GSM modem doesn't power off when Neo does

[Erin Yueh]

Hi List,

About GSM modem power off issue, we cannot solve this problem coz of 
hardware limitations. It's about some components and their connections 
like battery, PMU, CPU and GSM modem. In GTA01 devices, the battery is 
connected with PMU and GSM modem as below.

         ------> PMU -----> CPU (SC2410)
Battery |
         ------> GSM modem

Therefore, we cannot control the power status of GSM modem by PMU 
directly. Now, we are using an AT command 'AT at POFF' to power off the 
modem, when we shut down the device. Then, it can solov a battery 
problem bug#1003 'GSM modem is not powered down when Linux is shut 
down'. Frankly speaking, it's a stupid way to power off the modem. For 
example, if gsmd is dead, then we cannot send an AT command to modem. 
Also, we can only send AT at POFF this command 'once'.

We are deeply apologetic about this hardware design problem on GTA01 
devices. And, we modify these connections in GTA02 as below.

                      -----> CPU (SC2410)
Battery ----->  PMU |
                      -----> GSM modem

In GTA02 devices, we can control the power status of GSM modem from PMU 
and don't need this AT at POFF command to modem any more. We can just pull 
a pin to high or low and it can control the power status of GSM modem.

We feel very sorry about this gsm modem power off issue and try to 
improve it in our next device.

Best Regards,
Erin







Shawn Rutledge wrote:
> On Feb 15, 2008 10:46 PM, Mike (mwester) <mwester at dls.net> wrote:
>> Mike (mwester) wrote:
>>> I think it's time for another attempt to address a long-standing
>>> problem with the Neo and the current software.  Specifically, the way
>>> it works right now, no matter how you shut down the GTA01, the GSM
>>> modem is left powered up and running (it's wired up internally in such
>>> a fashion that even when the kernel powers off the CPU and all else
>>> that it can on the circuit board, the GSM modem continues to run).
>> [snip]
>>
>> Well, that was interesting.  Not much response, so I take it that this
>> isn't a particularly important item for most people.
>>
>> I'll just drop this then; my proposed solution didn't get any good
>> reviews first time around, and based on the lack of interest, I don't
>> see any real point in re-opening that controversy.  However, when
>> openmoko gets around to resolving this problem, I'll be interested to
>> see how the "official" solution is implemented.  Until then, keep those
>> Neos plugged in!
> 
> Of course it's important.  Maybe just a case that nobody knows who has
> responsibility for this, but everybody's hoping that somebody is on
> the case.
> 
> Where do you want to put that AT at POFF command?  gsmd should send it
> whenever it receives certain signals (like SIGTERM etc.) right?  Can
> you just go ahead and commit this change?
>