Some of our units fail in the field. The cause was found to be a hang in this circuit. A manual POR fixes the problem, but it cannot be recreated in the lab.

Hi William23 ,

Probably in the field you have long supply wires? If there is no shunt capacitor near Vcc pin, rapid load changes may cause huge swing in the supply line (due to wire inductance). Repeat your lab test with supply wiring identical to the field.

Add a shunt capacitor (say 10uF) across the Vcc and GND (optionally with 1 ohm series resistor to lower the Q value) to see if the failures in the field go away.

Cheers, heke

The 5V comes from an onboard 12-5V switching supply. I used a fast oscilloscope (1Ghz) and checked the VCC. I did not see a voltage drop. or any transients. 

I instrumented BST, INP, FAULT, TGUP. I saw BST go to 9V, and as soon as 3.3V appeared on INP, FAULT went high. 25us later FAULT went low. I did not see any change on TGUP.

I thought the only two failure modes where VCCUV and Current Sense. Are there any others? These both seem okay. Also, after the first symptom, I saw the regular pattern of FAULT going high, then low, every 26ms, around the timer interval.