[time-nuts] TIC resolution impact on GPSDO's performance

[Poul-Henning Kamp]

In message <45905B32.4000301 at xtra.co.nz>, Dr Bruce Griffiths writes:

>>
>> As far as I can see on TVB's site, the 8607 is about 5e-13 at 1000 sec,
>> so 1ns/1h sounds perfectly good to me.
>>   
>Only true when one only considers the data available on Tom's site.
>Quartzlock have a GPS disciplined BVA OCXO that achieves considerably 
>better stability, admittedly their carrier phase tracking GPS receiver 
>has a resolution of a few picoseconds for an integration time of 1 second.

I'm not sure what your point is here...  Obviously if you discipline
something with an Oncore, you have to choose a PLL timeconstant that
is optimum for the oscillator you are using.  If your oscillator is
very good, the oncore may not be good enough at any timeconstant.

But the 1ns resolution of the Oncore is a fact of life and for all
the "amateur" stuff we are talking about here, 1ns is plenty fine.

>>> Its not the coherence of the sawtooth but the coherence between the 
>>> oscillator clocking the timer used to position the PPS signal and the 
>>> frequency of the PPS signal itself.
>>
>> Yes, those are non-coherent, that is why we need the negative sawtooth
>> in the first place.  If they were coherent, we could just have
>> used a formula.
>>   
>But they are not sufficiently incoherent as hanging bridges and the 
>quasi periodic form of the sawtooth error amply demonstrate.

Your choice of the word "coherence" does not help the discussion
here because it implies that the signals are the same frequency,
which is not the case.

But if you insist on using the word coherence, then at least
use it correctly:  The hanging bridges happen because the two
signals are not coherent.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.