2018-06-25 16:17:51 UTC
One of my 5065a's has been running for several years and I have been
measuring the offset relative to a 5071a cesium (in Leo Hollberg's lab at
Stanford) since 1/1/18. I have been surprized to see that the ageing rate
is 10-100 times lower than the spec of 10(-11) per month. See the plot. One
way to look at the graph is to say that the 5065a has aged from about -397
microHertz in early January to about -392 microHertz in late June, or
roughly 5 microHertz in 5 months. The measurements were made at 10 MHz, so
1 uHz is 10(-13). That works out to an ageing rate of 10(-13) per month.
However, this unit was turned off for 40 minutes on day 92 (April 2, 2018)
to replace C23 and C24 in the A11 board. So maybe there's a small retrace
offset between the measurements before and after this.
The data consists of 418 points which are 10,000 second averages, which are
computed from runs varying in length from 1 day to 12 days. Most were 3 day
runs consisting of 10,000 points at 25 second intervals.
The gaps in the record are when a "new" 5065a from Corby was being tested.
This showed a much more standard warmup drift of slightly less than 10(-11)
over 5 days and a steady ageing rate of a fraction of 10(-11) per month.
The Rb was allowed to stabilize for 12 hours before the measurements began.
This is shows that the instrumentation, which was exactly the same for both
measurements, is not to blame. It was a sine-wave fitter but any counter
should give the same results.
Does anyone else get very low ageing rates from their 5065a's?