[Gta04-owner] GTA04 power management.

Dr. H. Nikolaus Schaller hns at goldelico.com
Sat Dec 24 12:45:48 CET 2011


Am 24.12.2011 um 11:39 schrieb Rainer Dorsch:

> Neil,
> thanks for the code and sharing the data of your experiments.
> can we do a bottom-up calculation? I.e. build up a spreadsheet like document, 
> which contains the power consumption of every component/subsystem in each of 
> its power mode (if there are more than on and off).

There is a (incomplete) table in section 5.20 on page 51 of the system manual for
the GTA04.

> Can you give a pointer to the OPTION GSM module and why it drains battery 
> although the device is off?

The Option module has a low-power mode where it still is capable to
receive incoming SMS and answer phone calls and wake up the CPU
to do appropriate actions.

This drains between 3 and 10 mA in average (resulting in up to
400 h at 1200 mAh).

In suspend mode this is not only ok but required.

If the phone is completely off, it should be 0 mA. But there is a hardware
bug in the GTA04A3 board that prevents that the modem can be completely
powered off after it was initialized once. Only removing the battery for a
short moment will reset this state.

> Not sure if it helps you with your problem directly, but it would certainly 
> give me (and maybe others are interested too) much better overview of how the 
> power is consumed by the GTA04 (and probably similar numbers hold for other 
> platforms...).
> Thanks,
> Rainer
> Am Samstag, 24. Dezember 2011 schrieb NeilBrown:
>> Hi,
>> I've been playing with power management and have not achieved the results
>> I had hoped for yet and am running out of ideas.  So I thought I would
>> summarise were I am at in the hope that someone can say something helpful.
>> I'm particularly looking at suspend-mode.  While runtime power management
>> (turning the CPU on/off dynamically as needed and adjusting voltage) is
>> attractive it can never save more power than full suspend, so suspend mode
>> will show us the minimum power usage without turning the device off
>> completely.
>> I'm current measuring about 8%/hour battery discharge while in suspend
>> mode which corresponds to approximately 100mA current drain.  When the
>> device is off I lose about  1%/hour which is around 10mA (the precision of
>> these measurements isn't great so there is a lot of rounding here) which
>> is presumably attributed to the OPTION GSM module (and matches what HNS
>> said it would be).

Looks ok.

>> So that suggests around 90mA still being used by something else.

Hm. This reminds me on the power demand of the Boot-ROM mode. I.e. before
U-Boot starts.

>> I've turned off the power supplies for WIFI, BT, GPS antenna, sensors.
>> I've turned off VDD1 and VDD2 to the CPU, and the various power supplies
>> for USB and the battery charger (VINTANA, VINTDIG).  All the leds are off.
>> All that should be left on are VDDIO aka 1V8, and REGEN aka 3v3.  It seems
>> that these need to be on by design.
>> With the GPS off (it is, I checked) and the LEDs off and none of the
>> other peripherals enabled, 3V3 shouldn't be providing much current as
>> discussed previously.  The only things of any significance I can see
>> driven 1V8 are:
>>  - The 26MHz clock.  This seems to be on all the time where I (naively)
>>    would have expected it to be enabled by the CLKEN output from the
>> TWL4030. I assume there is a reason for that?  How much current does it
>> use?

I think this isn't very much but will check in the next days. Maybe this could
be improved for future board revisions since the 26 MHz clock has a separate
enable input which could also be controlled by the CLKEN.

Currently the TPS65950/TWL4030 chip should just gate the clock so that
it is removed from the CPU.

>>  - Some power to the CPU.  If I read it correctly it is only supplying
>>    - DDS - the display subsystem
>>    - POP-DDR and POP_FLASH - the Package on Package memory.
>>    I would expect the DDS to be off (not sure how to check yet) and the

What I don't know for sure is if the code to control the TD028 display is
completely correct. I.e. it may not correctly switch off the LDOs within the
display module.

But you could easily test that by operating the GTA04 without display.

>>    memory to be in self-refresh mode.

Memory in self-refresh could be some reason.

>> Any guesses were that 90mA is going?  I would understand a 10mA draw, but
>> 90mA seems a bit steep for a sleeping device.

One voltage that is not shown in your list is VMMC. The SD card may draw
some current.

>> BTW, only slightly related: it looks like it might be possible to get the
>> power management chip to turn 1v8 and 3v3 off and back on again in
>> response to warm-reset.  So possibly I could use that to get the GPS into
>> a known state if I really need to.

Yes, but I think it is better to check for output at the serial port. It was just
5 lines to add in the U-Boot "gps echo" command. Maybe a little more
complex in the more modular kernel driver model.


More information about the Gta04-owner mailing list