Management of Power Management ;)
sal_electronics at hotmail.com
Sun Aug 23 18:59:22 CEST 2009
Paul Fertser wrote:
>> 5. SHR suspend, GSM on: 0.08A
> This one looks strange, guys measured it to be ~45mA with uSD inserted and
> ~25mA without.
>> 6. Neo1973 shutdown from SHR power menu: 0.03A (BAD. I thing the GSM is
>> still on)
> Yes, it's still on, known problem. We're waiting for Mickey's fix. The
> problem is that frameworkd "forgets" to turn the modem off (gta02 has a
> hardware switch but for gta01 software needs to write AT at POFF several times
> to reliably turn the modem off). And it still looks too much, the expected
> modem current is about 8mA.
Now I think into addition of the problem with uSD drawing ~20mA there is
one more problem. I still can not account for the ~0.05A (0.08A - 0.03A)
in suspend mode. To me looks like other hardware is powered in suspend.
I don't fully understand yet how the phone works, but I would like to
think that everything should be turned off in suspend and the GSM should
be in stand-by.
In addition the 0.03A drawn by the GSM is kinda high. We have a 1200mAh
battery if the phone stays in stand-by all the time it will run out of
power in 40h. My old Palm can last much longer in stand-by in
comparison. I am wondering if we can make the GSM to draw less power
while the GSM is on and registered?!
>> The max charging current I measured is 0.32A with 50% discharged battery
>> and SHR in suspend mode.
> Considering the limit is 500mA and suspend current according to your
> measurements is 0.08, it looks like 100mA went nowhere. Weird.
Well I think because the battery was at 50% charged the PMU chip was not
using the full current to charge it. I need to read the specifications
and understand the logic of the pcf50606 first. Other issue here could
be that my USB port could not provide the full 500mA. I need to build
current probe for the USB port and measure the current there also.
>> The charge current after I disconnected the battery for 20s is as
>> expected 0.10A. One problem I notice is that u-boot 1.3.1 will power on
>> the phone as soon the battery voltage rises. So if you have dippy
>> discharged battery the boot might fail and device will hang and draw
>> residual current and it will drain the battery even dipper. I believe
>> the old u-boot never powered on the phone automatically.
> It's supposed that all charging should be done with Linux running, at least
> that was the general idea at the time of development. The system should
> ensure that it will never discharge the battery too much. Well, easier said
> than done.
> BTW, _never_ever_ take out the battery from gta01 running on USB power. It
> will destroy it for sure if you don't pull of the cord immediately.
Opps. I think I have done this many times! I hope I have not burn
something already! What part of the phone will be damaged in this case?
Also I have found out there is small backup battery connected to the
pcf50606, because I kept my phone for very long time with no power it
most likely is drained completely. What would be the symptoms for this
problem? I suspect that backup battery preserves some kind of RTC and
the state of the pcf50606.
>> this problems. I will need some help setting up my development
>> environment. Currently I am running Ubuntu 8.04LTS.
> For kernel hacking all you need is compiler, git and text editor. So, i'd
> recommend to install emdebian (i think ubuntu is compatible with it), with
> the help of emdebian install the toolchain, then git clone the sources from
> git.openmoko.org and start hacking
If I am understanding everything correctly so far, I should use the
kernel described here:
More information about the devel