I had written a short program to catalog a pi's parameters because I have 67 pies in total and was finding it hard to keep track of them, eg if I needed a 4 core armv6l for a project, which one to pick...? So for the first time in an age I actually plugged in every one of my pies to run this utility and save the output of various commands indexed under the serial number of each pi.
Out of 67 boards, 2 had corrupt SDs which I tossed and replaced with newly built Noobs.
However one Pi Zero W behaved badly: on booting it immediately showed a thermometer icon at the top of the screen and shut down, and after that would not boot again. A finger touch suggested gross overheating, so I waited until it cooled down and then made this video with a thermal camera:
The temperature jumps up to 80C in one minute. And it does this regardless of whether there is an SD inserted or not. Googling for Pi Zero W overheating I found quite a few reports although the general consensus was that it was always the users fault for whatever reason, eg shorting/blowing up input pins.
Not the case with this one. It had been put away working, and failed when brought back out and tested in a test rig that had successfully tested 40-odd pies and 20 zeroes. (I suppose being fair there is a very small chance that it was due to static discharge (I generally don't bother with a grounding strap when handling pies, thought they were more robust than that) but I didn't notice any static and with 62% current humidity it seems unlikely there would be any.)
I'm not asking for help debugging this - I might as well toss it - but I'm posting this for other people who have Pi Zero W thermal failures in case they have similar problems and it turns out to be something systematic with the chips.