>Since this is a battery-powered device, with a total draw approaching
>30mA (due to the need for some heavily compute-intensive code, and
>thus a high clock speed), I am either going to use disposable batteries,
>or rechargeables.  I would like to use rechargeables, but most of
>the batteries I'm familiar with (lead acid, NiCad, NiMh) off-gas
>too much during charging to reliably pot, which means I'm stuck with a bulky
>(and expensive) 1 ATA housing.

None of the nicads are supposed to outgass, they only vent if there's an
overcharge, but you have a bigger problem. The pressure will force
helium-o2 into the cells, and then when you depress, it will vent he-O2,
and water vapour, which will kill the cells after some number of cycles.

Lead acid can be sealed, but the chamber will eventually have to be vented.
A catalyst sponge might serve to convert the H2-O2 into water in a
controlled manner.

I think you might be stuck with Li-Ion in this environment, and you'll
still have to look at the problem of gas getting into the cells and then
carrying something else out with it. Thionyl Chloride is NOT something you
want venting into the cabin!