Mosfet failure (Drain-Source shorted)

Hello everyone.

I´m having problems with switching mosfets.

Some of them appear to be damaged after some time, staying in ON state regardless control signal state.



The system is powered by a 7.2V lithium pack.
3V_2 rail is used to power an SD card.
12V rail is used to power 2 current loop sensors. Each sensor is driven by the circuit below:



Q43B, Q44B and Q23B are the problematic parts.
Some of them fail after a few weeks, some of them after a few months, but always showing the same kind of failure, short between drain an source.

I haven´t been able to reproduce the issue. I hammered the circuit turning the transistors on and off thousands of times but could not get to any conclusions yet.
I detected high current transients when turning Q43B on, but as I said before, I couldn´t break it.

At the moment I only have theories about what is happening, like high dv/dt between drain and source or ringing at the mosfet gates, but I do not know what else to do to confirm where the problems are.
I know those pull-up and pull-down resistors values might be excessive, but I haven´t designed the system, so I can´t explain why those values where used.

This is MOSFET datasheet:
**broken link removed**

I would really appreciate any suggestion or comment, as I am stucked.

Thanks

Q23B is operated at the edge of Vgs,max rating (12V). Possible damage reason is either momentary overvoltage (Vgs or Vds) or overcurrent. Could there be input voltage spikes, e.g. when disconnecting the battery? Perhaps you should place a TVS diode.

Hi FvM.

I missed that Vgs little detail. Good point.
I assume there are some significant transients on Q43. I need to do some more testing.
The battery is never disconnected.

I also found something but not sure is it is enough to damage the transistor.
When 12V rail is enabled, Q23B/Q24B are on for a short period of time until 12V voltage becomes stable. I suppose this is because RC filter made by gate pullup resistor and gate capacitance. I detected a voltage drop (around 1V) on that resistor. The picture below shows 12V rail voltage (yellow trace) and IO_PIN1 (green trace) when U10 is enabled:



- - - Updated - - -