Bridge failure - which went first, board or IGBT?

[GrantV]

Greetings and salutations 

The last time I ran my coil I overdid things slightly and was rewarded with the usual flash and bang, which I assumed at the time meant that I had blown an IGBT.

Testing the bridge, I have indeed lost 3 IGBT's, however, unlike the last time that I blew an IGBT, this time none have any physical damage to their casings.

I am attaching photos of my board showing that the tracks off Q4 (full bridge) on both the collector and emitter have been vaporised.

My question is, could this have been a board failure due to an overcurrent/overvoltage/overbps, or, would the failure of the IGBT have caused the board damage?

Secondary question: My coil runs beautifully at very high bps (1500 to 1800) and when it blew, I was likely pushing the bps past 1800. Could this have caused the failure, or, is high bps ok as long as voltage and current are within limits?

Thanks for your comments in advance :-)
Grant

[Mike]

I've seen identical failures on my bridge. I can't see a way that a PCB failure would cause the IGBT to fail. When the PCB fails it becomes an open circuit and prevents further current from flowing. When the IGBT fails however they normal fail short circuit. In this case there's nothing to stop very high currents flowing which will destroy the PCB.

As for your high repetition rate question, higher BPS means higher duty cycle (assuming you maintain the same on time). This results in the IGBTs heating up more. From here multiple things can go wrong. You could exceed the ability of the part to transfer the energy to the heatsink causing failure. The other thing that changes when you heat up the IGBT is the turn on and more importantly turn off delay increase. This can cause your zero current switching to be further from ideal causing more heating and can ultimately result in shoot through which at elevated temperatures especially, tends to be fatal for the IGBTs.

[davekni]

I've seen identical failures on my bridge. I can't see a way that a PCB failure would cause the IGBT to fail. When the PCB fails it becomes an open circuit and prevents further current from flowing. When the IGBT fails however they normal fail short circuit. In this case there's nothing to stop very high currents flowing which will destroy the PCB.

As for your high repetition rate question, higher BPS means higher duty cycle (assuming you maintain the same on time). This results in the IGBTs heating up more. From here multiple things can go wrong. You could exceed the ability of the part to transfer the energy to the heatsink causing failure. The other thing that changes when you heat up the IGBT is the turn on and more importantly turn off delay increase. This can cause your zero current switching to be further from ideal causing more heating and can ultimately result in shoot through which at elevated temperatures especially, tends to be fatal for the IGBTs.

Great answer!  Probably clearer than what I would have written.

[GrantV]

I've seen identical failures on my bridge. I can't see a way that a PCB failure would cause the IGBT to fail. When the PCB fails it becomes an open circuit and prevents further current from flowing. When the IGBT fails however they normal fail short circuit. In this case there's nothing to stop very high currents flowing which will destroy the PCB.

As for your high repetition rate question, higher BPS means higher duty cycle (assuming you maintain the same on time). This results in the IGBTs heating up more. From here multiple things can go wrong. You could exceed the ability of the part to transfer the energy to the heatsink causing failure. The other thing that changes when you heat up the IGBT is the turn on and more importantly turn off delay increase. This can cause your zero current switching to be further from ideal causing more heating and can ultimately result in shoot through which at elevated temperatures especially, tends to be fatal for the IGBTs.

Hi Mike, thanks for that! It does of course make sense that it was the IGBT's that went before the board, I just wanted to throw it out there only because I had found no physical damage to them at all, and there are always lessons to be learned :-)

Thanks also for your answer on the repitition rate. As I am using my own digital ucontroller driven interrupter, it does reduce the on time in track with the increasing bps thus I can never exceed my set max duty cycle. ie. if duty set at 10% bps=200 pw=500 but at bps of 2000 the pulse width will have droped down to just 50usec. Thus, I figured I was relatively in the clear to push the repitition rate higher than 'usual' :-) ?

[davekni]

Thanks also for your answer on the repitition rate. As I am using my own digital ucontroller driven interrupter, it does reduce the on time in track with the increasing bps thus I can never exceed my set max duty cycle. ie. if duty set at 10% bps=200 pw=500 but at bps of 2000 the pulse width will have droped down to just 50usec. Thus, I figured I was relatively in the clear to push the repitition rate higher than 'usual' :-) ?

Sounds like a good design.  Yes, in general you should be fine at high repetition rate since you have kept duty cycle constant.  However, two possible high-repetition-rate issues come to mind:
1) The anti-parallel diode in FGH60N60SMD has half the current rating and just under 1/4th the power dissipation capability of the IGBT.  Primary current ring-down after enable ends is conducted through diodes.  Thus diode current duty cycle does increase with repetition rate.  The initial failure may have been with a diode, which then fried the opposite IGBT and then the IGBT with the failed diode.  You may be able to determine if diodes are good or not by dissecting parts.
2) Depending on GDT+wiring leakage inductance, sometimes there are Vge rings or other anomalies at the end of bursts.  Resulting IGBT stress may be tolerated at low BPS, but generate too much total heat at high BPS.

Good luck with your repairs.

[Hydron]

1) The anti-parallel diode in FGH60N60SMD has half the current rating and just under 1/4th the power dissipation capability of the IGBT.  Primary current ring-down after enable ends is conducted through diodes.  Thus diode current duty cycle does increase with repetition rate.  The initial failure may have been with a diode, which then fried the opposite IGBT and then the IGBT with the failed diode.  You may be able to determine if diodes are good or not by dissecting parts.

This is one of the reasons I picked up some FGY75N60SMD - the diode is still rated less than the IGBT, but by a smaller margin. This is potentially important not just with high BPS, but if free-wheeling or a phase shifted QCW bridge is used - in both these use cases the diode gets a decent workout.

[GrantV]

Sounds like a good design.  Yes, in general you should be fine at high repetition rate since you have kept duty cycle constant.  However, two possible high-repetition-rate issues come to mind:
1) The anti-parallel diode in FGH60N60SMD has half the current rating and just under 1/4th the power dissipation capability of the IGBT.  Primary current ring-down after enable ends is conducted through diodes.  Thus diode current duty cycle does increase with repetition rate.  The initial failure may have been with a diode, which then fried the opposite IGBT and then the IGBT with the failed diode.  You may be able to determine if diodes are good or not by dissecting parts.
2) Depending on GDT+wiring leakage inductance, sometimes there are Vge rings or other anomalies at the end of bursts.  Resulting IGBT stress may be tolerated at low BPS, but generate too much total heat at high BPS.

Good luck with your repairs.

Hey Dave, LOL thank you, I am quite proud of my interrupter :-)

You see, this is exactly why I love this forum so much, post what initially sounds like a no brainer question and all of a sudden you find yourself learning about things you would never have thought of before  Certainly true in my case as I would not have given the anti-parallel diode a seconds thought! I'll have to go do more homework! Happy days.

Re the repairs ... nope, I'll be scrapping that board thank you! Apart from that I do have spares of that board, I'm using the opportunity as an excuse to put my dual bridge board to the test :-) pics later