This is confusing. I put the box out Saturday night at 7:30pm, 33*F outside at the time and the box was reading a solid (even throughout) 75*F. The lid was uninsulated for this test, everything else insulated as it appeared in pics on pg2 of this thread.
The temp dropped pretty fast on the BMS which was to be expected since it's on a shelf above the cells, and closest to the uninsulated lid. Definitely doesn't respond quickly (possibly at all) to the warmth rising from the bottom.
At 5am it was 28*F outside and the probe (wedged between cells, beneath the BMS' platform) was 64*F, the BMS was 56.3*F.
Thing is, for all I know, it could have been the same temperature in there had I applied no heating at all.
The plate has been 80*F +/- 0.2* the entire time, still is - the thing has been sitting outside since Saturday evening. Obviously soaking up some sun in the day time which isn't helping my testing. I don't own a chest freezer and this thing would shatter the glass shelf of my refrigerator, probably collapse it even, so I think I'll just have to do more testing in the wild. If I get an alert that the probe or the BMS fell below 45*F I'll make adjustments.
So I think I've been maintaining the plate at 80*F for 39 or 40 hours now and have used 32 amp hours powering it + router + microcontroller + a screen on a raspberry pi (didn't disconnect before throwing it all in the can for testing). It's pretty much used 4-10 watts the entire time. Even when the BMS was 92*F yesterday from the sun blasting the lid, the plate chugged on at 4-10 watts (probe was like 74*F at the time)
80*F at the plate is obviously not a temperature I'd maintain in real life, but I'm pretty sure if I set it to maintain 50*F the bms and probe will drop too low on cold nights. I'll see what difference lid insulation makes... probably somewhat significant. I just have to be certain whatever I put up there doesn't fall off, like adhesive failure, especially if I use metal lined stuff.
None of this would be an issue if I just made the plate heat based off the probe+bms average temperature, but that would end up with the plate getting far too warm. Tricky situation that I can't wrap my brain around programmatically to resolve. As in how to write sound logic that operates the plate on BMS+probe temps, but tapers off if the plate itself gets too warm.
Convinced this will work but I wish my BMS was enclosed in the battery area instead of on a shelf above it. And TL;DR because I'm at the office and don't want to start working