How Mosfets Burn Out with Heat

A MOSFET can fail in an overheating manner that ends up as a shorted blob. You can also punch the gate oxide and get a resistance from the gate to the channel. If you apply too high of a dV/dt at the edge of turn off you can get a reduced drain breakdown voltage etc.

I have seen some where one pin was _gone_. Seriously.

You may want to check a bit more carefully. In my experience a common overheating failure mode is a gate-drain short. This means the MOSFET turns on from the drain voltage and stays on hard enough to pull the drain voltage down to about Vgs=5V, etc. While you could call that a "short," it's actually a slightly specialized version of a short.

In that case, they tend to fail open.

Best regards, Spehro Pefhany

ftp://66.117.156.8/ExFets.jpg

John

Nice collection... :) " I call this one 'Poppy'. And I've named this one 'Cracky'. That's 'Snappy'. And that one is 'Poofy'....." :P

A good reason to wear eye protection. *;(

Most circuits I make can't draw enough power for an explosion but seem to have enough power to 'blobbify' the die.

D from BC British Columbia Canada.

I'll check for that in the future.. It just seems I discover 3 pins shorted out more often than just 2 pins.

For example: I recently killed a FCPF11N60 TO220. No external damage. Removed from circuit.

Gate to Drain: 2ohm Gate to Source: 2ohm Drain to Source: 0.1ohm (probe wire resistance)

However, this one didn't die due to a slow rise in temperature. That mosfet was working fine until I goofed up a test. It fried instantly after power up.

D from BC British Columbia Canada.

It's a typical end result of all failure modes, if bonding wires don't fuse first.

Mosfets will still function at temperatures where the die-bond solder is liqufied and migrating out of the pckage seams.

You're supposed to be able to recognise these temperatures before this happens, if all nuts and bolts are secure. It's a detectable and recoverable condition. (how many times.......?)

RL

I know that. That's why it doesn't make any sense, I tried IR, VISHAY = parts, they all behaved the same. I tried their sample circuit, none of = them can survive the Kickback 1.2KW transformer.

Yup Fully heatsinked.=20

their spec....

Mosfet.

I forgot to mention that I did find the solution for my Mosfet problem, = by using 4 MOSFET's, they seemed to be Ok. 4 X 300W =3D 1200Watts = total.

You never know. On the one with the vaporized leg they thought so, too, because it was battery operated. But then there was this huge capacitor ...

Right but WHY WHY? did they rate it at 90Amps? 150V? I put in only 15V =

amps.

Did you provide somewhere for that heat to go?

Someone should try to *read* the spec and apply Ohms's law

Not to someone who doesn't read the specs.

... or somebody is misreading the spec?

If I recall correctly, MOSFETs which are designed for power-switching service are spec'ed by their maximum standoff voltage, and their maximum on-current when driven into full conduction, and their maximum power dissipation. The latter is *not* the product of maximum voltage (at zero current) and maximum conduction current (at minimum voltage drop across the conduction channel).

If you try to use one of these MOSFETs in an application where you aren't switching it hard-and-fast, you can end up dissipating more power in the MOSFET junction than you'd expect, and may exceed the max-dissipation or junction-temperature limits, even if neither the voltage nor the peak current approaches or exceeds the spec limit. You end up pushing the part out of its safe operating area, and it releases its magic blue smoke.

In some ways, yes, in other important ways no. Bipolar parts have the same sort of safe-operating-area limitations as MOSFETs. In some ways they're even more trouble-prone, if you ignore SOA rules, as they're more prone to second-breakdown thermal runaway.

Quite right. Drop the drain voltage or pulse the drive, then touch the FET with your fingers (an instrument that is always available unless you are really careless) and you should find that the FET is getting too warm. Having found that, review your circuit carefully; sketch VI curves, calculate switching losses, check worst case threshold voltages, etc.........

Of course... But sometimes I get math lazy and just pop in the best guess and let'r rip.. Then poof! Most often I find mosfet pins all shorted out after a failure. Not entirely a bad event :P It gives me an idea of how much more 'upstream' damage can happen in cause of mosfet failure. Ex. Exploding driver chips.

D from BC British Columbia Canada.