I had an elusive problem show up after months of testing on three units in parallel. As luck would have it, only at a customer (Beta) test site. CPU halting, letting go of IO pins, and not resetting despite using watchdog timer(s).
Power supplies were up. Log of power supplies (by Teensy code) showed no problems. Turns out the pin socket strip I was plugging the T4.1 into was barely connecting to the +5V power pin. Normally fine, but sometimes dropping out - but you'd have to continuously measure right on the Teensy board itself to see that.
Could also be made to fail by pushing sideways on the Teensy, or low humidity (put it in a test chamber with a dehumidifier! The first clue, but, not ESD after all). Once the +5 input dropped enough to shut down the CPU, there must have been enough 10's of mW of heat dissipation in that pin connection resistance to hold it in the bad connection state, because power cycling caused it to restart - I surmise it cooled the connection down enough to reconnect. Also thus looking like a software or WDT problem.
The part used was Samtec SLW-124-01-G-S. These looked better for the standard PJRC square pins, but no, they have a single fork with one pair of tines, despite being gold plated beryllium copper they spread after a few insertions/removals.
Went back to this part which we've used for years: Mill-Max 801-43-024-10-001000.
Here are micrographs of both of these with Teensy pins. I'm curious if anyone else has a favorite model of socket for Teensys?
Power supplies were up. Log of power supplies (by Teensy code) showed no problems. Turns out the pin socket strip I was plugging the T4.1 into was barely connecting to the +5V power pin. Normally fine, but sometimes dropping out - but you'd have to continuously measure right on the Teensy board itself to see that.
Could also be made to fail by pushing sideways on the Teensy, or low humidity (put it in a test chamber with a dehumidifier! The first clue, but, not ESD after all). Once the +5 input dropped enough to shut down the CPU, there must have been enough 10's of mW of heat dissipation in that pin connection resistance to hold it in the bad connection state, because power cycling caused it to restart - I surmise it cooled the connection down enough to reconnect. Also thus looking like a software or WDT problem.
The part used was Samtec SLW-124-01-G-S. These looked better for the standard PJRC square pins, but no, they have a single fork with one pair of tines, despite being gold plated beryllium copper they spread after a few insertions/removals.
Went back to this part which we've used for years: Mill-Max 801-43-024-10-001000.
Here are micrographs of both of these with Teensy pins. I'm curious if anyone else has a favorite model of socket for Teensys?
