[FSO] Speed gps time to firxt fix up?

Helge Hafting helge.hafting at hist.no
Thu Feb 12 11:20:16 CET 2009


Daniel Willmann wrote:
> Hi,
> 
> On Sat, 03 Jan 2009 13:50:18 +0100
> Fox Mulder <Quakeman1 at gmx.net> wrote:
> 
>  > Since a few weeks i'm using fso-gpsd without any big problems so far.
>  > The only problem is that the TTFF (time to first fix) is always >1
>  > minute and sometimes it needs quite a lot of time. The only solution
>  > for me to be sure to get a fix within <30 sec is to use the agps tool
>  > which downloads the almanac and ephemeris from ublox for my region.
> 
> yes, I tried integrating hot-start into ogpsd, but somehow I had
> problems with supplying ephemeris. Just some days ago I noticed
> something that made me think the culprit is still the time calculation.
> The SVs that zhone claimed were visible suddenly made a jolt and were
> positioned differently. It was more than one, so it wasn't just one
> almanac that got updated. I'm not completely sure, though, didn't look
> too closely.
> 
>  > The problem is, that the gps chip doesn't have any internal memory why
>  > it needs to download all relevant gps data every time it is powered
>  > up. So i thought if it is possible to save the relevant gps data
>  > before shutting down the gps which can be done by fso-gpsd or kernel
>  > automatically. Than after starting the gps fso-gpsd/kernel feeds back
>  > the data to speed up the TTFF. This would be the same way most gps
>  > receivers work.
> 
> It actually does have memory, but lacks a backup battery which is
> supposed to power the memory of the GPS while main power is off.

Could this power be taken from the FR battery, for a while? I assume the 
gps unit draws less current when it only keeps the memory alive and 
don't track.

Restarting the gps software (or stopping one app and then run another) 
is sort of common, we could avoid startup time. Something like: when gps 
becomes unused, stop it but keep the memory powered for 10min in case 
the user wants to use it again soon. Turn it compeltely off to save 
power after that, or if the phone suspends.

> ogpsd already stores and restores almanac, GPS/UTC timeoffset and
> approximate last position. It's not restoring ephemeris because I had
> weird problems with that (TTFF increased to several minutes).
> 

 > The manual claims down to 5 seconds TTFF if you supply ephemeris and a
 > precise timepulse.
 > Without the timepulse TTFF is supposed to be around 24 seconds.

Can the weird problems be avoided by not providing the time pulse? 24s 
is better than the current 1-2 minutes.


Also, the gps can get weird problems while powered. When I drive through 
a tunnell, it looses the fix. (As expected!) When emerging from tunnels, 
it gets a new fix reasonably fast, usually.
But there is a double tunnel where I live, that really makes problems 
occationally. I drive through the first, and loose the fix. The distance 
to the next is _slightly_ shorter than the time usually needed to regain 
the fix. Then I get into the next tunnel. Sometimes, this results in 
10min or so to get a fix again after emerging from the second tunnel. 
Looks like this unit does not like being interrupted when it almost gets 
a fix.



Helge Hafting




More information about the community mailing list