[Gta04-owner] GTA04 testing status

Vasco Névoa vasco.nevoa at sapo.pt
Tue Jul 5 11:12:07 CEST 2011

Dr. Schaller

Remember the three rules of electronics: Decoupling, decoupling,  
decoupling!! :)
If I learned anything from my experience with GTA02 is that Openmoko  
did not pay enough attention to capacitors, and this basically yielded  
an unreliable device all around.
To this day I cannot operate reliably more than one set of peripherals  
on it at the same time; I have to pick between GPS / GPRS / WIFI /  
sdcard. If there is a mistake to learn from, this is it.

The other point where it failed completely was testing: had it waited  
a few months longer and tested it more thoroughly, the history of the  
Neo would have been quite different.
No amount of software hacking can compensate for poorly produced hardware.

Don't let "time-to-market" pressure get to you. Keep doing the job  
right, release only when absolutely confident. We have faith in your  
May the force be with you (and your great reports with us)!!


Citando "Dr. H. Nikolaus Schaller" <hns at goldelico.com>:

> And just a few notes from the lab...
> 1. Soldering issues
> We have four boards (two in "module" configuration
> and two in full GTA04 configuration). Out of these
> four only one did not show any soldering issues. I.e.
> 25% yield. This is a little low and we have sent one board
> to X-Ray inspection to better understand this problem.
> The issues appear to be short circuits of power supply
> lines to GND. Two times under the WLAN/BT module
> and once under the CPU.
> 2. Modem / Internal USB
> it took 3 days experimenting with the kernel drivers
> to find out that the USB3322 controller chip did
> not respond at all. Because it was not optimally
> soldered and had no power supply.
> It took so long to find since it is a subsystem hidden almost
> completely from the kernel. Because it is connected
> through ULPI to the EHCI controller. And the kernel
> only configures and handles the EHCI controller. And there is
> no error message - the interface simply appears to
> have no devices plugged in. After digging deeply into
> data sheets and kernel driver code we were able to
> add code to directly read the ULPI registers - and
> it turned out there was nothing to read (all 0x00).
> After heating the chip for reflow it now works well and
> we can see USB test signals at the pads where the
> UMTS modem will be connected.
> Next step will be to solder a USB socket and connect
> some device and check if it is recognized with lsusb.
> 3. Sound (ALSA)
> we have the headset working and the Earpiece.
> Sound quality appears to be very clear.
> And even the Vibrator motor can be enabled
> to "play" the sound.
> Where we have found a problem is the handsfree/
> music speaker amplifier. There, we have soldered
> wrong capacitors. This has the effect that we can
> see some signals on the oscilloscope if we disconnect
> the speaker. But it is not audible at all. Next step is
> to replace these 4 capacitors.
> Recording (internal microphone or headset) hasn't
> been evaluated yet.
> So hardware validation is a slow process which sometimes
> reminds me of the old ages where you had to punch
> your software onto a deck of punch cards. Deliver it
> in the evening to the computer center which worked
> through them over night. And only next morning you got
> the syntax errors from your FORTRAN compiler. Then
> you could fix something and had to wait until next morning.
> Our university did still run such a system so that we still
> could experience that - even in the ages of Unix and PCs :)
> Now, we would like to find the most critical problems
> before production of the Early Adopter units starts.
> I hope you can still wait some more days...
> Nikolaus
> _______________________________________________
> Gta04-owner mailing list
> Gta04-owner at goldelico.com
> http://lists.goldelico.com/mailman/listinfo/gta04-owner

More information about the Gta04-owner mailing list