[time-nuts] RFTG GPSDO and other GPSDO questions...

[Alan Kamrowski II]

Hi Everyone,

 

I've had some fun and learned quite a bit trying to get my Lucent RFTG
rubidium module to discipline itself to a GPS module.  It has an interesting
way of doing this that I'm not sure is the best or the worst, but it is
constantly figuring out an average phase error between the gps 1pps signal
and its own 1pps signal generated by dividing the Rb clock.  It can make
both time corrections (to correct the phase of the 1pps generated signal)
and frequency corrections by changing the control voltage on the rubidium.
I've been trying to figure out exactly how it works, but can only guess at
some things.

 

It measures an average phase error over a 5 minute interval right now, and
clearly it is using a 10Mhz clocked device to do this.  It usually bounces
between a phase error of +40nS or -60nS and it can only make time
corrections in 100nS increments.  My theory is that if the average phase
error is large enough that it can adjust for it then it will do so, for
example an average phase error of 150nS would mean that it could adjust by
100nS leaving only 50nS.  It tries to adjust the 1pps (time correction)
every 15 minutes.

 

It also calculates the delta between these average phase errors and I
watched them for an hour and you will see things like +6nS delta followed by
a -6nS delta and adding all the delta's together for the hour I got 0.  I
would think it might actually use these delta.  I would think looking at
this delta for a long period of time might be helpful for it to determine a
frequency correction, but perhaps that isn't how it works.  Maybe it looks
at the adjustment it made (100nS) and the time between adjustments, say 10
hours to determine the frequency adjustment.  I can't think that would be as
accurate because it makes time adjustments on a 15 minute schedule.  You can
also tweak the time correction schedule and frequency correction schedule so
I'm not sure they are relatable to each other or not.  I just know that my
unit has been running for >300000 seconds and still hasn't made a 100ns
adjustment.  It has displayed no time adjustments and three frequency
adjustments each for "0.0000e0".

 

This got me to wondering - how do other GPSDO units work?  Do they take days
and weeks to build up enough data to discipline themselves or are there
faster methods?

 

Thanks,

 

Alan