[time-nuts] ✘NEO-M8N vs. NEO-M8T
MLewis
mlewis000 at rogers.com
Mon May 21 21:51:26 EDT 2018
Having a linux box (Pi) dedicated as a time server should mean you have
consistent delays?
To offload time server requests so they don't affect disciplining
response/timing, would it be worthwhile having one Pi dedicated to being
disciplined by the GPS, then have that pi discipline a second Pi that
handles time server requests?
PPS-Client measures the delay of a Pi GPIO port and over time adjusts
the PPS to compensate. I'd think PPS-Client could be modified to process
the sawtooth. And possibly to adjust for the delays in writing to system
time?
The M8T's two EXTINT (External Interrupt) pins keep nagging me that the
resulting timestamp (Timemark UBX-TIM-TM2 message) suggests that such a
GNSS chip timestamp (not the iTOW value) should be able to be used to
advantage, say to measure the net delays in getting a PPS to discipline
system time.
- Would the difference between a timestamp of Linux system time and a
chip-internal timestamp provide a meaningful and worthwhile adjustment,
to system time or to the incoming PPS?
- Would successive pairs of timestamps provide the net delay in writing
such a corrected system time, allowing for a further refined correction?
- Should such a correction be an absolute adjustment based on the pair
delta, or should a period of deltas be smoothed and applied over time?
- Would/could such a correction (measuring and adjusting based on the
end result of system time) provide a superior result in system time than
a sawtooth corrected PPS triggering a write to system time?
Michael
On 21/05/2018 3:21 PM, Gary E. Miller wrote:
> Gregory!
>
> On Mon, 21 May 2018 19:06:17 +0000
> Gregory Maxwell <greg at xiph.org> wrote:
>
>> My best guess is that the magnitude of sawtooth error is just not
>> large enough to matter for typical applications of linux PPS.
> No need to guess. I recently posted that the RasPi 3B granularity is
> 52 nano Seconds and the PPS offset reported by UBX-TP is double that!
>
> So, clearly it matters.
>
> I'll do more data logging to get harder numbers.
>
> RGDS
> GARY
> ---------------------------------------------------------------------------
> Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
> gem at rellim.com Tel:+1 541 382 8588
>
> Veritas liberabit vos. -- Quid est veritas?
> "If you can’t measure it, you can’t improve it." - Lord Kelvin
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts
mailing list