Joe  wrote:
> Lithium cells have a linear-ish discharge curve, from 4.2V down to
> around 3V. Any sensibly designed lithium battery will cut off its
> output at the chosen lower bound, because if it completely discharges,
> it's dead forever. Not a 'steep' decline, a 'fall off the wall' decline.
>
> It's a good idea to have some independent means of anticipating this
> point.

But lithium ion cells don't work for this application.  We need 5V.
One cell is too little, two cells is too much.
One cell would start at 4.2V which is too low to reliably run a Pi.
Two cells would start at 8.4V which is dangerously high for the Pi, and the
proposed 5V cutoff would only come in below 2.5V per cell which is bad for
the cells.

As mentioned 6V lead acid doesn't work for a starting battery - it might be
OK for a 'leisure' battery that is fine with deep discharge, but again we're
on a part of the curve where the battery is absolutely flat and declining
fast.

Possibly LiFePO4 might be better, but a pair of cells at 3.6V fully charged
is still too dangerous for the Pi.

4x alkaline cells might just about manage it - 1.6V per cell when new, ie
6.4V total, which is a bit dangerous but maybe OK.  A gradual decline - I
don't know where the Pi low voltage detection is but flat would be about
1.1V per cell or 4.4V total which is probably detectable.  So if you don't
mind throwing away cells at 20% capacity it might just work.

Theo

--- SoupGate-Win32 v1.05