After spending the week up in Seattle ingressing and lighthousing we got back into the board for some battery performance testing.
I used my custom VISA software and the Rigol RM3058E DMM , added a logging and simple alarm mode to it so it'll log the data every so often and let me know when it hits a certain voltage.
Graphing it out, we haven't investigated what happened after 20 minutes, it might have been after we disconnected the programmer.
We made some changes to the board to add some test pins, and here is a useful warning i'll forget again, if you've left a project for a while, make a change and things don't work as you recall, check first if what you thought is meant to happen, actually did. Otherwise you'll be biased by the it did work, we soldered some stuff, it no longer works,. In this case the USB popped up saying unrecognised device, i had changed small elements of the code but nothing that should break it, went back to a known working firmware ( add the hex files and tag source control so you know working firmwares, add text/comments to describe state and what working means)
Anyway after poking for an hour and swapping things out, and a xtal that didn't boot once, and i tried a firmware that i'd written from scratch, and it showed the USB OK, so software issue, checked against source control and see what was changed , nothing major. it was late and we looked at other stuff. Next morning I wake up have coffee and take another look, wait a tick didn't this firmware not actually have USB host implemented .... yep... derp moment. so it never had it and since it'd been sitting soak testing and i always have at least two unrecognised USB devices connected to my dev PC, bias again.
So the USB is fine, some wasted time. Also ra ! gnd is an evil eagle command, i need to stop using it, its bitten me twice now.
the new battery test just ended at 2017-02-12 13:37:09 2.979585V (nice timestamp!) and nice smooth curve