[time-nuts] How I got my FE-5680A to lock in Sydney, Australia
csteinmetz at yandex.com
Mon Feb 10 19:57:27 EST 2014
>Sorry, but I didn't accidentally transpose the 'high' swing numbers.
>That's why I was so confused about the results.
No need to apologize, the facts are the facts! I take it that you
recently received the GPS. If so, it is probably still settling into
its new natural frequency after being off for a while and jostled
around in shipping. It appears to have drifted far enough that it
now scans through 10MHz in the normal position. At some point, you
may find that it no longer locks in the inverted position because it
has drifted too far to scan through 10MHz in that position (according
to your figures, it was right on the edge when you took the data you posted).
>The counter I'm using is one that I designed myself. As an
>alternative to its internal 1MHz XCO timebase (with dividers down to
>1Hz, 0.1Hz etc), it also allows you to use an external source of 1Hz
>pulses like those from a GPS receiver or a GPSDO, or a Rb Freq
>Reference. The dividers operate on these 1Hz/1pps pulses as well, to
>provide gating times of 1s, 10s, 100s or 1000s. It seems to work pretty well.
For short gating times, I'd think that GPS PPS jitter might be a
limitation. (That's the trick of a GPSDO -- the quartz [or rubidium]
oscillator controls the short-term stability, at time intervals where
it is better than the PPS, and the PPS controls the long-term
stability at time intervals where it is better than the
oscillator.) Specs for GPS engine PPS jitter seem to run from a low
of around +/- 5nS to several times that, which would translate to a 1
second stability of 10^-8 to 10^-9 at best. A good quartz oscillator
(or a well-designed GPSDO with a good quartz oscillator) should get
you to 10^-11 or even into the neighborhood of 10^-12. You'll add a
little jitter dividing by 10M, but it shouldn't be anywhere near as
much as the 3-4 decades by which the quartz oscillator beats the PPS at 1S.
More information about the time-nuts