[sdiy] Buchla filters (was DAF)

[Grant Richter]

> 
> Your mentioned trick is the basic but crucial routing in S. Ciani's famous
> 7-up ad soundtrack (sound of pouring limo into a glass). That is, at least
> *I* got an essentially very similar sound with it - since Ciani did it on a
> Buchla 200 system I'm wondering about the FM part, so by the occasion, a
> question to Buchla-players: I suppose she used the bandpass model 291 because
> the 292 lopass-gate circuitry would be too slow for this FM (according to
> Grant Richter, 200 Hz is max).

That would be full power bandwidth. The Vactrol FM response then rolls off
at (I'm guessing) - 6 db/octave. So FM frequency effects go up much higher
than 200 Hz but the depth then becomes dependent on frequency. The is still
some FM happening at 2Khz but not as deep.

This is for the slow Vactrol brand. The Hamamatsus are perhaps 10 times
faster and sound like an OTA. They also have a correspondingly higher FM
range. But the "plucked" effect is produced by the slow Vactrol. If you
build a Borg filter with a Hamamatsu opto-isolator it sounds more like an
MS-20 rather than a Buchla.


> I know that it goes down to 30 Hz with range
> adjustable from a semitone to four octaves. That's quite a special BP which
> as evident in the ad can make for a very decent LP - will someone explain the
> concept behind this to me: I had been thinking the 292 was the "main" filter,
> but then, isn't it, the "frequency gating" idea simulating a quite natural
> overtone behaviour, more intended as the polishing final part in the chain?
> I'm curious.

The 291 is the bandpass filter. Also a Sallen-Key? It has two separate 12
db/oct sections, each with it's own current control. This makes it a "true"
bandpass filter and is equivalent to the Serge variable bandwidth design
(I'm not sure about the rolloff - I think the Serge is also 12db).

There are a number of different ways to think about bandpass filters. A
resonator that has a gain of 1 at all frequencies except the pass band,
isn't technically a filter (it doesn't remove anything. just adds gain at
certain frequencies), but the audio effect is similar.

Also you can have bandpass with -6 db/oct sides with a much quicker rolloff
around the passband when regeneration is increased (or damping decreased in
the case of a state variable). But your passband gain, general rolloff and
rolloff near the passband all are interdependent.

To make a "true" bandpass you should have independent lowpass and highpass
sections like the Moog 901 A, B and C 24 db/oct setup with the filter
coupler. The Buchla 291 has a similar concept using Vactrols with 12 db/oct
independent filters in a fixed series mode. This allows independent control
of center frequency, pass band width and passband gain (labeled Q on the
291).