Rockbox mail archiveSubject: Re: eetimes sourced article on battery fuel-meter prediction methods
Re: eetimes sourced article on battery fuel-meter prediction methods
From: Björn Stenberg <bjorn_at_haxx.se>
Date: Fri, 12 Aug 2005 23:48:04 +0200
nobby nobbs wrote:
> Is it possible to use the dual system proposed in the article for the
> iriver and archos prediction algorythms? (or is a similar system allready
> in place?)
Rockbox' system for "fuel gauge" prediction is (at least for Archos devices) mapping the battery voltage onto a pre-measured battery discharge curve.
The system described in the article includes coulomb-counting, a process where you keep track of the precise amount of current used by the device at each point in time. Thus, you can track how much energy is extracted from the battery and thus how much should remain.
We don't have any way to perfom such tracking, since we cannot measure the current. Our only input data is battery voltage.
Also, the problem they point out with voltage reading is that the measurement will differ greatly if you read the voltage with or without load on the battery (i.e. in standby mode the device says the battery is nearly full, yet when you turn it on the battery is nearly empty.)
Rockbox, on the other hand, doesn't have a standby mode. We always have load on our battery when we read it. And even if the load varies a bit (spinup, cpu boost) the variations are noway near as big as those the artice use in their example (1/3C).
-- Björn _______________________________________________ http://cool.haxx.se/mailman/listinfo/rockboxReceived on 2005-08-12