Hmmm, wouldn't the most efficient way to get power into the device then to fix the regulator to 3.7V, and feed the power into the battery terminals? Maybe even build a (almost) hollow "fake battery" with cables out to your LiPoly battery. Of course, you really want to make sure the connection is stable else suffering sudden, complete, and filesystem-chewing poweroff.

Else I would agree, having a full battery should make the tablet only take power from the charger input, so overall, your idea is probably pretty good.

I really wouldn't even think of doing that.. fiddling on that level is for RC enthusiasts, and they're prepared to blow up the battery now and then, so they wear protective gear. It's safer and good enough to just go through the intelligent charging circuitry of the device itself.

What he ^ said...

Oh, I think I was misunderstood. I meant to feed the 3.7V directly into the tablet, not the battery. In fact, no battery would be involved except for the external one: feed its output through your 3.7V regulator into the contacts inside your tablet, bypassing the internal battery (in fact, keep it somewhere else). (Yes, I realise it's still a risky stupid idea :-) but no battery charging would be involved.

Though I don't know what the third terminal of the battery does, or how the tablet uses it. Current sensor? Temperature sensor?

Although I do like the idea of pushing efficiency to the limit, I don't think I want to play guinea pig on this one. I'm pretty happy with the efficiency I'm getting at the moment, but feel free. Also, this involves taking out the battery, which means rebooting, which is something I do only very rarely on my computers and don't like at all.

And no, I don't know what the third terminal does either, part of the reason why I'm passing.