Re: Clarifications about PPS SHM content

[Martin Boissonneault]

Hi Gary,

On 2020-04-06 15:01, Gary E. Miller wrote:

> Yo Martin!
>
> On Mon, 6 Apr 2020 03:53:32 -0400
> Martin Boissonneault <address@hidden> wrote:
>
> > Note: GPSd is 3.20 compiled from tarball, NTPsec 1.1.8 too,
> >       
> Give gpsd git head a try.
>

As you probably know by now, I did. Sorry for finding that annoying bug! Should I bother recompiling from head?

Result: no change, still ~750ns jitter

From http://ve2mrx.dyndns.info:10180/ntp/day/

                    Percentiles......                                           Ranges......                                Skew-   Kurt-
Name                   Min      1%      5%     50%    95%     99%     Max         90%     98%   StdDev      Mean    Units   ness    osis
Local Clock Offset  -2.732  -1.599  -1.045  -0.121  1.259   2.061   5.238       2.304   3.660   0.751       -0.018  µs      -3.36   9.281

The order for a new RasPi 4B 2GB has been placed. I intend to transfer ADS-B functions to a Pi by itself and let the NTP server do only time-keeping.

I don't expect the Pi 4B to be better than the 3B+ at timekeeping, but does anybody have a comparison? Numbers?

I expect a D'Oh! moment when I start building the new time-server. I'll probably stumble on an unrecorded change that messed my time quality... Part of learning!

Thanks,
Martin