[time-nuts] Question on disciplining oscillators
james.h.maynard at usa.net
Tue May 2 20:15:39 EDT 2006
I think SJ's answer (quoted below) didn't really address the question
posed about ionospheric and tropospheric corrections. He spoke to the
time constants for the Rb oscillator's phase-lock-loop rather than to
the question of applying ionospheric and tropospheric corrections that
are broadcast in the GPS signal.
I'm also trying to discipline a Rb oscillator (SRS model PRS10) to the
one herz pulses from an M12+ timing receiver. I am using both the
ionospheric and tropospheric corrections.
I think though, that I may have have set too long a time constant for
the PLL time that disciplines the PRS10 to the M12+T's 1 Hz output. (I
have it set to about 18 hours, the maximum value.) I'll experiment some
more with it.)
Poul-Henning Kamp advised me to try a little higher mask angle, as a way
of reducing jitter due to multipath on the measurements from satellites
that are low on the horizon. I currently have the mask angle set at 15
degrees above the horizontal, but I still see about 9 or 10 ns of jitter
on the 1 PPS from the M12+ -- even with Rick Hambly's new "CNS Clock
II" receiver with the hardware sawtooth correction option. I am not yet
seeing the 3 ns or so jitter that Rick claims to have achieved.
It puzzles me. Perhaps I need to install a ground plane under the
antenna, or even a choke ring!
James Maynard, K7KK
Salem, Oregon, USA
SAIDJACK at aol.com wrote:
> Hello Javier,
> this depends on your Rb oscillator, and if you use an external DAC circuitry
> then the DAC and DAC reference thermal sensitivity as well.
> Typically, the Rb including the DAC and DAC reference will be better than
> the GPS for time frames from 1000 to 10000s. The GPS will probably be better
> than the Rb above 10000s and should thus be allowed to correct the Rb at such
> longer intervalls.
> So try starting with a cut off frequency of around 3000s, then try 10K or
> even 20K intervalls for really good Rb's. Keep in mind that your DAC and DAC
> reference likely have 2-5ppm temperature coefficient themselves, and may thus
> affect your Rb accuracy.
> Simple OCXO's seem to work well with a time constant of around 1000s or less
> since they are much worse than the GPS at those times and benefit from the
> GPS corrections above 1000s measurement intervalls.
> time-nuts mailing list
> time-nuts at febo.com
More information about the time-nuts