[OLPC-devel] Battery Charge End Sensing

[John R.]

On 7/13/06, Jim Gettys <jg at laptop.org> wrote:

> I suspect that putting a conventional battery as a bridge would be the
> way to solve this, rather than complicating the power supply in each
> OLPC.

Not on point exactly, but here's a history lesson/data point:

I'm an afficianado of the old TRS-80/Tandy laptops. These devices have
no hard drive, they just dedicated part of their memory to a simple
filesystem.

They have a main set of 4 conventional AA batteries, and an internal
non-user-accessible rechargeable NiCd RAM backup battery. The main
batteries or wall plug keep the RAM backup battery charged all of the
time. When your main battery runs out, which it does every 20 hours of
use, the laptop does not function, but the RAM state is preserved. I
have gone at least 2 months with dead or removed main battery but the
RAM stays alive just fine.

The innovation was that your RAM backup battery is out-of-phase so to
speak from the main battery. It is always ready when you need it, when
your main battery is drained .

Another laptop, the Cambridge Z88 had a similar thing but it used a
cap. So it would only last long enough to swap the battery (or connect
to mains power), sort of like the Palm III devices.

Some of these old laptops used coin batteries which seem to need
replacing more often than the main batteries.

I think the M100/T102 had the best solution. 20 some years later users
are only now starting to need to replace the NiCds.

Such a backup battery might be interesting for suspend-to-RAM on OLPC.
If the user can suspend-to-RAM it removes need to suspend to flash. If
you battery is dead, you would have lots of time before any data loss.
Relying on only one battery means that you lose your suspend state in
a very short period of time unless you interrupt the user and force
suspend operation fairly early.

-- John.