[time-nuts] M12 100 Hz vs. 1PPS measurements
Tom Van Baak
tvb at LeapSecond.com
Mon Dec 8 02:52:32 UTC 2008
> Interesting results - thank you. Is there a small problem with the
> graph plotting routine though? On the left of the 100Hz plots, at
> some tau there are two points above each other (5 + 7 * 10^-2),
> and other tau have no data points (4, 6, + 8 * 10^-2).
Right, those first few points do look off by one. Another clue was
that at tau, MDEV must always equal ADEV. The bug is fixed
and I refreshed the plots for you. Good eye. Thanks.
> Nice and useful....
> What about Jupiter's 10 KHz output under same conditions?
> Predrag Dukic
Yes, a Jupiter is next on the list. My expectation in all these cases
is that more pulses per solution does not give any additional timing
accuracy. The inexpensive OEM GPS receivers that we typically
use compute one solution per second so that defines a certain limit.
You could say averaging solutions over time (minutes to hours) is
what gives greater precision to a GPSDO, not averaging pulses
(milliseconds to seconds).
However, that's not to say more pulses can't be useful in alternative
GPSDO designs. One that comes to mind is James Miller (G3RUH)
Jupiter-based GPSDO which, in spite of its extreme simplicity, has
great performance. See http://www.leapsecond.com/pages/gpsdo/
More information about the time-nuts