[time-nuts] Surplus Guidelines, was: Rubidium Standard

[Tom Van Baak]

> When computing a 1,000s tau Allan Dev., do the measurements need to be
> taken back-to-back, ie a 1,000s Tau is 1,000 readings taken in one
> 1,000s window?

Yes, back-to-back. To compute ADEV for tau 1000
seconds you need several or many measurements
1000 seconds apart. The minimum is usually 4, but
you get better results the more measurements you
have. If you ran for one day you'd have about 86 of
those 1000 s back-to-back measurements and from
that you could compute the ADEV quite nicely.

Note you can skip any number of measurements so
in this case you can also compute ADEV for tau
2000 by just looking at every other one, etc.

I'm not quite sure what you mean by 1000 readings
taken in one 1000s window. Do you mean a 1000
measurements, each 1000 s apart, for a run lasting
a million seconds? Or 1000 measurements, each 1
second apart, for a total run time of 1000 s?

> What do you call it if you take 1,000 readings in a 10,000s window and
> compute the Allen Dev. from that?

That sounds like a thousand readings at an interval of
10 seconds to me. From that you can compute ADEV
for any tau from 10 seconds to about half an hour.

See also:
    http://www.leapsecond.com/tools/adev1.htm

> cheers,
> glenn

/tvb