[time-nuts] Phase microsteppers

Tom Van Baak tvb at leapsecond.com
Thu Mar 10 21:30:43 EST 2005

The next question is, after you get the C-field set
to your liking, how to make further time or rate
adjustments. It mostly depends on what your needs
are. If you are in an isolated lab and need accurate
time a 5061A will do the job nicely at the microsecond
level; and probably stay within a millisecond for years.
You don't have to do a thing, except guard against
accidents like power failures.

If you are a cal lab and want to know you have a
1e-11 or 1-12 frequency standard you also don't
need to do a thing. You can probably turn it on and
off on demand and expect a retrace within 1e-11.

If you are a time nut this probably isn't good enough.
You want to know how far your Cs is drifting in time
and you probably want to get it back on track; you
want to keep your own UTC.

There are several ways to do this. One is to keep
the frequency as is and just make time steps. Any
time your Cs is off by 100 ns or more just bump the
time in 100 ns steps using the 1 PPS sync input
or thumbwheel switches. You can do this as often
as you like; the greater your error tolerance the less
often you need to do it. Most Cs with 1 PPS have a
built-in advance and retard mechanism. The current
model 5071A has both a digital phase and a digital
frequency adjustment built-in.

The second way is to keep a "paper clock". For
many purposes, it is sufficient just to know what
your cesium time or frequency error is -- you don't
physically have to adjust the clock time or rate to
compensate for the amount of error.

With this method you just keep track of your phase
error (e.g., against GPS) with a PC or even pencil
and paper. Then if you make measurements using
your Cs you simply subtract the known frequency
or time error.

I do this with one of my clocks. After several years
of hands-free operation the poor thing is off by a total
of 182.5 microseconds now. But I still use it as my
main 1 PPS source to make TI measurements. I
keep a log of the error. If it's important I just subtract
the 182.5 us from any data I collect when I use that
standard as a reference.

Often making one-line software or Excel corrections
in raw phase data out of a TI or frequency counter is
much, much easier than trying to keep the cesium
standard accurate to a ns.

In summary, since it's not possible to keep a cesium
clock exactly on-time anyway, why bother trying. What
you can do instead is continuously track the error;
mathematically these are equivalent.

Next -- if you don't like accumulated time error, or don't
like time steps, or don't like paper clocks -- if you really
need to adjust your cesium output to the ns level, then
use a phase microstepper.


More information about the time-nuts mailing list