[time-nuts] Frequency Stability of An Individual Oscillator:Negative Values?
Tom Van Baak
tvb at LeapSecond.com
Thu Apr 22 22:03:43 EDT 2010
Right, when working with 3-cornered hats, negatives show
up now and then, especially if your adev calculations are
based on tens or hundreds of points instead of thousands
or tens of thousands. Or if the runs are made with different
equipment or in different environments. Also, I remember
the closer any two of the DUT are in stability the more likely
you'll run the negative situation. It's no big deal, really. To
get a textbook perfect 3-hat for a wide range of tau requires
some work; not all random triplets of DUT will give you this.
Remember too that each ADEV calculation has an implicit
uncertainty. If you plot error bars you will see more clearly
why every 3-hat doesn't always work like you want.
For more information on 3-hat see this the Riley page:
More information about the time-nuts