[time-nuts] Where does the Z3801A 1 PPS come from?

John Ackermann N8UR jra at febo.com
Tue Mar 15 16:12:38 EST 2005


Thanks for the report, Tom.  I'm leaving on a business trip tomorrow, 
and have been slightly frantic for the last evening or two, so haven't 
been able to advance my testing further.

It makes much sense that the divider and synchronization circuitry would 
be in a PGA; it's the perfect tool for something like that.  And I'm not 
going to stand up and defend my '390 divider chain!

You mentioned earlier a spec'd jitter of the 58503 PPS down in the 
picoseconds.  Do you know what that's referred to (ie, second-to-second, 
or long term)?  I'm seeing several tens of ns of noise in both my 
Z3801As over timescales of hours to days.  If the '503 holds to 
picoseconds, it's doing something much more advanced than the '3801A.

John
----
Tom Van Baak wrote:
> Hi John,
> 
> I have 40 hours of data now on the phase difference
> between the 1 PPS and 10 MHz outputs. Attached
> is a plot using a similar scale as yours. With 5 min
> averaging the standard deviation is under 10 ps. So
> I'm convinced it's a simple divider.
> 
> To further confirm this I pulled the antenna cable for
> a few hours to see if both outputs keep the same
> relative phase even as the 1 PPS drifts in phase and
> the 10 MHz drifts in frequency. They do! Looking at
> the plot you can't even tell when I did it.
> 
> The same is true coming out of holdover, during fine
> frequency adjust and all. This would suggest HP does,
> in fact, use a temporary frequency offset as a means
> to fine adjust the phase within the 100 ns period.
> 
> As a final test I power cycled (!) the receiver during
> this run and, sure enough, the 1 PPS and 10 MHz
> keep their tight phase difference. The divider is
> probably jam reset to the right 100 ns once after
> the GPS 1 PPS and OCXO are stable and then
> the frequency tweaked to the approximate ns.
> 
> So my conclusion is the 1 PPS output is exactly
> as stable as the 10 MHz output and both are
> derived from the OCXO.
> 
> Bill pointed out PCB traces going to/from U32 so
> that PGA must contain, among other things, the
> 10 MHz to 1 PPS divider and output buffer.
> 
> In the old days one would use a string of 7490's
> to implement a divider. I imagine the modern way
> to do it in the 90's was a PGA (7 synchronous
> resettable decade counters can't be that hard).
> 
> As you know I use a PIC-based divider, but that
> just shows I'm more of a software person than
> hardware. ;-)
> 
> Anyway, thanks again for bring this query up. It
> turned out to be interesting. I don't think many
> people would notice it because even with your
> tempco (?) the difference was just a ns or two.
> 
> /tvb
> http://www.LeapSecond.com




More information about the time-nuts mailing list