[time-nuts] Using GPS 1PPS for accurate period measurement

Azelio Boriani azelio.boriani at screen.it
Fri Dec 2 04:20:46 EST 2011


I agree: better using a cheap counter (using the GPSDO to sync it). However
to do a time interval measurement your sampling frequency must be stable
enough between the reference PPS and the PPS-under-measure. The trick is to
offset the PPS-under-mesure enough (say 100mS) to gether some stable
samples from your sampling clock. Moreover, to get the best from averaging,
I think that the sampling clock must not be synchronous with either the
reference or the PPS-under-measure.

On Fri, Dec 2, 2011 at 8:39 AM, mike cook <michael.cook at sfr.fr> wrote:

> Le 02/12/2011 01:53, Bob Camp a écrit :
>
>  Hi
>>
>> Ok, 62.5 ppb at 1 second would be 62.5 ns. that sounds right for a 16 MHz
>> clock.
>>
>> Your accuracy will be related to the offset between the two 1 ops events
>> (divided TCXO and GPS PPS) and the accuracy of your crystal.
>>
>> With some luck you can get the crystal frequency to<  1 ppm by counting
>> it against the GPS pps output. If the TCXO pps is within 100 ms of the GPS
>> pps, you should get about 100 ppb accuracy.   The trick is to keep the TCXO
>> pps that close. Can be done, but it takes some fiddling.
>>
>>  I don't think that the measurements  have to be done concurrently.
>
>  The GPS pps signal will have a low ns jitter,  possibly defined in the
> data sheet, that can in this context  be ignored if measurements are take
> over a long period as can that for the RTC.
>
> So for me, I would characterise the micro first,  counting cpu cycles per
> GPS PPS period. If that is done over enough periods, a good indication of
> the measuring devices rate against a known standard will be known. (1)
>  Then measure the RTC pps using the same procedure. (2)
> Finally adjust the result from 2 by that of (1).
>
> However, even that will be more costly than renting a counter.
>
>
>
> ______________________________**_________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/**
> mailman/listinfo/time-nuts<https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts>
> and follow the instructions there.
>



More information about the time-nuts mailing list