one/F noise source ckt?

[Magnus Danielson]

From: Thomas Hudson <thudson at tomy.net>
Subject: Re: one/F noise source ckt?
Date: Wed, 30 Aug 2000 08:36:38 -0700

> 
> There have been some good discussions of 1/f
> noise on the music dsp list. Unfortunately,
> other types of noise made me unsub from that list.
> There are some good links in this message:
> 
> http://shoko.calarts.edu/~majordom/music-dsp/1998b/msg00119.html
> 
> I also have some code lying around based on an
> old Marvin Gardner/Scientific American article
> for generating 1/f random numbers. 
> 
> I've never seen any way to generate it electronically,
> which seems strange considering the first link
> in the above message would lead me to believe
> it is everywhere :-)

The above article is correct in that the 1/f curve is the power distribution
and NOT the amplitude distribution. The diffrence is highly important since it
gives -3 dB/oct instead of -6 dB/oct. Now, a normal RC filter is -6 dB/oct, so
it will not be enougth.

Doing a true -3 dB/oct filter REALLY IS as hard as fetching half an electron!!!

One method to approximate a -3 dB/oct filter which is available for DSPs is to
create a -3 dB/oct frequency distribution and then use the inverse Fourier
transform in order to generate the impulse responce that could be used in a
FIR filter. There are naturally limits with using this method, but it can be
used. Hmm... I wonder how the coefs will look!!!

Cheers,
Magnus