Meddling with GPS

Andy Green andy at
Sun Jul 13 18:30:58 CEST 2008

Hash: SHA1

Hi folks -

Spent the afternoon meddling with GPS about this highly variable
behaviours it has, but I am not sure it did anything useful.

I never see the bad behaviours with an external GPS antenna connected,
only with internal antenna.  If you connect an external antenna to the
internal antenna connector using the internal antenna cable, then it
also works fine.

Since it seems the firmware can act differently depending on signal
quality, and it seems it can illegally remember things from previous
sessions even when the power is cycled, it's hard to tell from which
direction any change in behaviour comes from.  For example if things got
better, is it because of the last change or because it has simply
randomly incrementally reached a critical point with gathering data from
the satellites?

I decided to look at the internal antenna and took apart an internal
antenna by removing its can, I noticed the phantom voltage on there used
to power the internal LNA was a bit low at 2.63V.  I decided to try to
increase this, which I did by changing the level on LDO5, initially
trying for 3.6V.  But LDO5 can only generate something below 3.3V since
its input is IO_3V3.  When set to 3.2-ish, we get 2.85V on the internal
antenna power.

Unfortunately the GPS behaviour can be so sticky and variable that I
didn't manage to make a test that pinned this down as having been
effective.  However I put this patch on andy branch and maybe it's worth
a try by anyone with internal antenna GPS issues.  By default this patch
sets the GPS voltage to ~>3.2V instead of 3.0V.

Another thing I tried was sending a UBX command CFG_RST each powerup to
clear the battery backed up memory region of all its data, but this did
not seem to do anything about the seemingly sticky behaviours between

- -Andy
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora -


More information about the openmoko-kernel mailing list