To get more insight in the performance of a frequency counter when measuring difficult frequencies, a slow (500 s) sweep with center of 10 MHz and a span of +/- 5 Hz was generated The frequency counter is set to 1 s gate time. Both the counter and the signal generator where locked to the same external reference. Attached picture shows the linear residue of the frequency difference. CVS file with the raw frequency measurements also attached. The slow oscillations outside +/1 Hz deviation could be caused by the limited sample speed and DAC resolution of the used signal generator. Between +/- 1 Hz deviation the measurement noise increases as the contribution of the linear regression in the counter seems to decrease. The ADEV calculated after removing the sweep trend is about 2e-11 for tau from 1 to 10 s Outside the +/- 1 Hz deviation the ADEV seems to be well below 1e-11 with tau is 1 s Does this measurement make sense? Any suggestions on how to improve this stress test? Erik.

Hi Indeed DDS generated signals can have some “interesting” artifacts as both the DAC resolution and the numerical resolution get in the way. Exactly how they manifest is very much a “that depends” sort of thing. Folks have various things they do (dithering is one of many) to try to reduce them. Since this stuff is typically locked away inside the signal generator, you really can’t change most of it. About the only thing you can do is to shoot the signal through a narrowband filter. If you can find or build a device with 10’s of Hz bandwidth that might take out some of the problems. The other alternative is to phase lock a cleanup oscillator to the signal generator. One of the recently discussed sub $5 eBay OCXO’s might be an easy thing to get a hold of. Set up a PLL with as low a bandwidth you think will hold lock. Will either approach work? You would have to know a lot about the internals of the firmware in the signal generator to answer that question. Bob > On Jun 7, 2023, at 8:46 AM, Erik Kaashoek via time-nuts <time-nuts@lists.febo.com> wrote: > > To get more insight in the performance of a frequency counter when measuring difficult frequencies, a slow (500 s) sweep with center of 10 MHz and a span of +/- 5 Hz was generated > The frequency counter is set to 1 s gate time. Both the counter and the signal generator where locked to the same external reference. > Attached picture shows the linear residue of the frequency difference. CVS file with the raw frequency measurements also attached. > The slow oscillations outside +/1 Hz deviation could be caused by the limited sample speed and DAC resolution of the used signal generator. > Between +/- 1 Hz deviation the measurement noise increases as the contribution of the linear regression in the counter seems to decrease. > The ADEV calculated after removing the sweep trend is about 2e-11 for tau from 1 to 10 s > Outside the +/- 1 Hz deviation the ADEV seems to be well below 1e-11 with tau is 1 s > Does this measurement make sense? > Any suggestions on how to improve this stress test? > Erik. > <slow_sweep.png><sweep.csv>_______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com