Battery charging (was Re: Daily use of Neo1973)

[Shawn Rutledge]

On Wed, Feb 20, 2008 at 4:50 PM, Marc Bantle <openmoko at rcie.de> wrote:
> Torsten Schlabach schrieb:
>
> > power management is that broken that even if I leave the phone plugged
>  > in to the charger on the desk, it seems to use more power than it gets
>  > and dies.
>  That I observed aswell. From what was said earlier on this list, Neo
>  consumes up to 500mA under certain conditions, which doesn't leave much
>  for charging.

Yes I finally figured out a few weeks ago that my powered hub cannot
supply enough current, so it has to be plugged directly into a
motherboard USB connector.  Even then the actual current going into
the battery is relatively slight because the phone is taking so much
current to keep running.

But after leaving it plugged in (and running) for a couple weeks, the
strange thing is that /sys/devices/platform/.../chgcur still gives me
a largish positive number (> 1000), which makes me wonder if it also
doesn't know when to quit charging the battery.  This is pretty bad
for most batteries, and LiIon batteries are the least tolerant of it.
The /sys/.../batt_voltage is also a bit over 4 volts, which seems high
too.

Isn't the PMU supposed to take care of the battery autonomously?  Can
it be configured wrong?

My dustbuster for example died after a year or so because the charger
has no control circuits at all (just a transformer and rectifier) and
the "docking station" encourages the user to keep it on charge all the
time.  (I didn't manage to convince my wife to charge it only when it
was dead, at least not from day 1, so that was the result.)  But
NiCads can take that kind of abuse for a while.  LiIon's cannot.

(Now think how many dustbusters are being sold.  It's very
irresponsible engineering, which ought to be illegal because of the
pollution it causes.)

>  Another thing I observed: Even with usb-network running, there's no
>  guarantee, the device is charging. The green fill level on the battery
>  panel will show up then and
>  /sys/devices/platform/s3c2410-i2c/i2c-adapter/i2c-0/0-0008/chgcur is
>  drifting around between negative an 17000. So obviously charging is
>  turned off. When working with ssh shell you may not even notice.

I don't think I've seen that problem.