The MSR atomic clock has drifted sufficiently in the past 6 months to reach the edge of its 1uS tolerance, causing 1PPS Timing errors.

Note this is a gradual drift as opposed to the major jumps we had seen in 2023/2024. Those jumps were large fractions of a second each, requiring rsyncing the atomic clock back to the timing master. Daniel says this is not needed in this case of a gradual few microseconds per year.

I have adjusted the nominal time difference for the atomic clock from 0ns to 1000ns, keeping the tolerance at 1000ns.

Attached trend shows that the timing master itself is not the cause of the drift, as verified by the end station CNS-II GPS independent clocks.