[time-nuts] Cross-Correlation Results

John Miles jmiles at pop.net
Sat Aug 4 23:30:44 UTC 2012


> Magnus inspired me to try my own cross-correlation experiment with the
> TimePod today.  I used two fairly normal OCXO frequency standards as the
> reference, and a Wenzel 5 MHz ULN as the device under test.
> 
> By doing single-reference measurements of each OCXO versus the ULN, I
> was able to plot their noise.  Since the ULN has performance near the
> noise floor of the TimePod, I could then test how well the correlation
> works to allow measurements below the noise of either reference.
> 
> Conclusion: it works *really* well.  Plots and a little write-up are at:
> http://www.febo.com/pages/cross-correlation/index.html

Cool stuff.  That's an impressive ULN -- I've never seen one do -125 at 1 Hz
before!

The oscillators used for dual-reference measurements don't have to be
especially great.  The 10811 in your 5061B is probably a lot quieter than
the -155 dBc/Hz floor imposed by the 5061B's buffer amp.  As happened in
your plot, the worse oscillator of the pair has the most influence over the
noise floor at a given offset.  They don't need to be matched in
performance. 

(BTW, something's funky about the way your images are getting scaled... in
both Safari and Firefox, they seem to want to take on the aspect ratio of
the browser window itself, which I've never seen before.  Is that
intentional?  It makes them look oddly squashed unless the browser window
proportions happen to be just right.) 

-- john, KE5FX
www.miles.io




More information about the time-nuts mailing list