[time-nuts] FE-5680A results

Murray Greenman denwood at orcon.net.nz
Tue Nov 29 12:43:42 EST 2011

There's a very nice way to separate diurnal thermal sensitivity from long 
term ageing. It applies equally well to Rb sources as to OCXOs, which is 
where I've used it.

What you do is set up to make frequency observations against a more stable 
reference (of course in this case a GPSDO or a similar), and then make 
several observations per day for a week or so, preferably at or about the 
same time of day each day. Then you plot the results using a spreadsheet, as 
ppb offset on the vertical axis and time of day (NOT total elapsed time) on 
the horizontal axis.

What will result after several days of observations is either a zig-zag or a 
spiral, depending on the relative sizes of the thermal and ageing effects. 
You can extract the ageing rate out quite nicely by adding a column with an 
estimated ageing rate which you add to the measured offset. If you then plot 
this against time of day you should end up with a static rough retraced 
circle or line when the estimated offset is correct. The width and height of 
the result gives you an estimate of the thermal sensitivity with time of 

If you are able also to plot against ambient temperature instead of time of 
day, the same process applies, and you can extract the thermal sensitivity.

Murray ZL1BPU

More information about the time-nuts mailing list