Tom Henry Tunable Noise Source

Grant Richter grichter at execpc.com
Wed Nov 17 23:16:11 CET 1999


Hi Juergen, Don, Martin,

I have a poster which was put out by the 
Federal Scientific Corporation that shows all this probability
distribution stuff. Titled heavily enough 
"Typical functions in the time and frequency domain
with amplitude probability density and cumulative distribution"
Whew!

I have scanned it in and put it up at

http://www.execpc.com/~grichter/probability.htm

the graphic is large ~111K so wait for it to load.
I think it addresses all Juergens original questions
and is also just a good reference
(I hang it over the workbench)

Grant

----------
> From: Haible Juergen <Juergen.Haible at nbgm.siemens.de>
> To: Grant Richter <grichter at execpc.com>; jhaible <jhaible at debitel.net>;
Martin Czech <martin.czech at intermetall.de>; synth-diy at mailhost.bpa.nl
> Subject: RE: Tom Henry Tunable Noise Source
> Date: Wednesday, November 17, 1999 4:35 AM
> 
> 	>> (3) How is the pattern (distribution ?) of a Sample and Hold
> changed when
> 	>>       you're sampling a white-noise-FMed VCO instead of sampling
> the
> 	>noise
> 	>>       directly ?
> 	>
> 	>I can answer this one for certain.
> 	>
> 	>White noise has a gaussian (bell shaped)
> 	>probability distribution relative to amplitude.
> 	>The further from zero volts the lower the
> 	>probability of getting a sample.
> 	>
> 	>A triangle of sawtooth wave sampled from
> 	>a VCO modulated by white noise has and
> 	>equal probability distribution. All voltage levels
> 	>have equal probability of being sampled.
> 
> Hi Grant, thanks for your answer!
> 
> This is what I was *hoping* for: a means
> to get equally distributed random samples.
> Now if I'm sampling a fixed frequency triangle oscillator
> at a fixed sampling rate, I get equal distribution of
> discrete levels, but they form a regular pattern in time,
> depending on the VCO freq and sampling rate.
> This is obvious for LFOs (the famous OB-8
> sampled LFO waveform), but it even happens with
> very high VCO frequencies. (I guess the sampling
> folds any high VCO freq. partly into the baseband.)
> Now with white noise modulating the VCO freq one
> can destroy the regular pattern quite easily.
> But there's still questions remaining (for option (3)):
> What amount of FM with noise will be enough to get
> no audible repetition in the S&H pattern ? And will
> the equal level distribution of a sampled triangle VCO 
> be in any way disturbed by the noise-FM ?
> What I *fear* is that if you get away from regular patterns
> you might also loose the equal distribution. But it's
> a very nonlinear operation, so I don't know the answer.
> 
> JH.



More information about the Synth-diy mailing list