[time-nuts] Clock quality: alternatives to ADEV

Magnus Danielson magnus at rubidium.dyndns.org
Sat Apr 12 10:36:47 UTC 2014

On 10/04/14 19:24, Richard (Rick) Karlquist wrote:
> The trouble with ADEV is that if you average
> a long time it papers over anomalous events
> like crystal jumps.

ADEV is about characterizing noise powers. The better variants such as 
TOTADEV and TheoADEV will be even more efficient at suppressing a 
crystal jump. It's just not the right tool for crystal jumps.

>  An alternative measure
> might be to, instead of averaging, simply
> keep track of the worse case change in frequency
> during 1 sample period.  Sort of like peak jitter
> versus rms jitter.

MAFE (Maximum Average Frequency Error) might be a suitable measure to 
apply, it's the frequency relative to MATIE (Maximum Average Time 
Interval Error).

MTIE (Maximum Time Interval Error) has been used for a long time 
characterize longer term systematics.

Maybe just doing a MFE (Maximum Frequency Error) might be the tool at hand.

Just to get the ideas rolling.


More information about the Time-nuts_lists.febo.com mailing list