MOSFETs being destroyed

[akis]

I have a simple 6S battery monitor/controller that cuts off the battery if there is undervoltage (cells drained too much) or overcurrent (some short or error condition). It is not perfect because it does not monitor individual cells however it is there to protect against gross faults. It cuts out at a relatively high voltage (22V) to save any weak cells.

There are two batteries protected by their individual monitor that power a device. The device may cause issues because: (a) it has two large capacitors 2,200uF for each of the supplies. (b) it puts the independent supplies in series.

The problem is that randomly, the pass MOSFETs blow (and always in the on state, which means the batteries now can be left to drain too much).

I have used a few different MOSFETs and they all blow at one time or another. I have 6 such battery packs, with two batteries each, and all monitors (12) have blown their FETs.

I have tried:
PSMN4R3-30PL : VDSS 30V, Id 100A, RDSon 4.3mOhm, VGS(th) 1.7V
PSMN017-80PS

I suspect that the Arduino pin D2 does something stupid and blows the FET gate, maybe. Or maybe the device's 2.2mF caps could be charged and then discharge badly into the battery monitor somehow. I have tried to simulate this condition but did not get anywhere.

So first things first, how do I go about protecting the gate ? Do I add a cap to the source to smooth out possible spikes? Add a zener to prevent voltage spikes?

I attach the schematic which shows an individual battery monitor. Pin 1 is battery V+ which also feeds the device, Pin 3 is battery V- and Pin 2 is device return.

I also attach another schematic simulating how the two batteries power the device together. R1/C1 and R2/C2 are the "device" being powered by two independent 6S batteries.

[OwO]

If you are putting two such battery packs/power monitors in series then when a mosfet opens it can see double the battery voltage, since it now sees two batteries in series with the device. You need a mosfet that can handle >60V.

[MasterT]

My understanding, that R3, R4, R7, R8 are to "slow down" initial charge current spike? Than it's wrong way to do it. Very likely MOSFET failed due overheat /overpower during this start-up period.
Better to drive MOSFET as fast as possible, removing all 4 resistors, and adding   special MOSFET driver IC to accelerate on/off timing.
To reduce current spikes use an inductors in series with probes 2 & 4.

[akis]

On the simulation diagram, R3 and R4 are gate resistors to limit current through the gate when switching - they are driven from a digital output pin of an Arduino, as shown on the other diagram. On the simulation instead of the Arduino I have placed a fixed 5V supply with a manual switch on-off.

R7 and R8 are current sense resistors. Not shown on the simulation diagram, but the rest of the circuitry can be seen on the proper schematic. The outtput from these resistors drives/is read by an analogue pin  on the Arduino and the software can tell the overcurrent condition.

If you are putting two such battery packs/power monitors in series then when a mosfet opens it can see double the battery voltage, since it now sees two batteries in series with the device. You need a mosfet that can handle >60V.

I thought that, but I cannot simulate it (ie I cannot see an electrical path that places > 24V on the MOSFET). That idea was exactly the reason I tried to simulate the circuit. In any case I then used the PSMN017-80 which is 80V, and those still blow.

That is why I then turned my attention to how I am driving the gate, it is a digital pin of the Arduino, and I know they are tristate, can it be that somehow the gate sees a lot of voltage as the Arduino powers up? Should I place a zener to protect the gate?