Re: 3.22 not sending PPS to SHM

[David Taylor]

On 15/01/2021 18:39, Gary E. Miller wrote:
> Yo David!
>
> On Fri, 15 Jan 2021 07:57:48 +0000
> David Taylor<gm8arv@yahoo.co.uk>  wrote:
>
> > In this configuration, the PPS is coming over the GPIO, and is
> > independent of the serial connection. Here, ntpd type 22 driver does
> > not need a serial connection.
> I write the howto, I'm aware of that.
>
> But you fail to note that the type 22 does not use KPPS, so you have
> a noticable increase in jitter.
>
> > On most of my systems I'm using that configuration, with multiple
> > local stratum-1 and pool internet sources.  The PPS is a nice add-on
> > for much better time-keeping, but it's not essential.  In practice
> > there have been no problems - except once when a GPS-jammer equipped
> > vehicle drove past!
> In addition to the loss of accuracy, there will be instances of
> seconds jumping.
>
> RGDS
> GARY

Gary,

As both GPSD and NTP(22) use the same /dev/pps0 source, how is the 
jitter not the same?  I recall discussions on the NTP lists that the 
current implementation is better than the old kernel PPS, not worse.

Reported accuracy appears fine, and I've not been aware of any loss of 
seconds.

Cheers,
David
--
SatSignal Software - Quality software for you
Web: https://www.satsignal.eu
Email: david-taylor@blueyonder.co.uk
Twitter: @gm8arv