Switchmode unusual IRS2113 failure

[fourtytwo42]

I am used to replacing IRS2113 drivers after a power MOSFET failure but the other day after experiencing ear-piercing squawks from a full bridge transformer the fault was surprisingly tracked down to an IRS2113 LO pin stuck low BUT the MOSFET was fully working.

Quite how the device failed I am uncertain though it did occur after a heavy transformer secondary short circuit that was shutdown after a very short time by protection.

My only guess is the pcb layout parasitics under such heavy load caused some nasty unusual voltage shifts around the IRS2113 that it did not appreciate!

My chip replacing skills were tested as the chip is deeply buried in a dense pcb, many years ago I learnt the art of chopping the dead chip out to save the pcb (that being more valuable than a dead chip). In case it ever happens again I socketed it whilst I was at it (with a nice turned pin variety) mostly because changing a chip once without pcb damage is good, trying it twice or three times is just asking for trouble IMOP.

[boB]

What I didn't like about those old IR211x  line of drivers was that their Vcc rating was too darn low.

What is it ?    Like,  18V max ?

boB

[fourtytwo42]

Hi Bob

Well given most MOSFETS don't need more then 12-15V that seems adequate (I think it's 20 btw). You say old but I believe the same design is now used in chips from others (e.g. Infineon) who took over the original companies concerned or licensed there technology, these things originated in Harris / Intersil I believe.

From what I know (that isn't much) all these drivers seem pretty much the same BUT I may have missed something, what ones do you use or recommend ?

[perieanuo]

mosfets don't like overvoltage, irs2113 has mosfets, case closed. you said yourself you heard fluctuations, added with some limit pcb design flaw and some limit capacitor VCC side and trace inductance, that's it. making a rock-solid power stage is not easy, i saw 3000 amp igbt's exploding from some mistakes like this. in the field, we mumble 2 minutes than replace the drivers and the filtering electrolytic caps (newer development, we verify 100n/10n smaaaaall and possible cracked smd caps) adn wait another cycle of voltage fluctuation mains side or the caps to die again. they called it now planned obsolescence

[fourtytwo42]

mosfets don't like overvoltage, irs2113 has mosfets, case closed. you said yourself you heard fluctuations, added with some limit pcb design flaw and some limit capacitor VCC side and trace inductance, that's it. making a rock-solid power stage is not easy, i saw 3000 amp igbt's exploding from some mistakes like this. in the field, we mumble 2 minutes than replace the drivers and the filtering electrolytic caps (newer development, we verify 100n/10n smaaaaall and possible cracked smd caps) adn wait another cycle of voltage fluctuation mains side or the caps to die again. they called it now planned obsolescence

Yes unfortunately when large currents flow with a very fast di/dt then slight layout irregularities become major problems. Myself I would go for parasitics causing an excessive voltage between VSS & COM of this chip that I have seen data sheet warnings about. I don't think it was VCC as there is another IRS2113 on the same power rail along with some other chips non of which suffered a failure. What interested me was the driver failed without a corresponding power MOSFET failure (the thing it was driving).

[perieanuo]

mosfets don't like overvoltage, irs2113 has mosfets, case closed. you said yourself you heard fluctuations, added with some limit pcb design flaw and some limit capacitor VCC side and trace inductance, that's it. making a rock-solid power stage is not easy, i saw 3000 amp igbt's exploding from some mistakes like this. in the field, we mumble 2 minutes than replace the drivers and the filtering electrolytic caps (newer development, we verify 100n/10n smaaaaall and possible cracked smd caps) adn wait another cycle of voltage fluctuation mains side or the caps to die again. they called it now planned obsolescence

Yes unfortunately when large currents flow with a very fast di/dt then slight layout irregularities become major problems. Myself I would go for parasitics causing an excessive voltage between VSS & COM of this chip that I have seen data sheet warnings about. I don't think it was VCC as there is another IRS2113 on the same power rail along with some other chips non of which suffered a failure. What interested me was the driver failed without a corresponding power MOSFET failure (the thing it was driving).

yep you may be correct, but don't forget each final stage of this driver is kinda push-pull configuration, if the lower mosfet dies first, you shutoff the power mosfet driven by that stage, so the mosfet goes happy to sleep, it's a 50-50 chance to stay alive for power mosfets