[Openec] RFC: One-Wire-Filesystem (owfs) for charging battery on XO and openec, (long)

Frieder Ferlemann frieder.ferlemann at web.de
Tue Sep 4 14:08:23 EDT 2007


Hi,

Albert Cahalan schrieb:
>> (does waking the CPU change the charge current into the
>> battery?)
> 
> Yes, at least for a B2. Charging takes a long time when
> the laptop is in use.

I see. Eventually the low speed charging mode is not affected,
i.e. delivers about 400mA into the battery regardless
whether the laptop is on/off?

>>> It's nice to charge without booting Linux.
>> (In case of XO it will already booted?)
> 
> I mean this:
> 
> 1. start with XO fully off
> 2. add battery and external power
> 3. let XO charge (do not boot)


Yes. Although for the scheme proposed
the story would have to continue somehow like this
(appending to your points:)

1. start with XO fully off
2. add battery and external power
3. let XO charge (do not boot)   in low speed mode until x Joule
   have been delivered or the internal resistance of the battery
   is below y Ohm or at least z Volt are reached.
4. boot XO and await clever instructions from the charging algorithm
5. react on these instructions


So at some time the EC would boot the XO to ask for clever
instructions because on its own it does not know how to
charge the battery so the maximum cycles/lifetime will be reached.

The "does not know" actually falls into two parts:

a) it does not have the clever charging algorithm,

b) it does not and _cannot_ have the data the clever charging algorithm has.

The data includes the charging history of that specific battery,
the charging histories of an ensemble of batteries (same production
lot), and an electrochemical model of the cells themselves.

Also as linux user space code can use floating point on a large number
of voltage data (for denoising and it might have a good thermal model
of the cells even if high speed charging mode results in varying
current to the battery) things like a sudden dV/dt of a single cell
might be detectable early (and then stored into history).


Summing up I think a "very clever" algorithm in linux user space will
win hands down over an "extremely clever" algorithm on the EC.


Greetings,

Frieder


More information about the Openec mailing list