Perrier, It's a good question.
Better setups use various forms of calibration methods to compensate for systematic biases. This can be done for instance by insertion generated side-band signals from a separate generator, where the power and offset frequency is set and a know relationship of what it should be can be calculated, and then the value detected deviates from this and this is then used for corrections. The system noise can also be measured. With this, consistency with known phase-noise levels can be tested from a number of sources. Those sources can be measured using a number of techniques, including cross-correlation techniques. All measurement methods have their flaws, and the cross-correlation has it's flaws as we have been discovering and investigating the last 5 years or so. Getting accurate numbers becomes hard as you reach towards the thermal noise floor, but above it we can solve most of the issues with cross-correlation based setup. Accuracy at the thermal noise floor is very very difficult. You want more details? Cheers, Magnus On 2019-08-23 21:07, Perry Sandeen via time-nuts wrote: > Gentlemen, > A question for my curiosity and education. Traditionally, the standard was > 10X better than what was being calibrated. > > Since this is at the theoretical limit, how might it be calibrated so the > numbers provided are accurate? > Regards, > Perrier > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
