges C2, and all voltages over 500V and up to 1000V are charged with Q3 and C2 and C3 in series.

After charging, the capacitors are discharged by R1 and R3 depending on which circuit was used.

All IGBTs are driven with a UCC21520 driver by means of a microcontroller. Isolated power supplies are obtained from an NMK1515SC DCDC converter with 3000Vrms isolation during 1 second.

^{simulate this circuit – Schematic created using CircuitLab}

The problem is that charging at high voltages (over 800V typically,) Q1 is failing, going into short-circuit and therefore C1 gets exposed to 800V and ends up exploding.

Sometimes it also happens in Q2, although less frequently. I suspect various things could be happening but not 100% in any of them and not even sure if it could be some other thing:

• When Q3 is closed V_CE of Q1 changes from 850 to 0V and then slowly back to 850V with the charging of C2 and C3. This without Q1 beign activated and could be the cause of the failue of Q1.
• Right now not using a resistor betwenn the IGBT driver and the IGBT gate which could cause voltage overshoot and cause the IGBT to fail. But this should occur on the IGBT beign triggered and not another one.
• Maybe IGBT model is not the most appropiate.

Any other ideas would be very much appreciated.

IGBT datasheet.

IGBT driver circuit:

Answer 1

Without some form of current limiting external to the IGBT, turning it on when there are hundreds of volts on the supply is going to generate a pulse of current due to the physical nature of a discharged capacitor. That current can only be limited by the supply itself and the ESR of the capacitor. The IGBT data sheet says this: -

So, if the pulse of current exceeds 20 amps, you could be in trouble. The IGBT you have used is extremely fast at turning on with a current rise time of typically 3.5 ns and that, to me, suggests it is a candidate to consider.

Answer 2

Problem seemed to be the accidental turn on off Q1 possibly due to Miller current. In order to avoid it we have connected the pull-down resistor to -12V so as to have a negative VGE during turn off. New schematic is shown where V6, V7 and V8 represent the drivers voltage when turned on. When turned off the driver outputs -12V and prevents IGBT accidental turn on due to induced voltages at the gate. 10Ohm gate resistors where also added to slow down IGBT turn on and avoid overvoltage events during turn on and turn off

^{simulate this circuit – Schematic created using CircuitLab}