We have a design with an UCD9240 and three single-phase power rails. We have found that after the system warms-up (UCD9240 case temperature with a thermocouple attached ~65°C) we get semi-random rail failures.
Some rails will come up, but at least one of the rails will fail to start. This shows a very strong temperature dependency, 0 failures below 55°C or so, ~ 50% failures around 65°C and ~100% failures around 70°C. We see multiple rail faults being logged, but looking at another system we notice that these might happen on start-up or shut down.
We are trying to debug the system, and we applied external heat to a separate, known-good UCD on a different system with four rails (the ones in the failing system have a known calibration issue). We started getting failures at ~75°C (unfortunately we seem to have damaged two PTD08A006W in the process, even though their logged internal temperatures never exceeded the indicated 80°C maximum).
According to the datasheet I should be able to approach or exceed 80°C without too much problem.
Is there something in the set-up that could be causing this problem? How can we avoid or recover from these rail failures?