How do you kill a SMART MOSFET?

[Rufus]

So I have a board with a couple of these http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00000218.pdf MOSFETs. They low side drive solenoids pulling about 2A from a 28v supply.

The board includes a 33v tranzorb across the supply and blocking diodes and 22v tranzorbs which clamp the MOSFET drains to supply plus about 24v absorbing the solenoid stored energy rather than letting the MOSFET overvoltage clamps handle it. The MOSFETs are driven hard on or off and barely get warm in normal operation.

A board came back with a dead MOSFET which had got hot enough for long enough melt the solder and ooze black encapsulant out of the sides. Nothing else on the board was faulty.

The only way I can see to destroy these MOSFETs is to put excessive power into the overvoltage clamp or pass a pretty huge reverse current through the intrinsic diode and I can't see how either of those things could have happened.

Thoughts anyone?

[ArtemisGoldfish]

Maybe something cooked the MOSFET's internal protection circuitry?

[mikeselectricstuff]

Low gate drive voltage causing too much dissipation and insufficient supply voltage for the protection circuitry to work properly?
Sounds like it passed a lot of power for a long time - catastrophic failure tends to just render a part dead short, which either trips/current-limits the PSU or causes the device to disintegrate, either way they don't tend to dissipate sufficient heat for long enough  to melt anything.

[scrat]

A board came back with a dead MOSFET which had got hot enough for long enough melt the solder and ooze black encapsulant out of the sides. Nothing else on the board was faulty.

Are the protection 22V tranzorb still working, too?
Does the off switching need to be so fast, that freewheeling diodes were not used, or are the solenoid clamps not accessible?
BTW, am I missing it, or the datasheet doesn't specify the clamping diode max current?

[Rufus]

Low gate drive voltage causing too much dissipation and insufficient supply voltage for the protection circuitry to work properly?

It is supposed to be a protected MOSFET not a protected low side switch so you would hope that if there is enough gate voltage to turn it on there would be enough voltage for the protection circuits.

Worth investigating though.

Were the protection 22V tranzorb still working, too?
Does the off switching need to be so fast, that freewheeling diodes were not used, or are the solenoid clamps not accessible?
BTW, am I missing it, or the datasheet doesn't specify the clamping diode max current?

Nothing else was damaged and yes the solenoid current does need to be collapsed quickly on turn off. The over voltage clamp on these devices is a zener between drain and gate so the MOSFET turns on to do the clamping. You are limited by MOSFET continuous and pulse power dissipation ratings. The other protection circuits obviously can't protect the MOSFET while it is trying to protect itself from over voltage.

[xygor]

Maybe the solenoid core was forced to move in an unusual way.  Inductor voltage, V = L * dI/dt + I * dL/dt.  If the solenoid were to move in an unanticipated way, the second term (change of inductance over time) could cause a problem.

Also, transzorbs have a fairly soft knee.