Werner Almesberger werner at openmoko.org
Tue Apr 14 06:11:28 CEST 2009

Steve Mosher wrote:
>   I guess to start I would have to release the A8 gerber to the  
> community. If we choose this path I'd work with DR. N to get some sort  
> of legal status for the organization...

By the way, some form of organization to look at may be the Consumer
Electronics Linux Forum (CELF). There, companies with an interest
in Consumer Electronics, such as Sony, Philips, etc., get together
and coordinate and run joint projects.

> Long ago Wolfy and I met with the Atheros guys.. talked about 6002.
> seemed the way to go, cant recall whats the production status was.

For the last eight months or so, Atheros have urged us to switch to
the AR6002. Production status should be good - already when I was
working on making AR6001 work with SPI, Atheros indicated that AR6002
modules were available from at least one source.

What I don't know is if Atheros' internal development process also
changed from the AR6001 to the AR6002. And of course, following
Atheros' urging and switching to the AR6002 in GTA02 wouldn't really
have solved our problems. In fact, it would have made things a bit
worse, with then two modules and stacks to support.

> Evaluate for me the difficluties in
> GTA02 -G +E... using the edge module from gta03.

Hmm, as far as I know, there has only been light testing of the
communication functions, but the results were encouraging so far.
I think Mickey did most of it.

In terms of electrical interfaces, it's almost identical to the
Calypso, with analog audio and UART for control and data

One difference is that the MC75 is really a 3 V design while our CPUs
(both 6410 and 2442) use 3.3 V for this sort of I/O. Logic levels are
compatible, except that a 3.3 V output from the CPU would exceed the
3.05 V maximum the MC75 allows. There's a 1 k resistor on TX, which
may or may not be sufficient, but RTS is driven directly from a 3.3 V
domain from the CPU. So this would either need a level shifter or
further clarification from EE or our supplier.

Similarly, IGT can rise up to the battery voltage and thus ought to
go through a transistor, not directly to the CPU. But all these are
small issues.

I've heard rumors about thermal issues but they were fervently denied
on further inquiry. Might have been a mis-interpretation of what the
data sheet says.

One rather annoying bit is the placement of the module's I/O connector.
The mating piece has to extend beyond the limiting plane of the module,
so this means that there's a ~1.2 mm gap between module and PCB.

Given that the module has its own PCB, plus the gap, it's about 2 mm
thicker than the Calypso. We could probably just cut a hole into the
main PCB, though. Maybe add a soldered-on FPC with the connector so
that the gap disappears as well.

In the x and y direction, the MC75 is slightly bigger than the
Calypso is wide. So the hole in the case would have to be reduced by
about 2 mm if putting the module there.

A more feasible choice would be to make the electronics currently
under the main can (reduced after glamectomy, NORicide, and
deamplification) swap places with the Calypso area. The MC75 would
then fit quite comfortably, provided that we can make a hole for it
(see above). Interaction with GPS and uSD would have to be examined.

- Werner

More information about the Gta03 mailing list