Another Fet Failure thread

[DaleKramer]

BSC057N08NS3 G 80v 100a Infineon Mosfet Failures

I am driving a single fet with a Supertex HT0740 driver.

Here is gate signal into the HT0740 (yellow) and gate signal from the HT0740 (green) with the fet removed from the circuit:

Here is gate signal into the HT0740 (yellow) and out of the HT0740 (green) measured with the scope probe attached directly to the gate on a trace less than 1mm from the gate.

Note, the input to the HT0740 (yellow) now has a 1V/div scale.

In this case the fet does not fail as the load is disconnected from the circuit.

And here is the fet failure when a load is connected.

The load is a motor controller that draws a few hundred milliamps at rest and has about 1000 uf of capacitance to charge when fet closes:

What could be the cause of this failure?

I suspect it has to do with the slow gate turn on time but it appears that the fet is slowing the gate down as the gate signal without the fet is turning on in microseconds.

Thanks,
Dale

 

[Alan B]

Turning a FET on into a capacitive load causes huge current peaks. Model your circuit in LTSpice to learn and avoid blowing hardware. You may need to slow the turn-on enough to prevent FET failure from excessive current peaks.

 

[Jeremy Harris]

Alan B is spot on. If the FET turns on quickly (or tries to) into a big capacitive load then the current at the instant of turn on is only limited by the FET on resistance, the equivalent series resistance of the capacitor, the internal resistance of the battery and the resistance of the wiring.

Let's guess some values and see what the instantaneous current might be at the instant of FET switch on. The supply voltage is about 43 V, let's say the battery is about 20 mohms internal resistance, the FET is around 10 mohms Rdson, the wiring is around 10 mohms and the capacitor ESR is around 20 mohms. This aren't great figures, so probably give a slightly conservative instantaneous current. For a tiny period of time the current at turn on will be about 43 / (0.02 + 0.02 + 0.01 + 0.01 + 0.02) = 537 A....................

This current only flows for an extremely short period of time before it starts to fall very quickly, but it'll blow the FET pretty fast, as you've found out.

If you slug the gate turn on down so that the FET turns on slowly, then it will act like a current limiting resistor during that initial capacitor charge time.

 

[DaleKramer]

Thanks guys,

False alarm for now, at least until I have to drive 16 of these in parallel.

Turns out I had put an unneeded diode in my keyswitch line that disabled my anti-spark resistor. What should happen is, when my ignition key is turned into the spring loaded start engine position, this places a 47ohm 5w anti-spark resistor across the fets. By the time my PIC has booted and sends a signal to turn on the fets there should be no volts across them and then you can release the ignition key.

Since I only have to turn these fets on and off once per vehicle use, it does not matter if they turn on slow, as long as it does not damage them. I am hoping the HT0740 will turn all 16 on in parallel.

I will be back if they do not.

Thanks,
Dale

 

[BatteryMooch]

If you slug the gate turn on down so that the FET turns on slowly, then it will act like a current limiting resistor during that initial capacitor charge time.

Good points Jeremy and Alan...
Be sure to model or calculate the power dissipated by the FET as it passes through its linear region (i.e., when it's acting as a resistor) though. It's easy to blow the FET by turning it on too slow, requiring it to dissipate huge amounts of power before the gate is fully saturated on.

[Edit] Oops, Dale and I posted at the same time. Hoping you can drive those 16 without problems!

 

[DaleKramer]

Just to follow-up, here is the gate and D-S with 16 parallel fets each driven through a 100 ohm resistor and with a 15v zener on the output of the HT0740:

I think I am happy now, looks like Miller time is about 5ms.

But there is always that possibility that the gate could drop out at full power (about 160 amps total for all 16fets).

Anyone think I will lose the fets if that happens?

Notice D-S volts is like nothing at turn on now, dah....

Dale

 

[HypnoToad]

In laments terms. What's the worst case scenario is a FET blow while you are riding?

Will it damage the motor at all?

Or how about the battery?

 

[ZOMGVTEK]

In laments terms. What's the worst case scenario is a FET blow while you are riding?

Will it damage the motor at all?

Or how about the battery?

Under normal use, using an off the shelf controller with sufficient overhead, FET failure should damn near never occur. Just don't run things near the limit and have solid connections. It would be very hard to damage anything past the controller should a MOSFET failure occur. Worst case would likely be some melted wires and potentially blown halls from overvoltage. That would be very rare, typically its just a few cheap FETs.

These guys are talking more about theory, and analyzing design failures, rather than real practice in a commercial controller.