Dear all,
Once again I would need some of your lights on the following problem. Really, I have to assess the stability of an oscillator. To do this, I have used a very stable oscillator (MASER) and I have compared its frequency to my oscillator through a frequency counter (sampling performed every seconds). The test was run for several days. Then I have plotted/calculated the Allen variance. The plot looks good but I would like to know the following:
a)Is there a good reference (book, article or else) addressing the measurement errors and their impact on the Allen variance calculation like the following (I have seen some documentations on the net but it is never really addressing these points):
a1)Usually the Allen variance assumes that each measurement is an averaging. Thus, having my frequency sampling set to 1 second, is it valid to calculate the Allen variance for 1 second (or instead the Allen variance starts to be meaningful at 10 sec or 100 sec….)?
a2)The Allen variance must follow some statistics (which one?) and as such what is the error calculation associated to each measurement?
a3)How many points do I need to derive a meaningful Allen variance: for instance if my measurement is one day long (86400 seconds) what is the maximum averaging I can calculate (could it be 10000 seconds corresponds to approximately 9 points or less?)?
a4)It is said that the reference oscillator must be 2 order of magnitude better than the required stability assessment but if the reference oscillator is not that great (only one order of magnitude better) is there some calculation to assess the error associated on the measurement?
That’s all....
Franck
Once again I would need some of your lights on the following problem. Really, I have to assess the stability of an oscillator. To do this, I have used a very stable oscillator (MASER) and I have compared its frequency to my oscillator through a frequency counter (sampling performed every seconds). The test was run for several days. Then I have plotted/calculated the Allen variance. The plot looks good but I would like to know the following:
a)Is there a good reference (book, article or else) addressing the measurement errors and their impact on the Allen variance calculation like the following (I have seen some documentations on the net but it is never really addressing these points):
a1)Usually the Allen variance assumes that each measurement is an averaging. Thus, having my frequency sampling set to 1 second, is it valid to calculate the Allen variance for 1 second (or instead the Allen variance starts to be meaningful at 10 sec or 100 sec….)?
a2)The Allen variance must follow some statistics (which one?) and as such what is the error calculation associated to each measurement?
a3)How many points do I need to derive a meaningful Allen variance: for instance if my measurement is one day long (86400 seconds) what is the maximum averaging I can calculate (could it be 10000 seconds corresponds to approximately 9 points or less?)?
a4)It is said that the reference oscillator must be 2 order of magnitude better than the required stability assessment but if the reference oscillator is not that great (only one order of magnitude better) is there some calculation to assess the error associated on the measurement?
That’s all....
Franck
