[time-nuts] Noise types and Allan Deviation

Bruce Griffiths bruce.griffiths at xtra.co.nz
Thu Jan 15 00:04:46 UTC 2009


Magnus Danielson wrote:
> Bruce,
>
> Bruce Griffiths skrev:
>   
>> Tom Van Baak wrote:
>>     
>>> A number of you have asked about Allan deviation recently so
>>> I thought I'd share this handy one-page chart that shows both
>>> ADEV and MDEV for the following noise types:
>>>
>>> - white phase
>>> - flicker phase
>>> - white frequency
>>> - flicker frequency
>>> - random walk frequency
>>>
>>> http://www.leapsecond.com/notes/Exploring_Allan_Deviation.pdf
>>>
>>> It's in high resolution PDF format so if it looks blurry on your
>>> screen either zoom by 2x or just print it on a laser printer.
>>>
>>> /tvb
>>>
>>>
>>>
>>>   
>>>       
>> Tom
>>
>> A table indicating the dependence of the ADEV of these phase noise types
>> on the measurement system bandwidth would also be helpful.
>>
>> There seems to be some general confusion about this.
>> In general, the measured ADEV, MDEV, etc., depends both on the phase
>> noise characteristics of the 2 oscillators being compared and on the
>> measurement system characteristics.
>>
>> Thus without correction for the effect of the measurement system noise
>> bandwidth, 2 different measurement systems measuring the same oscillator
>> pair will produce different results even though both measurement systems
>> are working correctly.
>>     
>
> There is yeat more effects which may compromise the measurement. Length 
> of measurement is certainly an issue. Stefano Bregni have made alot of 
> analysis on this as apparent in his book and in his articles. ITU-T 
> specifies that the measurement windown needs to be at least 12 times 
> longer than the highest tau being measured. Similarly, there is a 
> requirement on lower end of the tau scale. The number of samples used in 
> the gliding window will affect the values computed compared to expected 
> value. The use of drift rate compensation is yeat another issue.
>
> So the lack of annotated bandwidth is just one of several omissions that 
> will tweak data of the chart. There is more issues to be included for 
> real measures. I would point out some other peculiarities in the data.
>
> Cheers,
> Magnus
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>
>   
Hej Magnus

For the range of tau where phase noise processes such as flicker phase
noise, random walk phase noise, etc., dominate the measured values of
ADEV, MDEV etc, then the results are largely independent of the
measurement system bandwidth characteristics. Thus results obtained by
different measurement systems should agree in this region (provided that
the measurement system noise floor is sufficiently small).

Bruce



More information about the time-nuts mailing list