[ntp:questions] Re: Interaction of PPS and fudge time1 - how to fudge a GPS?

David Schwartz davids at webmaster.com
Mon Mar 28 02:17:08 UTC 2005


"John Pettitt" <jpp at cloudview.com> wrote in message 
news:GdI1e.12435$m31.127222 at typhoon.sonic.net...

> What I know so far is that if I run my GPS_NMEA without the kernel PPS
> flag then the time1 fudge seems to work in that the system syncs to
> GPS time + fudge factor.   However if I enable the kernel PPS flag
> then the system locks to the PPS and ntpq shows the GPS with an offset
> equal to the fudge factor.

    Right, so don't enable the kernel PPS flag.

> My assumption is that the kernel PPS (FreeBSD 5.4-BETA1) knows nothing
> about the fudge factor - is this correct?

    Yes.

> If this is the case how can
> I get ntpd to look at PPS but without the kernel being involved - if I
> just don't set the flag then ntptime shows PPSSIGNAL but not PPSTIME
> or PPSFREQ (I notice in the NMEA driver source enable_pps is only set
> if the kernel pps flag is given).

    Do not enable hardpps. You'll suffer a slight reduction in accuracy, but 
the PPS input will still be used by NTP to compute time offsets, just not by 
the kernel directly.

    DS





More information about the questions mailing list