[time-nuts] Cross-Correlation Results

John Miles jmiles at pop.net
Sun Aug 5 09:58:18 UTC 2012

> would an AD measurement also benefit from Cross-Correlation or would we
> need
> 2 cross correlated disciplined ocxos for that? Being able to measure AD
> below that of a single reference would be a dream which could help a lot
> avoiding to buy a H2 maser for low AD at all taus.

Unfortunately, since the process depends on averaging to reduce
differential-mode noise, it can only help with measurements that are
"ergodic" -- meaning that the characteristics of any given block of data
taken during a measurement can be treated as representative of the entire
measurement.  That's not true when you're measuring long-term stability.

In other words, cross-correlating PN analyzers rely on the DUT noise to
remain consistent over time.  That's the opposite of what happens with an
ADEV measurement, where you expect the carrier frequency to change somewhat
over time, and you are actually trying to log the statistical properties of
the drift.  

Now, you *can* use a "cross AVAR" function to remove instrument noise from
an Allan deviation plot, and you can also derive short-term ADEV from a
phase noise plot taken by a cross-correlating instrument.  The TSC
5120A/5125A analyzers do both.  This is only useful for lowering the ADC
noise, though -- it can't take advantage of dual references to measure a
more accurate DUT, beyond (I suppose) a 3-dB improvement over the reference
instability, if you have two H-maser references instead of just one. 

-- john, KE5FX

More information about the Time-nuts_lists.febo.com mailing list