Question on A5 power routing (was: Re: A5 Power routing "minor issues" for some modules)

Michael Shiloh michael at
Tue Feb 26 21:09:28 CET 2008

I can see using the GTA02 being used in research labs or in product 
development labs as an embedded and connected controller for 
prototyping, using the USB port in host mode to interface to arbitrary 
external electronics. In this case I would want to power the GTA02 via 
USB, with the battery installed to satisfy instantaneous current draw.

I realize that this is unorthodox. This is a prototype.

Assuming the user is comfortable hacking at driver code in order to 
enable this, is there anything in the hardware which prevents accepting 
power from the USB port in host mode, powering the GTA02 (except for 
those modules you mention below) and charging the battery at the same time?

Would any of the changes proposed for A6 prevent this?


Andy Green wrote:
> Hash: SHA1
> Hi folks -
> Thisafternoon I noticed that
>  - vibrator
>  - USB Host function (plug USB device into gta02)
>  - whole GSM subsystem
> are all powered specifically from Vb, which can be absent or unusably
> reduced during operation without a battery due to charger state machine
> actions.  Every now and again the charger wants to stop charging and
> make a measurement of the battery's own voltage.... no battery in there
> means no usable Vb then.  If we don't run the charger, there won't be
> anything put on Vb at all in the USB-no-battery case.
> It's not a huge deal though because you can run from USB with a battery
> in there okay, and it will be powered from the USB 99% of the time,
> including charging the battery.
> A solution can be to move these guys to take power from the VB_SYS rail
> which is the output of the power source summing stuff.  The chip has a
> 2.2A limit here on the chip from the battery to VB_SYS.
> However I talked it over with Tony and he remembers bad things happening
> with a similar idea for a switched GSM power rail on GTA02 A2, we dug
> one out and it used a TPS2023...
> It is rated quite similar to the Vb <-> VB_SYS switch in the pcf50633
> (also specified for 2.2A max) and likely has a similar Rdson for the
> switches (33mR for the TPS2023).  It's really not clear why that
> solution made trouble, the standard reasons would be insufficient caps
> and maybe too long a loop for the power, but eyeballing the board it's
> not too far away from the GSM stuff (nor is the pcf50633 in the GTA02-A5
> case), but that doesn't say anything about the routing.
> - -Andy
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Fedora -
> iD8DBQFHxAapOjLpvpq7dMoRAiMgAJ9UYN2NbyCVvDpwDuD4VobiTl5J6QCcCSnu
> sWAv1bKuKcswb0WC2L5pQsk=
> =X6xM

More information about the openmoko-kernel mailing list