Hi Morrie, I suspect your PM6690 is working correctly. High-end counters give better-looking results when the two frequencies (DUT and REF) are not exactly the same. I don't know if the Fluke documentation mentions it. But there is a short description in the HP/Agilent manuals: http://leapsecond.com/pages/53132/53132-reduced-resolution.gif
See also: https://www.febo.com/pipermail/time-nuts/2013-August/079662.html Remember that you are not measuring the Allan deviation of a OCXO alone -- your measurements are the Allan deviation of the DUT + the counter + the REF. When you get near the noise floor it becomes very important how much each of the three pieces contribute to the final number. The counter noise may differ between time interval, time interval average, and frequency modes. Note also that your 6e-12 number may be skewed by undocumented oversampling or averaging done inside the counter. That is, the true ADEV may be higher than 6e-12. But it sounds like you are just making pair-wise comparisons among OCXO so this is probably not a concern. To make short-term measurements with significantly higher resolution you may have to switch to an instrument designed to measure oscillator noise. A classic example is: http://www.wriley.com/A%20Small%20DMTD%20System.pdf /tvb ----- Original Message ----- From: <[email protected]> To: <[email protected]> Sent: Sunday, September 21, 2014 7:55 PM Subject: [time-nuts] A question about anomolous Alan Deviation Measurements. > I am using a Fluke PM6690 counter measure the Alan Deviation > of various pairs of 10 MHz OCXO's in use in my lab. > > I discovered that I can measure a (tau = 1 sec) low value of say 6.0E-12 > on a good set, but only If the crystals are tweaked so the delta is some > 0.5 ppm. If they crystals are tweaked to be within a mH, then the Alan > Deviation zooms to about 5 times higher. > > I suspect a flaw in my PM5590 unit or in its design, and I guess will > hunt around for something better. > > Does anyone have experience with this phenomena? Does one always > have to separate the clocks being compared to avoid this "beating"? > > Thanks, > Morrie Altmejd _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
