pseudo-random & slow random ?

[Don Tillman]

   From: Martin Czech <martin.czech at itt-sc.de>
   Date: Thu, 18 Sep 1997 08:22:25 +0200 (MET DST)

   Thats true. A shift register with length N has 2**N combinations.  So
   it won't take much more effort at all to get a very long sequence.  The
   question is how to set up the feedback and exor stuff so that all
   combinations are walked through.  

I would think the question would *really* be how to set up the
feedback, taps and all to get the best sounding noise source.  I mean
a long sequence isn't useful if it doesn't sound good or if it creates
a massive "whaaa-thunk" every four days.

If you just want a long sequence, simply use a long shift register and
XOR off the last two stages.  You'll do fine.

If you want to play around with feedback shift registers (they're
completely fascinating) it's tivial to write a program in your
favorite language to simulate it, with variable taps, variable length
and so forth.

(If you write it in Java, add a nice UI, make it an Applet, and run it
off a web page, you can share it with everybody easily.)

Or implement it on a spreadsheet.

(One of my earliest computer simulations in high school, around 1972,
was of a feedback shift register displayed on the front panel lights
of an HP 2114A minicomputer.  I think I toggled in the machine code by
hand for that one.)

You can also do an analysis (FFT, etc.) of the output noise signal.
Experiment with resistor taps off of various stages, etc. 

  -- Don

			"Try a vacuum tube!" 
			-- Eric Barbour, synth-diy, 18 Sept 97