[ntp:questions] ESR looking for good GPS clocks

Rob nomail at example.com
Tue Mar 6 19:55:49 UTC 2012

unruh <unruh at invalid.ca> wrote:
>> SiRF chips should be used in SiRF binary mode.  ESR should know that.
> ??? Are you claiming that the reports are better? You have tested it?

I have used SiRF binary mode (in fact I have written the original code
that gpsd uses in SiRF binary mode) and it works significantly better
than NMEA.

Partly because there *is* a data packet in SiRF binary mode that reports
the current time (vs the "time of last fix") and that has a defined
moment relative to the packet sending time where it is valid.

Of course there still is a fixed offset between the moment you receive
the last byte of the packet and the timestamp contained in the packet,
and that offset depends on the data rate, but it is all way better than
the situation with NMEA, because there is no wander such as between the
GPS fix process and the NMEA sentence emitter.

It is not good enough for 1ms accuracy.   For that (and better) you need PPS.

More information about the questions mailing list