Raspberry Pi NTP config with fudge factors

[Frank Nicholas]

I haven’t used the NTP configuration below for years.  I came up with it from some recipe(s) I found online (some commented in the file).  The "time2 0.350" was the generally accepted best number for the original Raspberry Pi A/B.  I personally have no idea how to tune NTP.  I just used what I thought was the best information I could find at the time.

My current NTP server is part of pfSense (FW/Router/NTP/etc., based on FreeBSD-10.3-RELEASE)  Their web interface is limited in some of the NTP/PPS/GPS settings you can easily get to.  If you make changes outside the web interface, you risk them being overwritten by the web interface.  I really need to get a good config nailed down & not touch it.

> On May 5, 2016, at 7:12 PM, Gary E. Miller <gem at rellim.com> wrote:
> 
> I use the 28 refclock, you the 20.

Regarding 28 ref clock, that is SHM?  Why that rather than NMEA (20)?

> I think for our purposes time1 and time2 are the same?  Basically the
> NMEA fudge.
> 
> I use a time1 of 0.480 for my Ultimate GPS HAT.  Just about in the
> middle of my 30 to 70 milliSec jitter.

I don’t remember what the difference is between time1 & time2.  I do remember it was for NMEA fudge  I just know the documentation I found always had time2 set.

> I find a short poll leads to a more unstable result.  Have you tried
> a longer poll?

I’ll look at this on my current configuration.  What’s a good way to “tune” my NTP config?  What’s a good way to monitor my NTP performance?

> Mode 17 is 9600?  Up that if you can.

Correct, 16 = 9600 & 1 = specified the NMEA sentence to get time from, I think...  I’ll look at upping it on my current configuration.  I had to limit the statements that came out, so they didn’t run over the second (some periodic (every 5 or 10 seconds) NMEA statements made it take > 1 second for a cycle).

> You'll do better if you hand pick some close chimers.  There is some
> bad shit in the pool, and you may be using something from Singapore
> or Bangalore.  If you must pool, at least use a US pool.  Like:
> 
> server 0.us.pool.ntp.org iburst
> server 1.us.pool.ntp.org iburst
> server 2.us.pool.ntp.org iburst
> server 3.us.pool.ntp.org burst

Thanks - I’ll look at this on my current configuration