[time-nuts] Time-Nutters-- Adding Rubidium to a Thunderbolt...?

[Poul-Henning Kamp]

--------
In message <CADHrwpfp1kuXWW8XzQpqK+m=47Z5PN4suMzfve_GjcjhZzDRvA at mail.gmail.com>, Dana Whitlow writes:

>Probably the easiest solution would be to feed 1 PPS from the T'bolt into a full-featured
>PRS-10 Rb, then fiddle with the loop parameters in the PRS-10 to best suit your needs.

Been there, tried that, not a good idea.

The PPS input of the PRS-10 is OK for "coarse" PPS signals like
those you get out of a GPS, because the sawtooth noise dithers over
the unlinearity and "missing codes" of the PRS-10 timestamping
circuit.

When you feed a PRS-10 a "perfect" PPS signal, you get really
erratic behaviour.

Only way to slave a PRS-10 from a "good" source, is to do
phase-comparison on the 10MHz signal, and after suitable
filtering and integration, use that to set the PRS-10 frequency
via the serial port.

That, on the other hand can give amazing results because of the
relatively good X-tals they put in the PRS-10.

-- 
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.