protecting mosfets in UcD

[zilog]

A few minutes ago I killed the output devices in my DIY UcD. They died during a swift burst of high output current (10-12A rms sine into resistive load), the mosfets being FDP3682. They often die for me during high power bursts, and I wonder what beside high Ids that might have killed them (I believe my deadtime still is on the high side, and the mosfets turn on slow). My gate drivers dont have any protection against negative or too high Vgs, should I add a zener of 15-20V and an antiparallel schottky to window Vgs?

[RX5]

I think there IS none... esp when your in the TWEAKING stage...

I have blown some FETs myself... even a solder shorting adjacent pcb tracks CAN kill ...

[MOER]

Do not use these Mosfets as their tr at 46usec is slow. You need massive dead time and the addition of about 10 ohm gate resistors to protect them. Compare the tr of the IRFB4212 at 7.7usec or the STP14NF10 at 16usec and that is why these fets work much better.

[Eva]

Note that you can't make direct comparisons between switching times of different MOSFET because they are strongly dependent on the gate resistor chosen for the test and not all datasheets follow the same proportion of resistance to gate capacitance.

For example:

FDP3682 is rated with Rg=16 ohms for Cin=1250pF
STP14NF10 is rated with Rg=4.7ohms for Cin=400pF
IRFB4212 is rated with Rg=2.5ohms for Cin=550pF

The latter is obviously going to show the shortest times, and the former the longer. In my opinion the FDP is the most suited part for cass D and the STP the least (taking body diodes into account).

[Baldin]

It's 46 nano seconds (nS), and can't at all be considered as masive. Think these MOSFETs are ok for a class-d application. Don't think thats the problem.

Are you using a fast anti parallel diodes over DS on the fets? (like ES2D)
Are you using snubbers on the mosfets? (like the 10 Ohm-100pF in the app note)
I think that it could be caused by high spikes.

Is it running hot at idle?

Could it be the drivers not dilivering enough current at high output, and therefore not making a clean switching of the fets, causing through shoot?

[RX5]

Eva,

can you name -other- important parameters in choosing MOSFETs for Class D operation??

Im sure many would benefit from that....

Regards,
Raff

[MOER]

Sorry about the typos nano secs is what I meant

[Eva]

Originally posted by RX5
Eva,

can you name -other- important parameters in choosing MOSFETs for Class D operation??

Im sure many would benefit from that....

Regards,
Raff

Gate drive requirements (total gate charge), body diode reverse recovery charge (sometimes hard to compare due to different test conditions), body diode maximum recovery dV/dt (immediate failure when exceeded), conduction losses...

[Tekko]

Hum i think thaty zilog have either gave up, lost internet or taken his life, because he havent logged in here since he posted this thread or been in the irc channel.

[zilog]

Originally posted by Tekko
Hum i think thaty zilog have either gave up, lost internet or taken his life, because he havent logged in here since he posted this thread or been in the irc channel.

Im at my parants, molding pieces of race faring for my motorcycle.