External GPS antenna question

[Al Johnson]

On Saturday 04 July 2009, Paul Fertser wrote:
> Al Johnson <openmoko at mazikeen.demon.co.uk> writes:
> > On Saturday 04 July 2009, Paul Fertser wrote:
> >> Sebastian Krzyszkowiak <seba.dos1 at gmail.com> writes:
> >> > On 7/4/09, Hammer Armin <ahammer at datazug.ch> wrote:
> >> >> With the internal gps antenna, the first fix last quite long -
> >> >> with the external - within 5min to 10min I get a fix.
> >> >
> >> > WTF? I'm getting fix easly with *internal* GPS antenna in 30 seconds -
> >> > 2 minutes...
> >>
> >> Because you already have almanac (and sometimes ephemeris) data.
> >
> > With a good sky view and suitable constellation TTFF from cold is ~40s
> > with the internal antenna, rootfs on SD and an early A05 without the
> > capacitor on SD. For it to take >2min I need heavy urban canyoning, being
> > indoors or in a vehicle, or an unusually poor constellation.
>
> Let me cite the site that looks to be authoritive enough [1]:

[snip quote from link]

> 25 * 0.5 min = 12.5 min . I assume this gives a good estimate of TTFF
>
> If you already have almanac (which is the usual case as almanac is
> usually valid for months), you'll get a fix in a little more than 30
> seconds (time required to receive ephemeris from visible sattelite).
>
> [1] http://gpsinformation.net/gpssignal.htm

~40s was a consistent lower limit from cold start TTFF tests made to test the 
kernel SD drive strength and idle clock. IIRC this closely matches the cold 
start TTFF specified for the Antaris 4. Consensus at the time was that the GPS 
was powered down for long enough between tests that there would be a true cold 
start since the version we use has no nonvolatile storage for almanac and 
ephemeris. Assistance (ie providing almanac and ephemeris) is supposed to 
reduce TTFF to 10-20s, still significantly longer then a warm start.

If you want to repeat the tests yourself the script and method are in the 
archives.