[sdiy] question about a Mankato filter

[David G. Dixon]

> Given the choice between optimising levels for use as an
> oscillator (i.e. all equal to 10 Vpp) and optimising them for
> use as a filter (i.e. equal loudness or whatever), I'd do the
> former.

That was the choice made by Thomas Henry et al.  It was definintely the
right choice for that particular circuit.

> Equal loudness is ill-defined and depends on the input. And
> filter outputs are certainly going into a mixer anyway.

> If all oscillator outputs have the same amplitude, you can switch
> between them without having to tweak the modulation levels every
> time. That's worth another 3.50 USD to me.

But you wouldn't need to spend the extra 3.50 for that; that chip is already
there.  The extra cost (and board size and complexity) would come from
adding the extra components (another 2164 and quad opamp, and a few
resistors) necessary to get the distributed gain.

> Or one could use three op amps for make-up gain and save a couple
> dollars. Inelegant ? Maybe, but in the big picture, an elegant
> interface is worth a lot more than an elegant circuit.

Ah, but you couldn't.  That's the problem.  As you say, you could take the
various outputs out to a mixer, but then you'd have to adjust those levels
between oscillator and filter.  There could be no smooth transition between
the two, except at the third stage (18-dB) output, where the gain is 1.

This is an interesting point, because distributing the gain equally has it's
downside, as well.  Part of the reason I'm revisiting 4-pole cascaded-stage
filters is to develop a new core for my ongoing OB-X-style multimode filter
project.  Using distributed resonance gain builds those gain levels into the
output of each stage, and thereby changes the responses one would get when
combining the stage outputs in various combinations for the different modes,
relative to what one would get from a more conventional 4P filter core.  For
example, in a conventional filter, all of the stages give 0-dB response at
DC regardless of the resonance level.  With distributed gain, these
responses all differ at DC depending on resonance, by as much as 3 dB at
self-oscillation.  Obviously, this changes the character of the various
synthetic filter modes.  Whether they will sound significantly different, or
whether one will sound better than the other, remains to be seen.