IGBT brick internal gate resistor power dissipation

[davekni]

This may seem like a strange concern, but I'm wondering about the possibility of internal IGBT brick gate resistor failure.  I'm designing a QCW for long 100ms pulses starting around 170kHz.  Planning to use CM600DY-13T half-bridge bricks.  Internal gate resistance is 1-ohm, with recommended 1-ohm external.  Running +-20V gate drive into 6uC/gate at 170kHz is 41watts/gate or 82watts/brick.  With external 1-ohm, half that 82watts is external and half internal.  Thus the internal gate resistors are running at 41watts/brick.  If the internal gate resistors are thick-film and relatively small, I wonder about long-term failure.  Thick-film resistors don't survive power pulses very well.

For those of you who have deconstructed IGBT bricks:  How robust to the gate resistors appear?  Is it possible to determine resistor element technology (thin-film, thick-film, or wire-wound)?  Are the resistors thermally connected to the base, or suspended above?  Should I be concerned about 41 watts for a CM600DY-13T brick gate drive internal power?

Thank you all in advance for advice.

[Mads Barnkob]

I have never seen internal gate resistor that look like a regular resistor.

I have only seen integrated gate resistors that looks like underneath, small island pads with a thin outline to the underlaying layer.

[davekni]

Mads,

Thank you for the picture!  I can't tell what the resistor technology is.  It is nice to see that the resistors are on the heat-sink surface along with the IGBT die.  At least that helps with power dissipation.

Measuring Mads image, gate resistors account for about 2.2% of the total die area.  If I assume similar ratio for CM600DY-13T with total package power limit of 4000 watts, then gate resistors could dissipate 88 watts.  That is well more than the 41 watts I calculate for my QCW design.  Encouraging to know - indicates my gate resistor power concerns are low-risk.

Anyone have images that differ in resistor type/mounting from the image Mads posted?

[Mads Barnkob]

At bottom of page. Here you can clearly see that the bond wires are just jumping pad on to the gate in the middle of the islands (no internal gate resistor): http://kaizerpowerelectronics.dk/technology/teardown-schneider-electric-frequency-inverter-altivar71/

Built-in drivers sit on base plate: http://kaizerpowerelectronics.dk/teardown/3-phase-igbt-module-teardown-fuji-7mbp50ra120-from-3-kw-omron-servo-drive-isolation-day-36/

CM600 that I removed RTC from, same internal resistor pads can be seen at the bottom

/>