[time-nuts] Higher measurement rate to improve TIC quality?

Magnus Danielson cfmd at bredband.net
Fri Apr 20 20:19:41 EDT 2007

From: Brooke Clarke <brooke at pacific.net>
Subject: [time-nuts] Higher measurement rate to improve TIC quality?
Date: Fri, 20 Apr 2007 17:00:13 -0700
Message-ID: <4629540D.40809 at pacific.net>

> Hi:
> I'm still trying to understand Allan plots.  For example looking at a Time 
> Interval Counter's stability.
> The one shot resolution of the Time Interval Counter is important since it 
> determines the stability floor on an Allan plot.  Can the stability floor be 
> improved by making the measurements faster?  That would move the one shot 
> resolution point to the left thus moving the line down.
> For example if the one shot resolution of a TIC was 10 ns when measuring a 1 
> PPS signal that defines a point at 1 second on the X-axis and at 10 ns on the 
> Y-axis.  Now if the measurement can be made 10 times in each second (the test 
> signal is changed to a 10 Hz square wave), then the point on the Allan curve is 
> now a facter of 10 lower.

The singel shot resolution defined a point in the plot, but considering how
interpolation then works on it, it actually defines a 1/tau line intrinsic to
the measurement which then is being broken by a handfull of perturbations
where the most fundamental is the timebase/reference of the counter.

Other parameters of interest is the amount of measurements being used. Too low
counts will certainly make the uncertainty much higher as you go up into higher
taus. Stefano Bregni have written some highly readable articles on this
available on his webpage.


More information about the time-nuts mailing list