Thanks for the replies everyone. As for adding more detail to my inquiry, I currently have a simple PCB containing PIC, CR lithium battery, thermistor and crystal. This PCB keeps time and records temperature threshold crossings(time and duration). I've used it down to -40 without problems, but when asked to ship it in dry ice it did not work. The ram would clear for whatever reason.
The idea of using an insulated container is interesting to me. My PCB is 3cm by 4cm. If I encase it in styrofoam 1cm thick and assuming I could control the internal temperature to above -40*C to keep all the parts within spec, and the ambient temp is -78.5*C for a temp diff of
38.5*C, k value of 0.01 for styrofoam and total surface area of 24cm^2 ( counting both sides). I get a total power disipation of about 0.1 watts. I used the formula H=k
*A*dT/L, A-area, L-thickness, dT-temp diff, and k-thermal conductivity. The power consumption of my product is easily 1000 time less than that, so I would essentially need to disipate 0.1W through a resistor in order to keep the temp up. But I don't know enough about batteries to know how much heat they dissipate as they deliver the current. This side effect of batteries would obviously help me here. Can anyone give me an approximation for this?
Also does any one have any design tips for the enclosure? For instance is stryofoam the best? I'm not too mechanical...
Since I currently use a 50mAh battery I would need to oversize this by at least a factor of 10 to get a reasonable amount of logging time. The other alternative, to place the thermistor on leads and place it in the shipment, while the electronics remain outside. This is obviously the best scenario, but I'm not sure if I can do it.