Simple but important Pre-Resonace on SSM and Envelope Slope Questions

[jhaible]

> I was just trying to emulate my DK-600's 2044 SSM filter sound on my
Waldorf
> MicrowaveII. It comes close however the Self resonance on the MWII hits
> very hard and fast over a range of about 1/20th knob sweep. It goes from
> almost nothing to 100%. On the DK-600, this pre-selfresonance range is
about
> 1/4 of the sweep and is simply the best sounding range to be in. In this
> area the filters starts to go high pass and hints of self resonace cause
the
> sound to sparkle.
>
> Can anyone explain exactly what is happening to the frequency spectrum as
> self resonance is approached ? Can you reference this to filter circuit
> behaviour if possible?

Im not familiar with either instrument, so I can only make some general
remarks.
There are so many degrees of freedom in filter design, that I would be
surprised
to find two different ones with similar behaviour, rather then being
surprised
about them being different.

Laying out the mechanical range of a potentiometer for pleasant acoustical
behaviour of resonance is one of the easier tasks - and still you find it
neglegted on some instruments. In case of the SSM2044, the data sheet
suggests a reverse audio taper pot for optimal behaviour. In a patchable
synth, with CVs going thru ADCs and DACs, you can mimic such courves
in software rather than using special pots, of course. Given that your
DACs have the right resolution, and given that the designer cares for
details like that to start with.
Now that's when we have the same filter chip in a circuit, and similar
input level etc. But I'm in doubt that the Waldorf synths use a 2044 filter
- do they ??
When we start to compare different filter topologies, there are a million
possibilities, and sometimes controveral design goals. Do you want
the self oscillation to override the input signal ? Or do you want to hear
the input signal almost unaltered even when the filter screams on the top
end of the audio spectrum ? You will desgign the mixer stage of input
signal and feedback signal accordingly: Use the plain filter input stage,
use an extra adder stage. Limit the signal in the extra stage or don't.
Introduce a compensation path for the input signal against bass loss or
don't.
Make the feedback loop frequency dependent, or don't. In case you make it
frequency dependent, make the corner frequeny fixed, or make it change
with feedback gain ...
Ok, that's what came to mind - I'm sure there are more points.
(And I was only talking about the classical 4-pole polygon filter.)

> Also, the MWII has fast envelopes but they have linear slopes. I prefer
the
> slopes of other synths that I guess are log/exp. Can anyone explain why
> these slopes sound better. Is there an advantage of one over the other?

I also prefer exponential slopes for normal playing. Why ? Not sure.
I guess it's what the ear (and brain) has learned to prefer over the
years, and expo shapes are also "smooth" in a way that linear slopes
aren't.
There are exceptions, though. Linear slopes can be more handy for some
weird modulations, so if you don't use them as standard for VCAs and VCF,
they can truely have advantages. (And no, exponential VCF and expo VCA
plus linear EGs do *not* perfectly emulate exponential EGs.)

JH.