Bug #1024 (oscillating re-camping), the latest news.

[Dieter Spaar]

Hello,

Timo Juhani Lindfors wrote:
> Joerg Reisenweber <joerg at openmoko.org> writes:
>   
>> Suggestion to FSO to improve sleep=2 workaround: Use temp reading (e.g. from 
>> battery, or even from calypso if we put that to operation) to switch between 
>> sleep=2 and sleep=4
>>     
>
> Does calypso have a temperature sensor? That'd be very cool. The
> battery temperature sensor can not be reliably used to measure outside
> temperature since the battery heats itself a lot.
>   

There is a temperature sensor in the GSM transceiver chip. However I
don't  think that this sensor is supposed to measure the temperature of the
environment, most certainly the purpose is to adjust some RF parameters
depending on the transceiver temperature (e.g. RF power). And don't forget
that the GSM modem is shielded, this probably cause the temperature
inside the modem to be different or follow the outside temperature only
slowly.

Additionally this sensor is not enabled in the GSM firmware, it always 
reports
0. We don't know why it is disabled, but lets assume its intentional. If we
enable the sensor, we might break something else because suddenly the
temperature is no longer 0. We don't know where and if the temperature
value is used inside the GSM firmware because we don't have the source
code for the relevant parts. The sensor is queried by a regular task of the
GSM firmware so I also don't think its a good a idea to query the sensor
on  our own and might get in conflict with this task. It might be possible
to make the raw value of the sensor available somehow. Besides the
required effort I don't think this would help a lot for bug #1024. Till
now it looks like only the Calypso chip itself is the part which is
responsible for the temperature related effect of bug #1024. If the
GSM modem is sending, I expect the transceiver chip to get warmer
than the other chips so the sensor value would not necessary reflect
the temperature which affects bug #1024. And of course so far it
seems that there is no exact temperature when bug #1024 begins
to show up, the phones behave different.

Instead of messing around with all of this and making things more
complicated, I thing the current solution to detect re-camping and
enable AT%SLEEP=2 is the better approach (maybe adjust the
algorithm slightly).

Best regards,
  Dieter