Yeah, but for how long?
In a past life.. I used to make engine controls for drag racing boats. Early on in the development process, I'd bolted a known working "Launch controller" into a test boat and if failed miserably. It worked on the bench, failed in the boat. Took it back to the bench and, of course, it worked again. Not that it was getting wet or anything. The entire product was a solid block of epoxy. In the end I found that the reason it was failing was that the mount didn't fit very well to the block and the forces of mounting it was putting odd stresses on the block -> transmitting these directly to the chip -> to the silicon. Screw on the base, part fails, unbolt it, part woks. Redesigned the mount to use rubber grommets and never saw that problem again. (Did toss out the cracked one though.)
Coud be a temperature thing?, Mounting thing?
-jim lee