[ntp:questions] high jitter on serial gps causes big time offsets

unruh unruh at invalid.ca
Fri Apr 5 16:49:26 UTC 2013

On 2013-04-05, Nickolay Orekhov <nowhere at mail.ru> wrote:
> Hello!
> I've got the following configuration:
> tos mindist 0.128
> tinker panic 0 stepout 60
> # TSIP,PPS reference clock
> server mode 10 prefer maxpoll 3 true
> fudge refid TSIP time1 0.08
> server maxpoll 3
> fudge refid PPSI
> The main goal: I want 1 ms or better precision always, even with gps
> quality going high and low.
> When satellites appear and disappear again, serial clock could be selected

What do the satellites have to do with it? The internal gps receiver
clock can freewheel over the few seconds that the sattelites come in or
go out, and any jitter there is on the nanosecond level, not
I thought you were using pps. That is always good to at least micro
seconds (taking interrupt jitter into account), not msec. The serial
port gps is used only to set the seconds ( and the jitter there should
certainly be less than a second) not the sub second. 

So, the sub microsecond PPS should ALWAYS be selected over the serial

> before PPS clock. Or maybe some gps receivers will show good quality before
> PPS will appear or will be selected.
> In general, I want ntpd to skip adjustments coming from specific server or
> ref clock according to some constant precision. For example, I know that
> gps serial jitter is about 30-40 ms. So I want ntpd to do no adjustments
> lower than this value. And when PPS will appear it will be selected.

Why? ntpd will do the best job it can with the output of the serial gps
. If they jitter by 30ms, then the ntp clock will only be good to about
that. So what. When PPS comes on line, it will be good to microseconds. 

I think you have developed some model of ntp and of gps which is not
what they actually are. 

> Can this be done? Or how do you solve this situation?

Solve what? You have not presented a problem. You have presented a
solution to an unknown problem and asked us to comment on it. 

More information about the questions mailing list