Power amplifier SMPS IGBT keeps blowing - Artcoustic PDA1000

[mikkele86]

hello.

I have two power amplifiers, both having their IGBT blown on the PSU.
The original mounted IGBT where marked with IRGB20B60PD1PbF and the gate driver is a IR2156STRPBF.

I looked at mouser and digikey and could not find any available IRGB20B60PD1PbF and somehow ended up using a Bourns BIDW20N60T.
I replaced the gate driver, the gate resistor and both IGBT. But as soon power is turned on, both IGBT shorts out completely (gate, emitter, collector) and the fuse on the AC line gets killed.

So I though i missed a capacitor or diode. I removed allmost every component between the bridge rectifier and trafo. compared them to a new component, didnt find anything of interest..
anyways i decided to replace the gate driver (again) , snubber, diodes and some SMD caps.
installed a new set of BIDW20N60T and the same thing happend again, both IGBT dead instant.


I removed the IGBT and measured the DC voltage coming from the bridge rect. 323v stable.
So im wondering if the BIDW20N60T was a bad choice ? maybe I should have used a IKP20N65H5 instead ?
Tips to what to look for or check is appreciated


I have attached my homemade schematic and some pictures.
Schematic_Artcoustic PDA1000_2024-01-06.pdf

my tools for measurement
FLUKE 123 20MHz scopemeter.
LCR meter DE-5000

Original IGBT and parts.


Parts removed and SMD caps checked.



Rectifier and Caps checked.

[stj]

what drives the gate driver?
i dont see a psu chip in the pictures - on the underside?

[mikkele86]

thought i attached at PDF schematic, but its gone. i will try again with the schematic.
But the Gate driver is the only IC on the board, nothing on th back

[stj]

do you have a component tester that can check caps for esr and leakage?

[mikkele86]

do you have a component tester that can check caps for esr and leakage?

I have my DE-5000 LCR meter, i can measure ESR and capacity but not leakage.

[CapLeaker]

I don't think your Bourns IGBT is a good replacement. The gate emitter threshold voltage is off, the rise and fall times are off. It's like the Bourns is too slow and it needs more gate emitter voltage than the original one.
Compare IRGB20B60PD1 to IKP20N65H5 and that to your BIDW20N60T.
You can get the IKP20N65H5 from Mouser... they have a few in stock.

[stj]

it's a shit design, if some of those film caps go leaky the igbt's can short each other

[mikkele86]

Thanks for pointing that out CapsLeaker.
there is a huge difference in rise and fall times.

IRBG20B60PD


Bournes


IKP20N65H5

Some measurement on C14 and C15 cap.
Looks good to me.

[CapLeaker]

I thought about these 2 film caps as well. I would build a incandescent lamp in series trick to limit the current going to the PSU, so you don't blow up parts all the time. Then I'd assemble this PSU but leave the IGBT's out and check the gate drive. Install the right IGBT's and see what happens.

See how close the original IRBG20B60PD is to the IKP20N65H5? The BIDW20N60T has similar values in most categories, but is much slower that the original. While the IGBT turns on all right, it may be too slow turning off (so it kind of behaves like a stuck on gate drive). The gate voltage is off too...

[mikkele86]

will try a incandescent lamp in series.

I got a 40W, will it do the trick ?

[petehall347]

will try a incandescent lamp in series.

I got a 40W, will it do the trick ?

that is LED

[mikkele86]

dang it. your right. need to go shopping then

[CapLeaker]

No. It has to be an old style incandescent light bulb, not a CFL or LED strings like this one you pictured. I use 60w, 100w and 150w bulbs, depending on how much I want to limit the current. Start with 100w.

[mikkele86]

got it. just made ordre online, couldnt find any local retailer which had old style incandescent light bulb.
ill will post again when things are ready

[Francesc V.]

I hope not to say something stupid, but:

If you would have an osciloscope, without installing the IGBTs, you could check gate signal..........and see if both signals are well synchronised. Before burning them again an again..... :-)

[stj]

will try a incandescent lamp in series.

I got a 40W, will it do the trick ?

-
i have been waiting for almost a year for somebody to post that!
it's not 40w anyway - it's gonna be 2w or 4w with that number of led-strips

[mikkele86]

Got the bulbs. used the 100watt, lights up for split sec.
Everything seems to be running without the IGBTs

I have taken some measurements, with COM/negativ from the oscilloscope to the COM on the PCB (pin 8 on gate driver)
hope its done right 😅

HO and LO gate at the same time.
HO - pin 13 channel A (upper)
LO - pin 11 channel B (lower)


CPH - pin 7 on gate driver

RT - pin 4 on gate driver


CT - pin 6 on gate driver

[CapLeaker]

The first pic that shows HO and LO, I only see one channel B and nothing much on channel A. That's not right. Also check the gate drive resistors.

[mikkele86]

Measured the gate resistor, 15ohm as stated. Path from Gate Pin on the IC to the pad where IGBT gate is, is also fine.
I did some reading on the gate driver and I think the driver is in a UVLO state. Under Voltage Lock-Out.

Im measuring +10.5v on the VCC pin on the IC.




Im measuring +4.5v on the SD (shutdown) pin


I did reach the Start up treshold VCCUV+ on VCC at +10.5v

I dont think we would see a pulse on HO before we enter a preheat stage

[CapLeaker]

O,k, that is squared off now. I guess it's time to put in the new correct IGBT's and fire it up using the light bulb trick. The light should come on for a blip and go out. As more current you draw, the light bulb should get brighter and stay on.
If you turn the PSU on and the light bulb shines bright there is something wrong.

[mikkele86]

its alive 🤩

thanks for the inputs