[time-nuts] Where does the Z3801A 1 PPS come from?
Tom Van Baak
tvb at leapsecond.com
Mon Mar 14 04:05:37 UTC 2005
> Very interesting, Bill. Thanks for that info. I'm a bit surprised as I
> would have thought that deriving the 1 PPS from the 10MHz would yield
> the best performance. Of course, this means that using the 1 PPS to
> measure the stability of the 10MHz oscillator isn't correct, so I've
> done a bunch of non-meaningful measurements recently. :-)
Not to worry. The positive way to look at your plot
is that you have confirmed the two outputs agree to
within +/- 1 ns peak-to-peak over more than a day.
That's good. And it means the difference between
using the 1 PPS and the 10 MHz output for any
measurement is on the order of a few parts in
10^14 at a day.
Nothing you measure is that low so my conclusion
is that you can safely use either the 1 PPS or the
10 MHz as a reference for any measurement you like.
Note that 1 ns is easiliy within thermal variations.
I'm running your experiment here just to be sure. It
would be good also to show the 1 ns excursions are
not something in your counter. Was this a 5334A?
> The CPU must be applying some smoothing to the GPS 1PPS as the jitter is
> significantly less than the raw pulses, but it can't be doing too much
> as there's still about ~50ns noise (over timescales of hours) in both my
> units. I wonder what the algorithm is?
The CPU, like all of us, does software averaging on
the raw Oncore 1 PPS (but I don't know the details).
> Another interesting experiment suggests itself... measuring both the CPU
> generated 1 PPS, and the 10MHz derived 1 PPS, against a third source to
> see the relative jitter of each.
I may have done this and will look for the old data. My
understanding was always that the 1 PPS output was
tied tightly to the OCXO. I guess I don't buy into the
idea that the CPU is generating the 1 PPS on its own.
To me, 1 ns agreement between two signals suggests
it's all hardware-based, not software. But I'd be happy
to learn I'm wrong.
/tvb
http://www.LeapSecond.com
More information about the Time-nuts_lists.febo.com
mailing list