I have an FTS 1050A quartz oscillator that performs, except for one parameter, far better than it has any right to.

Here is a plot of its stability compared to an Oscilloquartz 8607-008 BVA oscillator, which has superb short term performance, and an HP 5065A Rubidium standard that performs very well in the medium term. By blending the two plots at their crossover point, you can get a picture of the FTS performance.

The published stability specification is 1x10-12 from 1 to 100 seconds. This unit beats that by almost an order of magnitude, and stays in the 13s out to about 50,000 seconds.

This oscillator also has no discernable aging trend. This plot shows its relative phase versus the HP 5065A over 14 days, with the frequency offset removed:

The one place where this standard falls down is in phase noise performance. As shown in the plot below, while this FTS 1050A greatly exceeds its published specification of -116 dBc/Hz at 1 Hertz offset (at -125dBc/Hz, it's 5dB better than the Wenzel Ultra Low Noise oscillator!), its noise floor is about 20 dB higher than the spceified -160 dBc/Hz.

Given how quickly the floor is reached (at only about 30 Hz), it looks to me like there's a broadband noise that extends from the carrier all the way out; at 30 Hz offset, this noise source starts to override the normal noise mechanisms.

However, this standard has such superb stability that I don't want to mess with it and possibly break something else. So, I'm going to live with the phase noise.