I have designed a simple h bridge using a HIP4081 full bridge driver and Irf1405 mosfets. I had to modify it by putting schottky diodes to protect the hiside pins on the driver from undershoot. I have tested it within my means and it seems to operate as intended. However it occasionally burns out mosfets, 2 or 3 usually. I dont know if there is some transient issue that i am not aware of. I am running it on 24 volts, 20khz pwm on the low side. Mosfets do not
creat any heat prior to blowing. I have very short gate traces and the ground and power appears to be very stable. I am Stumped.
It's quite possible that you've got a transient violation of Vgs constraints. FETs aren't very forgiving of excess gate voltage, and I don't see any sort of proctction devices in your schematic. Try planting a zener of suitable value across the gate & source of your FET and see it that helps.
I tried a 14 v zener between gate and source and it still blows. Could the problem be related to the fact that i am trying to run a 12 v brushed motor on 24 volts max 50% duty?
I don't think the voltage rating of your motor has much do to with the issue, at least in a direct sense. That the addition of the zener didn't help strongly suggests (though doesn't prove) that excess Vgs isn't the issue; zeners still have a finite impedance, and it's possible that there's a transient current sneaking through somewhere that's popping your FETs.
Do you have a 'scope available? Problems of this sort are a lot harder to solve without one. If you have a resistive load (prefereably non-inductive, or at least much less so than your motor) available that'll handle a similar power level as your motor, see if driving that gives you the same results. If the problem goes away, that would suggest that the problem likely is related to the dV/dt that arises due to shutting off current flow to the motor's winding inductance. If not, take a close look at your control inputs.
Since you're not seeing a huge temperature rise prior to failure, violation of some voltage parameter seems most probable. You might want to try adding a bit of impedance back into the gate at turn-off, in order to give you a bit less dV/dt at the output nodes by slowing the switching transitions. Yes, you'll get a bit more switching loss, but it's better than blowing FETs all the time...