[time-nuts] Any Spectracom hackers out there?
John Ackermann N8UR
jra at febo.com
Sun Oct 24 14:22:39 EDT 2004
Brooke Clarke wrote:
> Hi John:
> Is the idea to be able to decode the WWVB time signal by using a known
> Do you have docs on the 8170?
> Have Fun,
> Brooke Clarke, N6GCE
If I understand the docs correctly (I do have a book with schematics),
the 8170 derives its master clock from the WWVB carrier, and steers its
1pps off the decoded WWVB timecode. The microprocessor will advance or
retard the 1pps in either 1ms or 100us steps to bring it in alignment
with WWVB's second tick.
When the receiver loses lock (which seems to happen every couple of days
due to diurnal effects) and the crystal free-runs, the time will jump by
a few milliseconds and it takes quite a while for it to get back in
step. (My guess is that the resting frequency of the crystal when it's
out of lock is far enough off 10.000MHz to cause a sudden jump, which is
then corrected when lock returns a few minutes later. But getting the
time synchronization back takes quite a bit longer due to the long time
constant of its loop.)
My thinking is to replace the 10MHz phase-locked signal with a local
one, and then have just the time-setting PLL at work. If the WWVB
signal is lost, the thing won't jump due to the frequency change. Of
course, doing this loses the ability to correct for errors beyond a few
milliseconds, but given the stability and low offset of the Rb, that
shouldn't be a real problem.
Spectracom had an option to automatically switch to an external
reference when lock is lost, and I could just duplicate that circuitry,
but I'm wondering if, given the availability of the high quality local
reference, it's worth doing the switching, rather than just running on
the Rb all the time.
More information about the time-nuts