[sdiy] VERY cool/bizarre: Buchla noise

[john mahoney]

> Something tells me the signal coming out of the MM5837 ain't true
random,
> right?  ...

Yep. It merely shuffles a few bits around -- so few bits that it
repeats quite obviously. The pattern repeats in less than a second on
old Mono/Polys. It sounds vaguely like a train. In fact, it's so bad
(All together now: "How bad is it?!") that Korg used analog noise
generators in later Mo/Pos.

It's only a 17-bit shift register. You need almost twice the bits (or
more) to avoid an audible repeat. Here's the MM5837 block diagram:
    http://www.bitshifted.com/images/mm5837-block.gif
(How appropriate is that domain name? Ha!!)
And the datasheet:
    http://www.bitshifted.com/PDF/MM5837.PDF

Oh... The noise *spectrum* of the 5837 is very balanced over the audio
range, when clocked appropriately, so it's good for certain purposes.
Somebody posted a page full of noise spectra, and the 5837 had one of
the flattest. I commented at the time that it was a great example of
how specs don't always tell the whole story.

The non-randomness is rhythmic, as you noted. So, the chip is a champ
or a chump, depending on what you need.
--
john (just a bit shifted...)