GPS/SD issue hardware fix vs. software fix
andy at openmoko.com
Mon Oct 20 15:40:31 CEST 2008
-----BEGIN PGP SIGNED MESSAGE-----
Somebody in the thread at some point said:
| Risto H. Kurppa wrote:
|> Some months ago we had this GPS time to first fix issue that was
|> handled with a software fix. However the new Freerunners ship with the
|> hardware fix (you can see a small capacitor in the connectors of SD
|> Is someone able to give us some test results hos much better the hw
|> fix is? TTFF with two FR's where one has and the other one doesn't
|> have the HW fix, using the same software.
| I don't think the hardware fix is any better. With the updated kernel
| and no hardware fix I see TTFF in line with chipset specifications,
| often around 40s, whether I have the SD in or not. The kernel update
| allows the 'fix' to be controlled in some detail, and I posted a script
| that repeatedly cycles through the available settings recording TTFF.
| This shows both the variation in TTFF with the different settings, and
| the variation in TTFF over multiple attempts at different times in the
| same location.
| Having said that I don't think there have been any definitive tests
| posted. I'll have a look to see if I have a test unit with the capacitor
| installed, but I can't promise anything.
Another guy on community list did also try to make a really scientific
comparison at the time, with GTA02 against other GPS devices at the same
location, but something seemed a bit wrong with his procedure.
Generally the soft fix for holding the interface silent when not in use
is going to outperform running the interface with the cap on one line,
and act the same as making it silent with the cap.
Accessing SD during GPS capture will perturb it somewhat with or without
the cap, I would expect you get marginally less disturbance with the cap
if you run the interface. But typically, the perturbation is of an
"insane" character if seen, you suddenly travelled 4km in 1 sec or so
and is then easily chopped out... and this only occasionally when SD is
run at a bad time during getting the fix.
So AIUI the situation is acceptable without the cap on GTA02. In future
devices we will test it with all the SDIO lines slugged with small caps
and will take more care about routing.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
-----END PGP SIGNATURE-----
More information about the support