PPS documentation

Achim Gratz Stromeko at nexgo.de
Mon Feb 27 19:59:23 UTC 2017

Achim Gratz writes:
> Feb 26 21:53:26 raspberrypi2 ntpd[20604]: refclock_params: kernel PLL (hardpps, RFC 1589) not implemented
> Feb 26 21:53:26 raspberrypi2 ntpd[20604]: NMEA(1) set PPSAPI params fails
> $ uname -a
> Linux raspberrypi2 4.4.49-v7+ #968 SMP Fri Feb 17 14:22:28 GMT 2017 armv7l GNU/Linux
> AFter that ntpd then falls back to non-kernel discipline.

…and entirely non-gracefully.  I've left it running with that config for
a day to see what would happen and it wildly oscillated between ±10ms
time offset.  Interestingly enough an averaged (via kernel smoothing)
version of the offset still kept within ±10µs.  In any case ntpd clearly
should have just ignored the request for the unavailable kernel PLL and
used it's own (just like flag3=0), but it wasn't happening.  I think it
just ignored the PPS signal altogether and was working with the NMEA

+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

SD adaptations for Waldorf Q V3.00R3 and Q+ V3.54R2:

More information about the devel mailing list