[Tinkerphones] Status GTA04A5 and miscellaneous

H. Nikolaus Schaller hns at goldelico.com
Wed Jul 27 10:59:13 CEST 2016

I have two small updates.

> Am 26.07.2016 um 11:20 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
> Hi,
> there isn't much to report at the moment about the GTA04A5.
> We are still waiting for all components to arrive and the PCBs
> to be produced. Let's hope that no news is good news and
> cross fingers that it happens soon.

I have got a feedback that all components should arrive by
August 22th.

Except the headset jack. This is currently not in stock at the
manufacturer or distributors. Estimated delivery in October :(

Well, we don't want to wait that long and it looks as if there are
brokers who can get them for us.

> On the software side we have published letux kernel 4.7.0:
> 	http://download.goldelico.com/letux-kernel/letux-4.7/
> It fixes the hdq driver so that we can again read the fuel gauge and
> battery state of charge. This is essential to be able to measure
> suspend current.
> Unfortunately we are still struggling with the charging system.
> The symptoms are quite puzzling. On some devices charging
> seems to work. While on others it simply fails, does not recognize
> cable insertion/unplugging and the ethernet gadget fails as well.
> But installing our letux 4.3 kernel on the same SD card and rebooting
> makes it work.
> What the real mystery is: there is no significant code difference in
> the charger driver between 4.3 and 4.4-rc (where it fails for the
> first time). So the difference lies somewhere outside the charger
> driver but we have not yet found where.

Thanks to the help from Andreas, we have found what is going wrong
and developed a partial solution. The problem is that the charger driver
assumes that the usb driver enables the OTG interface. This includes
a prescaler for the VBUS measurement. This should become enabled
if a cable is inserted.

But it isn't.

This has the following consequences for the charger:
* VBUS virtually remains at 0mV (although available)
* the automatic charging mode detects an illegal state and turns off after some milliseconds
* the measurement for the battery NTC is turned of as well and hence we see only 56°C

We have developed a hack for this (enable the prescaler before starting
charging). This makes charging (and interestingly also the ethernet gadget)
work again, so that we can better debug other issues.

It does not solve the underlaying problem of course, where we also
know that during suspend the USB driver does not turn off correctly.
Therefore we are not yet back down to the 25mA suspend current
we did have in the 3.7 kernel.

If you are interested in the patch it is here:


And the discussions:



More information about the Community mailing list