LM2903-Q1: Destroyed part, hint for root cause?

Hello FRAMziskus ,

It is difficult to tell from "resistance" measurements due to the wide variation in applied voltage/currents by various DMM's.

Normally you should see a "diode" across the supply pins in one direction, and a high impedance *if* the applied test voltage is less than the minimum operating voltage (or about 1.2V for the LM339 family). The inputs and output should show a diode in one direction and high impedance in the other direction from ground, and high impedance in both directions from V+ (the LM339 family does not have ESD diodes to V+).

If you are seeing a solid resistance measurement in both directions, most likely it is a melted junction (short). Typically we would expect "melting" to occur if the voltage is about 45V and the available current is high enough. Voltage just above abs-max can cause soft-failures that may not be immediately noticeable.

The LM2903/LM339 family is pretty rugged. The inputs can withstand up to 36V even with no supply. What kills these devices is negative voltage - or taking any pin below the GND pin by more than 500mV.

Pin 3 also looks damaged, though it does show a slight difference in readings depending on polarity (still some functioning junctions).

I would investigate if the supply voltage is exceeding Abs Max, and if the supply and/or Pin 3 is going negative at some point. Run through a complete power up-down cycle and watch for transients at power-up, operation and power-down (and wait a few seconds until all caps have discharged). Use a scope and monitor all four pins (V+, IN+, IN- and Out).

Can you provide a schematic and the conditions of the failure?