Roger Klaveness wrote:
 
> My thougths on self oscilations:
> In analog circuits there is always some noise present (thermic noise at least)
> whichs excites the filter .In the digital world if you have zero you have
> zero,and how much you try to amplify and filter zero you still get zero, so
> you must have something to excite the filter, you could try with some low
> level noise to simulate the thermic noise maybe

Benoit Mandelbrot's discovery of fractals was as a result of work he was doing
on the French telephone system. He was examining the characteristics of noise
in an attempt to eradicate, or at least minimise (obviously analogue) unwanted
signalson the phone lines. It might be an idea to try some fractal noise to
excite the filter. Perhaps the quality of the noise is as important to the
generation of a rich source of self-oscilation as the degree to which it is
low level.
I haven't seen Mandelbrot's results, but I imagine that within this low
amplitude range, there would be self-referential relationships between the
levels (and frequencies) of separate noise components, which lead to fractal
behaviour. The trick is to generate fractal noise....