rpi3 clock drift

[Peter Jeremy]

On 2019-Nov-28 23:51:53 -0700, Ross Alexander <rwa at athabascau.ca> wrote:
>I recall that some of the uBlox parts are designed for timing apps
>rather than navigation - they would be the ones to go with.

Yes, there are.  Unfortunately, they are going to be a lot more expensive
(the NEO-M8T is the cheapest and the bare module is $52 in 100-off).  The
cheapest board I've found is about €70.

>I have an identical (pi-2, ult hat, external patch antenna, debian)
>system running at work where I can get only a view of half the sky and
>it has significantly more jitter.  A broader sky view gives more
>confidence in the position domain (more birds, and they are more
>spread out), which translates into lower time jitter.

I've got a NEO-6M sitting against an external wall and it typically
sees 11-12 satellites and uses 8-9.  ntpd is reporting about 2µs
jitter (PPS input connected to a PC Engines APU2 - AMD GX-412TC) -
which seems good enough.  (I'm working on building a NTP server
around an ESP32 as a second datapoint).

>BTW, another *significant* source of jitter is the brand and age of
>the sd/mmc card used.  As they age, the write speed decreases and
>block write latency gets less uniform; this shows up as system clock
>jitter in the loopstats.

As with Ian, I don't understand this.  The only possibilities I can think
of is that the PPS interrupt priority is too low and/or you're not using
kernel PPS time capture.  (Is this on FreeBSD or Debian?)

-- 
Peter Jeremy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 963 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-arm/attachments/20191130/a3a2f6b9/attachment.sig>