[ntp:questions] What is the NTP recovery time from 16s step in GPS server?

David Taylor david-taylor at blueyonder.co.uk.invalid
Tue Oct 30 19:28:26 UTC 2012

On 30/10/2012 17:59, Rob wrote:
> When I wrote that part of gpsd, I implemented the proper checks so
> that the time is only put in the SHM after the GPS receiver has
> indicated that it has a valid fix on the satellites.

THanks for this information, Rob.  In this case the GPS appears to have 
a valid fix, but it is outting GPS time and not UTC.  There must be a 
flag in the Trimble TSIP which tells you this, as I've seen the same 
when monitoring with the Trimble Studio software.  It takes some time 
for the switch-over to occur.

> I would hope that the receiver has collected the GPS offset by that
> time, so that it will not put the wrong time value in SHM during a
> short interval.

For this receiver that hope is not applicable.  Perhaps gpsd should only 
show UTC, or have a flag to indicate between GPS time and UTC?

> However, it has been a couple of years and many changes have been
> made to the code, I cannot tell you if it may have been broken.
> Do you really see the problem you describe above, or is it only a
> hypothesis of what might happen?

Oh, I really see this.  I had occasion to restart my Raspberry Pi NTP 
server and the gpsd source was off by 16.0 seconds for a few minutes
after booting.  The GPS receiver is powered from the PC.  At the moment 
it has multiple Internet servers as well as the gpsd/ATOM reclocks.
Web: http://www.satsignal.eu

More information about the questions mailing list