gpsd-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Clarifications about PPS SHM content


From: Martin Boissonneault
Subject: Re: Clarifications about PPS SHM content
Date: Fri, 20 Mar 2020 17:35:25 -0400
User-agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:68.0) Gecko/20100101 Thunderbird/68.6.0

Hi Gary,

I noselected the GNSS serial refclock. I don't think I can optimize it better on my rPi:

 						Skew- 	Kurt-
Name 		 	 	Mean	Units 	ness	osis
Local Clock Time Offset 	-1.672 	ns 	-4.39 	12.12
Local RMS Time Jitter		277.112	ns	28.36	220.4
rPi 3 B+, Raspbian Buster, NTPsec 1.1.8, GPSd 3.20, DS3231 RTC, uBlox MAX-M8Q GNSS (fixed, GPS only, SBAS integrity only)

I just wanted to tank you for the help in getting those results. I subscribed to the ntpsec users list, so any further questions about it will be there now.

Again, Thanks
Martin

On 2020-03-19 14:55, Gary E. Miller wrote:
Yo Martin!

On Thu, 19 Mar 2020 13:44:31 -0400
Martin Boissonneault <address@hidden> wrote:

Thanks for confirming my intuition. This thought came from watching 
ntpshmmon. The answer caused the question ;-)
Good data does that.

Follow-up question: I recall someone said that all time sources have
an influence over the stability of the NTPd time, and if you had one 
unstable time source connected (like the GNSS serial shm?), the whole 
stability was affected even if you have a better, more precise clock 
selected.
Well, sorta, sometimes.  NTP decision making is PhD level obfuxcation.

ntpd will only use info from a source is it is part of the best cluster.
So a bad pool server will usually be ignored, but it may, every once in
a while, yield a good time and be included in the cluster, then be quickly
thrown out.

Within the cluster, the bext chier is picked, and gets the "*".  That is
then filtered and then used to discipline the PLL that adjusts the system
clock.

IFF a bad chimer gets selected will it throw around the system clock.  This
can happen, during startup, when you have only a PPS and a GPS chimer, with
little jitter history.  ntpd flips a coin, and it it selects the GPS first,
less than optimal things happen.

This is why you put a short poll on the PPS, and put it first in the
ntp.conf, so it gets selected first, if possible.

Considering my GNSS serial SHM has been (x) and is stratum
11, do you think it is negatively affecting my stability and should
be commented out?
I am unaware how stratum plays into NTP selection algorithms.  Best to
ask that on their mailing list.

I know my ntpviz graphs would look better without the serial shm ;-)
Easy: noselect.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
	address@hidden  Tel:+1 541 382 8588

	    Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

reply via email to

[Prev in Thread] Current Thread [Next in Thread]