[ntp:questions] Re: NMEA refclock - PPS not working

R Jenkins not at pub.lished
Mon May 15 07:34:21 UTC 2006

"Harlan Stenn" <stenn at ntp.isc.org> wrote in message 
news:ywn9ody0dx0z.fsf at ntp1.isc.org...
> The time reported by your NMEA clock is far enough away from your other 2
> sources that ntpd is not believing your NMEA clock.
> H


it's supposed to be working with PPS; that is the absolute time reference.

I'm using a Garmin GPS25 as per the example in the NMEA Refclock 
It has only the $GPRMC output string turned on, at once per second.

Others are using successfully using exactly the same hardware - e.g. from 
Dave Morgan in reply to my previous topic:
"I am running a gps25hlv direct with no extra conversion for the PPS but 
with a 2.4.29-NANO kernel on a Debian/Woody system."

The NMEA data is only 4800 Baud, sent after the PPS edge that it relates 
to - by the time the NMEA string has been received and validated, it's got 
to be waaay behind the 'real time' reference pulse edge.

In practice, looking at the serial lines with my vintage 'Interfaker' RS232 
monitor box, the GPS does not send the serial data until visibly well after 
the END of the 200mS PPS pulse.
This is the basis of any serial + PPS GPS, so presumably this is what the 
refclock driver expects.

The real problem seems to be that the PPS signal is not being seen somewhere 
along the way, or is somehow too unstable (in terms of kernel timing) to be 

Robert Jenkins.

More information about the questions mailing list