STM32F072 overheating and dying with no apparent reason

[Angel Sanchez Garcia]

Hi.

I'm experiencing a very weird problem with a design, a CAN bus keypad with RGB backlit keys, in which the MCU (STM32F072C8T6) would sometimes overheat to over 150ºC and stop working. This happens some times in first power up of a unit, or after a few hours working.

This design shares many details (mcu reset, clock, power circuits) with another device based on STM32F303 which works flawlessly.

What is really funny is that this only happens in some batches of micros, those marked with CHN factory code have never broken, this were used in prototypes and some repairs, whereas chips marked with MYS or TWN codes do break and it appears at least the MYS chips will always break after some time.

I have checked all signals and everything is at 3.3V or less, also checked the evolution of VDD when connecting power and it shows no signs of overvoltage.

What can be the cause of such a failure? I have run out of ideas to diagnose this problem.

[Uwe Bonnes]

All STM32 I have lost seem to be linked with overvoltage applied.

[Angel Sanchez Garcia]

Hi Uwe.

That is what I thought at the beginning, as the symptoms are similar to some occasions when I accidentally applied 12V to some GPIO pin, but I found no evidence of this.

Thanks for the idea anyway.

[Angel Sanchez Garcia]

Ok, it looks like I finally tracked down the cause of the issue.

The chinese assembler that did our first mass production used a counterfeit RECOM regulator, this part introduces some peaks of up to 7V in the 3.3V output voltage when there is a quick drop in the input voltage, destroying the MCU.