From: Uwe Freese <mail_at_uwe-freese.de>
Date: Wed, 18 Sep 2002 20:16:47 +0200

Hello,

I had an idea some days ago and because of Ed's mail I'll tell you
what came to my mind:

The battery consumption is more or less predictable.
The AJB consumes xx mA if hdd is [on/off], led backlight is [on/off]
and so on (that can be measured as I already did some weeks ago).

So in the battery thread (that takes the battery voltage four times
per minute now), we can look if hdd is on or off, led is on or
off,..., assume a current and say "ok, one minute with hdd on,...
means that xx mA is taken from the batteries".

When started with a full battery (after a complete charge) and with
knowledge of the battery capacity, we can sum up all the consumptions
and predict how full the batteries really are and how long estimated
running time is.

I know that this is not so easy. There are some things that have to be
considered:

- The battery consumption has to be saved in rtc and loaded after
  turning on the AJB again.
- Find a current value that has to be added when the charger is on
  (ok, only some more tests needed for that).
- If the battery has EMPTY_VOLTAGE, remember which capacity it had and
  save it to rtc. Then use this value for the next time. And tell the
  user which capacity the battery has somewhere in the "info" display.
- The estimation will work good if the user usually makes a complete charge
  and runs the AJB till the battery is empty (because resetting the
  battery status there to 0% / 100%), but won't work well if he ever
  charges some minutes and uses the AJB some minutes (inaccuracy will
  grow over time). So a "plausibility test" is needed with checking
  the voltage (if voltage says that batteries are much more
  empty/full, change the assumed current status).
  
So what do you all think of this idea?

BTW: "Battery monitor" is a "hack" for palm OS and is doing exactly
the same and it works very well!

Bye, Uwe.
Received on 2002-09-18