General Information

In search of the perfect filter

In the beginning was the state-variable filter. This ubiquitous filter deserves mention because the designers of the EML 101 added a pan pot to the output that allowed you to sweep through the lowpass, bandpass and highpass, which almost made it interesting. One of the problems with the state variable is the lack of self oscillation as resonance advances. It also has rather undramatic 12 dB/octave slopes for lowpass and highpass.

A Moog Model 901 A,B and C filter was a dual section filter with 24 dB/octave lowpass and highpass sections. It had a "filter coupler" which could switch the two sections in series and in parallel, in order to realize band pass and band stop filters. It also had a common modulation input which allowed the bandpass and notch filter to be swept externally. It had lots of drama to it because of the 24 dB/octave response which is much darker than the 12 dB/octave types. It also wouldn't oscillate as resonance increased because it used negative regeneration (feedback), which caused peaking but not oscillation.

The Korg MS-20 filter doesn't have a resonance problem because it is a Sallen-Key design with active positive feedback, rather than undamping or negative feedback. As a matter of fact, it squeals like a pig. Some of the MS-20s used a rather poor OTA though, which was both noisy and very nonlinear. It also had rather undramatic 12 dB/octave slopes but the Korg engineers very cleverly improved it by placing lowpass and highpass sections in series, with independent resonance controls. This allows all kinds of filter shapes and the dual, movable, very resonant peaks formed nice vocal like formants. The filter types were fixed though and you could not switch them in series and parallel with out custom modifications.

A Buchla 292 Lowpass Gate has a very dramatic, "woody" or string like sound. It also is a Sallen-Key design but realized with a Vactrol. The Vactrol is important for several reasons, it is a passive device and so does not contribute any active noise like an OTA. Secondly it is slow, it enforces it own slew rate on the modulation sent to it. This might seem like a problem but what actually happens it is behaves like an acoustic instrument, imparting its own characteristic to the tone. In general it has a "cello" like response, which is quite musical. They also don't have any resonance controls and the positive feedback is fixed at unity.

A 24 dB/octave filter is more dramatic than the 12 dB type. You can make a 24 dB filter by putting two 12 dB section in series. As the Moog and MS-20 showed, a dual section filter is just more versatile than a single section filter. But controlling the two sections together is a problem, using both hands just for the filter. So a single control that controlled both sections with one hand would be more desirable.

So the the ideal filter should have the following characteristics:

Controls on the Joy Rider

Note: The MOD knob is also a push-pull switch

Technical Data

Back of the Joy Rider showing power connections. Module uses +/-15 VDC at 100 ma. Three stripped #24 wires are provided for power connection.

• RED = +15VDC
• GREEN = Ground
• BLACK = -15 VDC