[ntp:questions] Re: Fixed-location GPS mode?

Terje Mathisen terje.mathisen at hda.hydro.com
Tue Jun 6 09:09:01 UTC 2006


Tom Van Baak wrote:
> The GPS 18 LVC doesn't have a zero-D mode. However
> it is a very nice, compact, easy-to-interface, GPS timing
> receiver with a 1 us 1 PPS accuracy spec.
> 
> Although this makes it a poor choice for work with atomic
> clocks it should be more than enough for NTP, yes?

It is very useful indeed, and at less than $100 for a complete NTP 
server (assuming you can locate any kind of spare PC at all, a 
1995-model Pentium is fine) probably the best value NTP Stratum 1 clock 
available.

> Recent GPS 18 LVC measurements I made showed it
> much better than spec; a couple hundred ns peak to
> peak and just 80 ns RMS for the 1 PPS output.

Thanks! That's very interesting, and sort of confirms what I've 
suspected for some years:

Since a working 3 m GPS PVT solution requires ~10 ns time, it really 
shouldn't be too hard to output the PPS signal within an order of 
magnitude of this. Giving away two decades (to ~1 us) seems sort of 
excessive, right?

Yeah, I do know that GPS systems like my Motorola Oncore UT+ has to 
slave the PPS pulse to an internal 10 MHz (free-running) osc, limiting 
absolute precision to +/- 50 ns (before the sw sawtooth correction 
term), but that would still indicate 100 ns as a (cheap) reachable target.

Terje

-- 
- <Terje.Mathisen at hda.hydro.com>
"almost all programming can be viewed as an exercise in caching"




More information about the questions mailing list