[LinuxPPS] Re: nmea driver

Udo van den Heuvel udovdh at xs4all.nl
Thu May 31 20:23:08 CEST 2007


Paul wrote:
> I am using your latest (12 May 2007) patch to NTPD. Hardware is a
> Jupiter board and a homemade interface. The nmea and pps goes on a
> serial port.

OK.

> I put the required 127.127.20.0 stuff in the ntp.conf and after a while
> the ntpd advertised itself as stratum 1.

1?
should be 0?

> So just to check I added a pool server - which was 1 second out.
> Unfortunately the problem appears to be me, because when I had 3 pool
> servers, they all agreed (well close enough) with each other, with
> offsets of a few mS., but not with my nmea which gives an offset of
> about -1000 (mS.) from ntpq -p and marks as a false ticker.

1 second is way big of an error. Something must indeed be wrong.

> The pps output is a pulse of about 25 mS. on these boards, so by
> changing the assert/clear on flag 2 I can change the offset to -1025.
> 
> Any ideas?

Not really at this point. That is why I am sending the reply also to the
LinuxPPS list.
My pulse length is longer (200 ms?) but should that mater much?

> It would be a shame to just use the pps signal and pool servers when I
> have nmea data.

Indeed. Please try the longer pulse length?

Do other readers have tips?



More information about the LinuxPPS mailing list