[ntp:questions] How exact are the PPS pulses of common GPSses?

unruh unruh at invalid.ca
Thu Feb 14 17:03:28 UTC 2013

On 2013-02-14, David Woolley <david at ex.djwhome.demon.invalid> wrote:
> unruh wrote:
>> If it were out by 1us, it would put you out by almost a km on the
>> surface of the earth.
> That's a requirement for the relative time from two satellites.  The 
> absolute time and any propagation effect that is common to all 
> satellites in view, can have a higher error than that that implied by 
> light travel time effects.  Time errors in common across all in view 
> only produce errors proportionate to the orbital velocity, not to the 
> speed of light.


> I once found an official Navstar document that suggests that, for time 
> transfer purposes, GPS is good to 50ns.  These were figures for SA off, 
> although the document was from the SA area.  I think they were for a 
> relatively short averaging time.

The evidence from the people who do radio astronomy ( who use gps to
provide the time base for long baseline interferometry I believe) is
that it is good to better than that. I cannot point you to the relevant
documents unfortunately.
Now it is true that this depends on the receiver, and I recall tests
that the more recent Motorola timing receivers at the time had an
unexplained 30ns difference from the older ones. But that is ns, not us.
Computers are good to us.

More information about the questions mailing list