[sdiy] Buchla 266 fluctuating voltages

[Magnus Danielson]

From: Scott Gravenhorst <music.maker at gte.net>
Subject: Re: [sdiy] Buchla 266 fluctuating voltages
Date: Sun, 30 May 2004 13:42:31 -0700
Message-ID: <200405302042.i4UKgUi14577 at linux6.lan>

> James Patchell <patchell at cox.net> wrote:
> >At 01:38 AM 5/30/2004 -0700, Scott Stites wrote:
> >
> >         The 5837 is not at all subtle, if you listen directly to its 
> >output, you can plainly hear it repeat about every one or two seconds.  And 
> >I have a tin ear, I guess you could say, because, in general, I can not 
> >tell if audio is being processed in an brand x or y module....they pretty 
> >much all sound the same to be...more of less.
> >
> >         Another alternative, if you need to have pseudo random noise, is 
> >to make your own from some HC164 shift registers and a quad xor gate.  32 
> >bit long sequences sound pretty white....and the repeat rate is hours....
> 
> This is very true.  Someone here, maybe Magnus, pointed out that the spectrum
> of LFSR generated noise is a series of lumps and that the more bits you use
> (with maximal length taps) the more lumps you get and the faster you clock
> the shift register, the closer together they are.  So as you increase the
> number of bits and increase the clock, you simulate a continuous spectrum
> more closely.

This is exactly what I stated. My point was also that long generator
polynomials isn't *that* difficult to implement either.

> Whoever posted this, please do correct me if I'm inaccurate about this.
> 
> I have a chart at http://home1.gte.net/res0658s/electronics/LFSRtaps.html that
> shows maximal tap arrangements for up to and including 168 bits.  I have found
> no data source that goes higher than this.

I posted a list of up to 64 taps. 64 taps is fairly cheap and is more then
enought. More than 168 is also possible and could fairly easilly be generated
if needed.

> What I would find interesting is at what number of bits can people no longer
> tell the difference between the noise quality of one LFSR and x bits more. 
> Eg. is it possible to tell the difference between a 32 bit and a 64 bit LFSR
> clocked at the same rate?  Can we safely say that 64 is ridiculous?

Yeap! Loop-time is long after your synth has disintegrated and definitively
after your loved ones have put you to your final rest after a long and
satisfying life. The noise is dens and the real worry is the upper clock-rate.

> I've built these up to 24 bits and even that sounds very white to my ears
> (not sure what metal my ears are made of...).

Having rasist ears? It seems to differ white from coloured! ;O)

(Bad joke, I admitt, but I could not resist....)

Cheers,
Magnus