IGBT snubber circuit

Before you get dependent on snubbers. STOP! Look at your circuit physically. If you are trying to switch faster than 1 microsecond your physical layout is essential. You should have a ultrafast freewheel diode as close to the IGBT as you can get it. Even if you think your load is resistive your wires , traces etc are not. THEN, your gate drive has to be very stiff. You have to hold it off when it switches off and the gate tries to follow the collector due to miller capacitance. Also, you need some capacitance near the igbt and the diode across the supply bus. That compensates for inductance in the supply line. look at your off spike, it should be less than your breakdown voltage. much less. Then you can look at your thermals. Do some measurements and calculate your losses. Usually when a device runs like yours does it is a thermal issue. The snubber just lowered your losses prolonging your heating issue. Again look at the physical. For a test, Put a large block of copper on your part without a sil pad. Then put a large sil pad on the block.

To summarize once you get your low inductance layout and your freewheel DIODE you need to look at thermals. THIS SIMPLE CIRCUIT WITH A FREEWHEEL DIODE SHOULD NOT NEED A RC SNUBBER, EVER.
It is not difficult to get simulation results with acceptable accuracy for snubber or main SW power loss analysis on 10KHZ converters. A few parasitic parameters and a right gate driving speed are needed.

More and more SPICE level-3 MOSFET and IGBT models have been used in industry. Nonlinear capacitive characteristics (Ciss, Coss, Crss and Cies, Coes, Cres) are accounted. Switching analysis results accuracy have been improved.

Other than the possible root cause on all the above comments, latch-up is still a possible root cause although this effect has been improved over ten years ago. Simulation cannot reflect latch-up, but it can remind us if there is a high du/dt, which may cause problems.

I used RCD snubbers for IGBTs on 10KW converters. One of my bottom lines was that keeping away from latch-up phenomenon, even short circuit has happened.