David Woolley wrote: > Richard B. Gilbert wrote: > >> >> My Solaris 8 SPARC system seems to be able to synch with the GPS >> receiver in about 30 minutes. I may reboot that system once a year > > How do you define "synch"? NTP has an infinite impulse response, so > will never completely synchronize. What Unruh means is the time until > the residual error is comparable with the measurement error specified > for the receiver, which is one microsecond for cheap GPS receivers and > 50 nano-seconds, according the the GPS performance specification. The > error also has to stay that low, e.g. the first zero crossing doesn't > count because the error will overshoot on the other side. > <snip>
I don't really give much thought to "defining" synchronization. If pressed, I'd say "remaining within one millisecond of the correct time". The GPS receiver generates a PPS signal accurate to within 50 nanoseconds. I don't have the equipment necessary to measure the loss of accuracy getting the time into the computer. I'm not sure I'd be able to use it if I did have it. _______________________________________________ questions mailing list [email protected] https://lists.ntp.org/mailman/listinfo/questions
