Once this sort of electronics get inside a box because they go outside, the problem is usually one of keeping them cool, particularly if any sunlight falls on to the box. The electronics will keep itself toasty warm - these chipsets run hot anyway, so I’d anticipate you could very cold before you need to take action. I do high altitude balloon flights with a Pi in a polystyrene foam box at -45℃ outside and as I’ve gone overboard with the duct tape, +30℃ inside.
Temperature wise, it’s probably fine if things have been running.
But loose power for a few hours in winter, and you’re now starting up from a cold-soaked condition.
If it merely “doesn’t perform well” when cold, letting it warm up is enough.
In theory one could have a power resistor and run that for a while before trying to start the concentrator if a temperature sensor reads on the chilly side.
It’s often start-up from extreme cold that is the problem…but if system can limp into power up - even if out of spec/not fully functioning, & then get a reboot/warm start self heating will often have taken care of things…xtals and oscillators can be a problem. I’ve seen telecoms systems in cold climes set to do that automatically (after 30 mins IIRC when they have been off for >few hours) as a matter of course…
Looks like the source code is now available in the main pycom repo - there’s a modified copy of the usual Semtech C code sitting in a subdirectory. (The software runs on whichever esp32 pycom module you plug in, the board doesn’t have it’s own CPU).
And they stuck GPL3 on it, so no reason we can’t play with it. If it works, that all would seem to make it far preferable to the locked-down TTIG.
May have to give this a look, or maybe see if I can get an on-hand ESP32 talking to an ordinary concentrator, since it’s SPI plus a few standard signals (yes, the irony is that this is interesting because it’s inexpensive hardware, but the last thing I can justify is buying yet another concentrator board out of sheer curiosity)