This is the phase measurements over approx 67 hours. A constant frequency offset of the OCXO have been removed.
The NELS stations are in nice agreement whereas the Chayka station has a totally different idea. I don't know yet if that is because the Chayka signal is very weak and my tracking software doesn't cope, or if that is the look of their signal.
The three NELS stations have very different noise patterns, and I suspect that the 6731M curve shows two micro-steppings, but I am not sure.
The Modfied Allan variance for the four signals. The "hump" at 1000s in the A7499M curve have been indentified and solved, it is due to a non-linearity in the interpolation of the 3rd zerocrossing which results in a 8nsec periodic noise signal as the zero-crossing slides through the sample clock period.
