GPS warm vs cold start

GPS's give satellite acquisition times based on warm start vs. cold start.
What determines whether it is a cold start or warm start?

I can see that if it has been on for a while then switched off to change
batteries and then switched back on again this would be a warm start, but
how long does it have to be powered off to be classed as a cold start?

Ta.

 

No, GPS receivers quote warm start and cold start specifications. Since those
specifications are individual to the brands and models, the manufacturers can
quote whatever numbers they want, calculated whatever way they want, based on
whatever assumptions they want, and power-off times of whatever they want. It
is not part of the overall GPS system spec.

---Bob Gross---

 

It's even a warm start without changing the batteries.

The definition basically goes back to the time when state-of-the-art
receiver technology was slow in aquiring signals from the satellites.
This is the equivalent of comparing an 8 MHz early 80's PC with todays
GHz machines.

The receiver had to look out for a maximum of 32 satellites, each search
could take multiple minutes. In order to speed this up, the almanac was
introduced as a memory the receiver would keep in order to predict the
satellites that would be visible for a given time and position on earth,
thereby reducing the unnecessary searches for satellites that are not
visible. The almanac would be valid a couple of weeks and be updated in
the background during regular operation.

If the receiver has no almanac, it has to execute a search on all 32
possible satellite PRNs, which will take considerably longer, because it
will be busy searching in vain for at least 20 satellites that are not
visible at all. Those old receivers could take 10-15minutes for a first
fix, or any time longer.

A cold start is a start without an almanac or with an almanac outdated
to an extent that it is not representing the actual constellation. It
depends on the receivers firmware implementation wether the receiver
will discard a very old almanach, or wether it will assume it as 'still
partly valid'. If he is unlucky, he will then discard a large number of
satellites from the search that are in fact visible. That way he could
search even for hours without aquiring a fix in a worst case scenario.
There's a bit of statistics involved because the actual constellation
can have any possible state, and the receiver may just give up on a
search and switch to another satellite right before the satellite he
gave up on turns up on the horizon (bummer).

We have seen such a behaviour from older Magellan devices that have not
been switched on for a couple of months or years.

On a warm start, a receiver will discard all satellites from the search
that compute invisible from the current almanac.

This searchtime optimization scheme looses it's importance with modern
receiver technology, where the searches are not performed sequentially
but in parallel. The effective search times are reduced by throwing
thousands of search channels at the signals instead of only one at a
time. That way even looking for 20 invisible out of 32 possible
satellites doesn't count too much.

Although of course it is nice to see the current visible satellites on
the display even before the receiver has computed a position.

Garmin and Magellan never mention the number of correlators/search
channels their chipsets use. Some chipset manufacturers (Trimble) quote
32 per tracking channel, SIRF-II uses 1920 in total, ATMEL Antaris uses
8192 in total. Forthcoming chipsets will have hundreds of thousands of
correlators.


- Carsten

 

Well, Dale quoted the definitions of these procedures from the NavStar
manuals.

However, the times a manufacturer gives for these clear definitions are
statistical figures obtained by measuring a series of starts under
conditions the manufacturer considers as typical. Some manufacturers
even show TTFF probability curves for their receivers, and varying
conditions like the number of visible satellites or signal level.

- Carsten

 

No, both are warm starts. Okay, after more than 30min off, the receiver
may feel cold in your hands, but that's another issue ;-)

The receiver is constantly updating the almanach, not just after a few
days. A full almanach transmission lasts 12.5 minutes. The receiver may
still be able to aquire a fix before that. Read Dales quote of the
definitions in the

Re: IS there room for passive augmented GPS (reception from cellular

thread.

- Carsten

 

However, the last time I looked, Garmin used different definitions of
the different types of start than the Navstar manuals used. So you
can't compare them without knowing what a particular manufacturer
actually means.

Dave

 

Not according to Garmin. A warm start is one where, while the receiver
has been off, the overall SV makeup and geometry hasn't changed much.
I.e., the SV's the receiver was using to get it's position when it was
turned off are still in view, not over the horizon yet, when it's turned
back on.

A cold start is when the above is not true and takes maybe 45 secs to a
minute to get position vs. maybe 15 secs or so for a warm start.

That's even more than a cold start.

Steve

 

The determination has to do with whether or not the sats the gps wants
to use for the position calculation have current orbit data. In some
circumstance you could be off for a few minutes and not have it, and
in others, you could be off for several hours and still have good
data.

In one case you have to download the current high precision orbit data
(which is quite slow, about 100bps), and you cannot calculate a
position until you have it. In the other you just go to calculate
position.

There is one other circumstance, and that is where you don't have a
current almanac, and then you can expect it will take about 15 minutes
to get a fix. You have to download enough of the alamanc to find
enough sats that you can get the precision data from before you
calaculate a position. That is the circumstance you are in if the GPS
has not seen the sky for 6 months or more, or possible if it was
stored without batteries, and the internal battery backup has run down
(Garmin units are famous for that)...

 

Correct except the data rate is only 50. The information is repeated
every 30 seconds and it takes 18 seconds to download it.

Street wisdom at one time thought the 30 minute boundary was significant
and it seemed to fit the anecdotal evidence on the older mulitplex units
but Garmin has stated that they never had a 30 minute aging algortithm
to determine the age of the ephemeris data.

Actually Garmin says they can get a fix in 5 minutes. It never needs the
full almanac to get a fix, only enough of the satellites (3).

You have to download enough of the alamanc to find

True but since it is only 3 and there are 12 channels to deal with it
Garmin says they can do it in 5 minutes on an average so long as the
clock is halfway close. It does take 12.5 minutes to get a full almanac
but that can go on in the background.

Dale