Junction Noise

[Magnus Danielson]

From: Tim Ressel <Tim_R1 at verifone.com>
Subject: RE: Junction Noise
Date: Wed, 6 Sep 2000 13:50:56 -0700 

> Well, you math weenies have one on me. I can tell you that I have studied PRN
> sources with FFT analysers, and the output falls into well defined buckets. In
> between the buckets there are definite holes: frequencies with no energy. My
> idea was to frequency modulate the clock with another unrelated PRN. This would
> smear out the energy and fill in the holes. This I believe would make for better
> "noise".

It is definitly so that the noise from there PRNGs is not *continous*, it isn't
and we should not be fooled to beleive it is. It consist of a number of evenly
spread frequency spikes. There are 2^N - 1 of these frequencies spread from DC
up to half of the clock frequency, so getting N higher will make the spectrum
denser, thus, more frequencies will occur over the same span. Each new D
flip-flop is worth doubled amount of frequencies.

So, the question is really, how dense would it have to be before we consider
it just as good as any other noise source?
Adding length is fairly cheap, dropping in a million more frequencies between
two frequencies is cheap, it is just a matter of knowing the "magic" tap
numbers. Luckilly, Synth-DIY seems to be the list to call on to get those ;D

> Does it ever bug anyone that we are discussing the vageries of *noise* quality?

Hum?

Oh, is this thread keeping the list "noisy"?

Cheers,
Magnus