Nlfilt is something I concocted some while back. The papers describing
it are on the Bath site:

Non-linear oscillators:
http://www.bath.ac.uk/~masjpf/fractal_rev.html
and
(more specifically nlfilt)
http://www.bath.ac.uk/~masjpf/HK_poster.html

If you look up nlfilt in the helpfile documentation, and follow the
'some hints  for nlfilt opcode) you will find a fairly detailed
distillation of the main points in the papers.


Internally, the non-linear operation is squaring, which means of course
that any values larger than +- 1 are certain to overflow very quickly.
Because of the constant added in some cases, even values below +- 1 will
overflow sometimes. This is of course an expected feature of this sort
of lon-linear algorithm! It also depends on the amplitude of the input,
which, at the very least, needs to be within +- 1.0, and preferably
within half that range.

I cannot answer the question of bugs without looking at the code,
however. I have to admit that since this was added to Csound I have not
done anything further yet with nlfilt myself!

Richard Dobson


vanDongen/Gilcher wrote:
> 
> Another strange effect excists when using nlfilt in an instrument where a
> global a-rate variable is assigned some value (not necessarily the output
> 
> ar nlfilt ain,ka,kb,kd,kL,kC
> 
> When ka or kb <=0 there is no sound generated after the first (I think)
> ksmps burst (there is only short tick. There is also an upper bound I
> think somewhere around 5.
> This effect (bug) doesn't happen when there is no ga variable present in
> the instrument.
> Also, I think in the manual the kL and kC values are swapped in the
> examples (the results were more interesting anyway)
> 
> I would like to hear more about how this filter works. Can anybody post a
> webadress where I can find the original article mentioned in the manual.
> 
> But csound is certainly fun to work with.
> 
> 



-- 
Test your DAW with my Soundcard Attrition Page!
http://wkweb5.cableinet.co.uk/rwd (LU: 6th July 1999)