[ntp:questions] Accuracy of GPS satellite signal ?

Richard Eich richard.eich at domain.invalid
Fri Jan 26 20:33:07 UTC 2007


hal-usenet at ip-64-139-1-69.sjc.megapath.net wrote...
> In article <MPG.202131877f6d3e0498ad41 at news.verizon.net>,
>  Richard Eich <richard.eich at domain.invalid> writes:
> >If I were to set up five or six global locations, each time-
> >synchronized to a local GPS receiver, how much variability would I 
> >expect there to be between the time as indicated by the GPS 
> >satellites used by each local GPS receiver.
> 
> What decimal point are you interested in?
>    nanaseconds? microseconds? milliseconds?

0.500us

> Are you interested in the time out of the GPS box
> or the time out of an NTP server connected to a GPS box?

The time out of the GPS receiver.  The GPS receiver will be 
physically attached to the device to be sync'ed; no NTP.

> How much money are you willing to spend?

It always gets down to that, doesn't it?

> The GPS side depends upon how fancy your gear is and how
> good your antenna location is.  The speed of light is
> a foot per nanosecond.  GPS position accuracy for consumer
> gear is generally within a few 10s of feet so time accuracy
> should be within a few 10s of nanoseconds.  The national
> standards laboratories use GPS to compare their atomic
> clocks so you can get very very good if you work hard enough.
> 
> With a bit of work, you can probably get a NTP server using
> GPS to be within a few/10s of microseconds.  That's good enough
> so that you can easily measure asymmetries or jumps in network
> routing.  (assuming similar setups on both ends)
> 
> Getting the time from a good NTP server to other systems gets
> interesting.

Thanks very much for your help.




More information about the questions mailing list