Killed half bridge - what went wrong?

[throbscottle]

I've been having nice success using a IR21531S as a speed control for a 24vdc motor which draws around 3A running and over 10A startup. A time-switched resistor limits the start current (3R3 for about 200mS, then 0R1 to feed a tacho in the future). The motor is the spindle motor for my PCB drill, finally moving from perf-board prototype to a proper board. The high and low side transistors are FDP047AN08A0 (75V, 80A, 4.7mOhm)

I've been running it from an old wall-wart with the speed turned down, for testing, and it's worked very well. Since today is the day I was going to use the prototype to drill the PCB which will replace it, I connected an unregulated supply consisting of transformer, rectifier and 3300uF capacitor (just what I happened to have lying around), 27v no-load, so it can run at full speed.

But what actually happened was that the high side transistor blew s/c, apparently at switch on. The only thing that was different was I'd missed off the on/off control connection to the IR21531S after previous tinkering, so it went straight to "on" instead of starting up in shutdown, but I don't really see why this would have caused the failure.

So can anyone suggest what might have actually happened at switch-on, here?

As always, thanks in advance

 

[Tony Stewart]

flyback avalanche failure?
protection?

 

[ZipZapOuch]

I've been having nice success using a IR21531S as a speed control for a 24vdc motor which draws around 3A running and over 10A startup. A time-switched resistor limits the start current (3R3 for about 200mS, then 0R1 to feed a tacho in the future). The motor is the spindle motor for my PCB drill, finally moving from perf-board prototype to a proper board. The high and low side transistors are FDP047AN08A0 (75V, 80A, 4.7mOhm)

I've been running it from an old wall-wart with the speed turned down, for testing, and it's worked very well. Since today is the day I was going to use the prototype to drill the PCB which will replace it, I connected an unregulated supply consisting of transformer, rectifier and 3300uF capacitor (just what I happened to have lying around), 27v no-load, so it can run at full speed.

But what actually happened was that the high side transistor blew s/c, apparently at switch on. The only thing that was different was I'd missed off the on/off control connection to the IR21531S after previous tinkering, so it went straight to "on" instead of starting up in shutdown, but I don't really see why this would have caused the failure.

So can anyone suggest what might have actually happened at switch-on, here?

As always, thanks in advance

Post a schematic. Hard to follow the written description alone.

 

[Papabravo]

The usual culprits are slow switching time due to inadequate current drive, and absence of dead time.

 

[throbscottle]

Schematic attached. It's part of a larger drawing so is a bit squashed up. Mostly pre-biased "digital" transistors are used for the switching, Q15's bias resistors do double-duty as the timing resistor in the inrush limiter on the right. The 24v supply goes to the terminals PL1 marked "motor HV". If it wasn't used there would be a jumper on PL9 giving a couple of internal options. The IR21531S is a self-oscillating half bridge driver with 6uS dead time. D16 and D18 enable duty other than 50% to be selected (RV4 is 100K btw). The switch on the left gives options of always on, turn off after about 15 seconds, or on/off on every cycle. R16 on the left gets it's input from a comparator, Q17 causes the input from the footswitch to be delayed until the inrush limiter has finished (it feeds into a flip-flop, not shown). D17 stops the 15.6v Vcc of the bridge driver from feeding into the 5v logic Vcc (Q13 collector must always have a supply or strange things happen).

"Stop" is for an emergency stop switch

It's currently built on perf-board but I tried to keep the bridge circuit as tight as possible and used a star ground.

It's the high side bridge transistor that blew.

 

[ZipZapOuch]

Schematic attached. It's part of a larger drawing so is a bit squashed up. Mostly pre-biased "digital" transistors are used for the switching, Q15's bias resistors do double-duty as the timing resistor in the inrush limiter on the right. The 24v supply goes to the terminals PL1 marked "motor HV". If it wasn't used there would be a jumper on PL9 giving a couple of internal options. The IR21531S is a self-oscillating half bridge driver with 6uS dead time. D16 and D18 enable duty other than 50% to be selected (RV4 is 100K btw). The switch on the left gives options of always on, turn off after about 15 seconds, or on/off on every cycle. R16 on the left gets it's input from a comparator, Q17 causes the input from the footswitch to be delayed until the inrush limiter has finished (it feeds into a flip-flop, not shown). D17 stops the 15.6v Vcc of the bridge driver from feeding into the 5v logic Vcc (Q13 collector must always have a supply or strange things happen).

"Stop" is for an emergency stop switch

It's currently built on perf-board but I tried to keep the bridge circuit as tight as possible and used a star ground.

It's the high side bridge transistor that blew.

That's exactly as I understood a very small part of your project when I read the text version.

 

[throbscottle]

It started life as a simple on/off switch!

 

[rjenkinsgb]

You don't have a flywheel diode from the right side of the motor to positive - a spike that polarity could possibly damage something?

 

[eTech]

Can't see value of CT. What is the operating frequency?

 

[throbscottle]

CT is 470p. Frequency should be in the region of 15kHz with RT being 133k. I thought it would be too high (but tried it anyway) but it seems happy.
rjenkinsgb - you might be onto something - the resistor R29 is wirewound. Could it conceivably have generated a spike on the first cycle? Big enough to kill a 75V mosfet on the other side of the motor? Which I suppose would still look like a short at that point. That diode is looking like a good idea...

 

[throbscottle]

I was having a bit more of a look at this. The transistors would be subject to a spike during the dead-time, however these ones (fortuitously since they were in my junk box) are UIS (aka avalanche) rated and have a low Qrr body diode, which are desirable characteristics to have in a bridge for this reason.

Despite this, could the sheer size of the motors inductance, combined with the low resistance of the supply, have been enough to exceed the UIS rating - despite the 3.4R series resistance? In which case, If I had a fast recovery diode across the bridge's lower transistor, would it have stopped the problem? (I'm saying 3.4R assuming it happened at switch on, but it could have been after the resistance dropped to 0.1R)

Tony - is this where you were going with your response? I am ignorant of many things!