I understand that the motherboard (I'm assuming that's the devboard) does not support temperatures below -25 Celsius. We don't plan to use that anyway, we will have our own power supply. I just need the data on the Carambola itself, regardless of external components. We'll make sure the power supply and all other external components support the needed temperatures.

So could you please clarify the temperature range for the Carambola board? Will it work or will it break at -20 Celsius for a few days?

Thank you,
Razvan

So taking into account what gedass said you're going to have to find a way to keep the carambola warm, or find a hardware solution more suited to your needs (although I guess that'll be quite a bit more expensive).
You can solder an antenna connector to the carambola board (the footprint is already there, there's just no connector soldered to it). For maximum range you'll need custom antenna anyways (a unidirectional antenna will boost the range considerably, although seeing that you're going to build a sensor mesh your antenna should radiate in at least two directions), so you can just as easily attach an antenna cable to that connector and have the WiFi antenna above ground while your hardware is safely placed underground with the batteries.

I'm curious, what kind of bandwidth can you expect from those sattelite uplinks?

For cost reasons, not all sensors will be equipped with sat uplinks - that's where the WiFi comes into play. The ones without sat uplinks will simply upload their data to another sensor that does have the necessary uplink capabilities. This also helps with positioning of the sensors - not all of them will have a clear view of the southern sky to be able to talk to the Thuraya geostationary satellite.

Thuraya GmPRS (the equivalent of GPRS in the GSM world - packet switched data) gives you 15kbps uplink / 60kbps downlink.

I'm still not sure what that -40 minimum temperature range that Gedass indicated was for. As far as I understand from this thread, if I plan to go below -10 Celsius I should consider heating or insulation. Is this correct?

I'm still not sure what that -40 minimum temperature range that Gedass indicated was for. As far as I understand from this thread, if I plan to go below -10 Celsius I should consider heating or insulation. Is this correct?[/quote]

Without additional heating system will start to boot down to - 42-47 oC.
If you really need exactly info about temperatures list I can repeat all tests in our laboratory after 1 week.I am on vacation now.

That's ok Gedass, no rush. We would need the board to boot up and be stable at that temperature for at least 72 hours (no reboots, USB, WiFi and GPIO operational - even with reduced performance).

I'm still a bit concerned by the fact that Ralink only rated their SoC for -10 to +55 Celsius. They must have run tests as well so they've probably encountered problems at temperatures outside that range. I'm pretty sure they would have said -40 to +55 if they could (although I doubt they would want to tell me what went wrong when they tested below -10 to limit their range at that value).

Thank you!

I suppose the temperature limits stated in datasheets are the maximum ensured operating conditions.
Which means if 1 out of 100 chips fails below -10°C they'll all be rated for -10°C.
As a company selling microchips in badges of tens of thousands of chips you don't want your customers to have problems with about a thousand of them, so you'll be super careful what you put in your datasheets.
So maybe you could ask the 8devices.com guys to hand select their boards for your specifications for a small fee or something like that.
I also couldn't imagine the carambola to fail once the temperature drops below -10°C if it's already powered on. The little sucker (sucker in the most respectful meaning of the word) gets quite warm to the touch under normal operation (I think around 30-40°C). But I'm guessing you're only going to boot them up for short periods to conserve power.
I still think carambola will work in this project. If I had a freezer that could reach such low temperatures I'd even put one of my boards to the test
Would be interesting to know its limits.