[time-nuts] Precision Freq Measurement in Short Time

Tom Van Baak tvb at leapsecond.com
Fri Jul 29 17:05:51 EDT 2005


> Hi everyone who has SR620,
> 
> I found a way to measure/compare two frequencies to 1E-12 
> in 1 second by SR620.
> 
> This is desirable in order to measure the Allan Deviation
> for tau = 1 second or adjust frequencies in real time.
> 
> By surfing the net I found an article from SRS describing
> a way how this can be done at:
> http://www.thinksrs.com/downloads/PDFs/Manuals/PRS10m.pdf
> 
> This is the PRS10 manual, refer appendix B for details.

Oh yes, I forgot about that. Thanks for the reminder.

See also the long April 2005 time-nuts thread called
"Difference between HP 5370A and 5370B" where
Brooke Clark mentioned this SR620 technique.

> The principle is:
> If RS620 is used to measure frequency, the RMS jitter is 
> is typically 100ps. However, the jitter will be reduced to 
> 25ps if we measure time interval. If we take 1000 samples 
> per second and average the result, the RMS jitter will be
> less than 1ps (25/sqrt(1000)).

You may be interested to know this averaging
approach is also implemented, internally and
automatically, in the 53132A (and 53131A?).

The single-shot TI spec for the 53132A is 150 ps;
yet it is advertised as a 12 digit per second
frequency counter. As far as I know it also does
this with many (I've not seen the exact rate
published) rapid phase measurements and
then the not quite "instantaneous" frequency is
calculated from the average. Does someone have
an internal design note on the 53132A to share?

This technique works for CW frequency sources
but not 1 PPS time sources.

Also the frequency or phase result you see is
not an instantaneous measurement nor a zero
dead-time measurement so what you get is a
kind of mixture of MDEV and ADEV but not
pure ADEV. But I do it all the time and it's
close enough.

More on the rest of your email later...

/tvb






More information about the time-nuts mailing list