Analysis of OUTPUT high GPIO in suspend / DL_GSM

joerg at joerg at
Sat May 10 16:00:18 CEST 2008

Am Sa  23. Februar 2008 schrieb Andy Green:
> > [   58.325000] GPJ06: OUTPUT      1 <== DL_GSM... what does it do? ***
> DL_GSM controls a couple of tristate drivers via an active-low nOE that
> are to do with communication to the GSM side as part of an elaborate
> scheme that looks like it can in the fullness of time just be replaced
> by DL_GSM and one other IO under a FIQ GPIO UART implementation,
> removing R4403, U4401 and doing it all from the CPU side.
> R4403 (100K to 0V connected to DL_GSM) means we burn 33uA / 108uW by
> having OUTPUT 1 in suspend ... hey it all adds up.  However setting it
> OUTPUT 0 or allowing it to float so R4403 controls it means this dual
> tristate driver (always-on BTW) is enabled to drive unknown levels into
> bad places.
> -Andy

Make R4403 NC!
If we need R4403 for fab, kill it after flashing GSM-module.

Booting with headset connected, uBoot gives those bad deafening chirps and 
clicks. (don't do this at home!! Will OM pay for my hearing aid? For any 
others'? ;)
The tristate drivers of U4401 are active by default when CPU is reset, due to 
R4403 pullDOWN(1).
I *strongly* recommend to make R4403 NC. So we have a chance GPIO DL_GSM is 
high Z at CPU-reset and U4401 blocks TX-IRDA to aerjack path.
This should be a pullUP, no pullDOWN!


ps: there are other very strange issues like 
O- after boot, left earpiece(!) is microphoning to right earpiece when tapping 
it with a finger (no that's not the mic and ALSA settings! That's left 
O- Sound only by left earpiece? 
- Left earpiece *gets hot* during boot???? 
Will have to check more detail on these issues.
O-(!!) [hw] Using headset for a call is completely useless, GSM RF feeds into 
mic _really_load_ so you don't hear anything else (missing RF block measures 
on earjack receptacle). Maybe can fix with a ferrite block near jack on 
headset cable? (compare charger for what I'm talking about)
O- [sw?] Headphone (earpiece_l) and device speaker SPK4102 are mutually 
exclusive. (Alas this is NOT headset and _device_earpiece_ are alternative) 
We can not disable headphone, so we mustn't activate speaker(-mode) of 
U4101-amp when headset connected. What about ringtone playback while headset 
connected, then?

used device: Prototype GTA02-113.

More information about the openmoko-kernel mailing list