Re: [gpsd-dev] [PATCH 5/9] Define TTFF

[Gary E. Miller]

Yo Hal!

First off, best to be very specific about the Almanac and the Ephemeris.

They are different, but both are needed.  They are also send on different
timescales.  The Almanac is course, but is valid for weeks.  The Ephemeris
is accurate, but only valid for 30 mins.  Each sat only sends it
own Empemeris, but sends all Almanacs.

I write a tool so gpsd can dump this data, but never got around to checking 
it in.

On Fri, 17 Apr 2015 11:45:52 -0700
Hal Murray <address@hidden> wrote:

> > +<p>The time required for a GPS to get a fix (Time To First Fix
> > (TTFF)) +can vary from under 15 seconds up to just under 30 minutes
> > (actually, +29 plus calculation time).  The main factors affecting
> > this latency are +(a) whether it has an almanac available,
> > +(b) whether it has satellite ephemerides available, and
> > +(c) whether it has recent fix available.
> > +Of course the quality of signal at your location matters as
> > well.</p>
> 
> Perhaps I don't understand what's going on, but I think it's
> misleading to claim something like a 29 minute upper limit.  I think
> that reasoning is based on 15 minutes for the whole cycle.

Not misleading at all.  29 minutes is pretty common.  Most GPS, less
so with modern ones, will wait for the start of the next 50bps cycle,
which could take 15mins minus a few seconds.  Then take 15 mins to grab
a whole cycle of he almanac.  Not to be confused with a full almanac
which can take hours.

The almanac tells the GPS rceiver which sats should be in view at
the current time.  This is important for old single channel receivers
to not waste time looking for a sat.

Modern multi channel receivers just try to listen to a bunch of sats at
once to try to grab their Ephemeris.

Most GPS cheat on their hot start times.  They assume it stored the 
lat/lon of the last fix and has an approximmate time of day.  So can guess
which sats to try first.

Ship a GPS from Shenszhen to Oregon and it often takes 15 to 30 mins
to locate.


> That assumes you can only hear one satellite and can hear it clearly
> for the whole time and that you have to start at the beginning of a
> cycle.

Sort of.  Each sat send its own ephemeris, and only its own ephemeris,
so you get the one matching the sat you are receiving.  To get all the
orbital parameters only takes about 30 seconds.

But without the almanac finding all the sats, and handing off as sats
move will be troublesome.

> The data is sent in chunks.  You have to collect them all.

Not really.  Some of the Almanac data is for weeks in the future, those are
not needed for a current fix.

> With one satellite, you can start in the middle.  So the nasty case
> takes N+1 chunk slots rather than almost 2x the whole thing.

Assuming smart firmware, not guaranteed.

> All the satellites are sending the same data.

Nope.  Each sat send only it own ephemeris.  Each sat sends all the
almanacs.

>  If you can hear
> several satellites,  you can potentially reduce the collection time
> by the number of satellites.  (But that assumes you can find them
> which is hard without the data you are trying to collect.)

Nope, they pretty much all send the Almanac in lockstep.

> If you are unlucky, some noise will trash a chunk of data so you will
> have to go around again.  If you are really unlucky, it will trash
> the same chunk several times in a row.

Or you get Almanac part from one sat, and part from another.

> How long is a GPS satellite in view?  If I want to listen to one for
> 15 minutes, what are the chances it goes out of sight?

Easy to see, just stare at your GPS sky view for a bit.  The sats are
in vaying orbits.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701
        address@hidden  Tel:+1(541)382-8588