[time-nuts] Random Walk Noise experiment

Bruce Griffiths bruce.griffiths at xtra.co.nz
Sun Feb 13 09:06:22 UTC 2011


ehydra wrote:
> I think the confusion is now perfect:
> http://www.mikrocontroller.net/topic/207061#2059725
>
> Let Google translate it from german to your language.
>
> Does the difference come from voltage vs. power spectrum?
>
Yes, integrating the power spectrum of white noise produces flicker 
noise whilst integrating it twice produces random walk noise.
In practice integrating the power spectrum requires implementing 
fractional order (=1/2) integration of the signal  ( voltage or current) 
and single integration of the signal is equivalent to double integration 
of the power spectrum.

Bruce
> Magnus Danielson schrieb:
>> On 12/02/11 21:02, Bruce Griffiths wrote:
>>> Flicker noise is not the same as random walk noise, the spectra differ.
>>> Using an AC coupled generator (eg a sound card) filters out the low
>>> frequency content.
>>>
>>> Zeners and transistors (biased at low current) can be used to generate
>>> flicker noise directly at least for low frequencies where it dominates.
>>> Generating random walk noise is more difficult, integrating white noise
>>> is one technique that can be used (at least in principle).
>>
>> Of course... *head-slapp*
>>
>> white noise has a flat power spectrum
>> flicker noise has a power spectrum of slope f^-1
>> random walk noise has a power spectrum of slope f^-2
>>
>> For random walk you need to do integration. If you do it in analogue, 
>> care in low-frequency cut-off comes in and below it you will have 
>> white noise. For digital it's a trivial, but you may end up with 
>> digital wrap-around but doing a low-frequency leakage you avoid it 
>> and end up with the same situation as in the analogue domain.
>>
>> So expect there to be a frequency limit for it if synthesized.
>>
>> Cheers,
>> Magnus
>






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