Hi Martin,
>I have a question I've been wondering about for a while, and after reading
>the following, it seems the right time and place to ask it:
>
>Jon (I think) wrote about the filters on a new synth:
>
>"The ION is noticeably missing any Sequential type filters. However, the
>Oberheim filter types were, IMO, the standout filter types on the ION. The
>Roland filters were nice too but for my money the Oberheim filters are VERY
>impressive...."
>
>Now, I don't know much about sound synthesis at all, but I do know that what
>a filter does is remove certain frequencies (in subtractive synthesis at
>least, which is all I know about). This being the case, in what way can the
>filters on some machines be said to sound "warmer" or "better" than others,
>when all they're doing is taking away some of the original "full" sound
>(which may well be "warmer" or "better" than the "original full sound" of
>another synth)?
>
>I'm sure I'm missing some part of the equation. There's resonance, for
>example, which although I think I know what it does (boosts the frequency
>around the cut-off point, no?) I'm not savvy enough to add in to my question
>about filters.
In theory you are correct: a filter allows certain frequencies to pass, and blocks others. In practice, though, there is no such thing as a perfect filter. Every filter design has tradeoffs, and the different filters used by Moog, ARP, Sequential, Roland, Oberheim and others all had sound characteristics that were based on these tradeoffs and the electronics used to make the filter. For one thing, there's the "steepness" of the filter; i.e., how quickly the filter changes from passing frequencies to blocking them. In electronics, this is measured in dB/octave, and is often referred to by the number of "poles" that the filter has. (Each pole is 6dB/octave.) Filters with less poles are easier to design and control -- certainly, in the analog days -- but less poles means that the filter allows a greater amount of unwanted frequencies through.
Then, as you mentioned, there's resonance. In our perfect filter, resonance doesn't even occur, but in any real-world filter with at least two poles, it can. Because the resonance adds an airy quality to the sound, and can even be used to create a pitch of its own in extreme cases, analog synth players came to depend on this effect to add character to the sound.
Then there are the esoteric properties of the filter like "warmth" and "metallic-ness." These are harder to quantify, but have to do with the fact that as different frequencies pass through a filter, the phase relationships of the frequencies change, both intentionally and unintentionally. Plus, there are unplanned byproducts of the circuitry that makes up the filter (uneven frequency response, oscillation due to amplifier feedback, component overdrive, etc.) If one were to program a virtually identical sound -- same base waveform, cutoff frequency, style of filter, number of poles -- on a Minimoog, ARP 2600, and Prophet 5, the result would be audibly different, because each synth's filter changes the waveform subtly during the filtering process.
In addition, in the digital world there is another factor in the digital resolution of both the native audio system and the filter modulation. Low-resolution digital systems are described as "grainy" (analogous to photo quality) and digital modulation sometimes includes discrete jumps between points which can be heard by the ear and seen on the oscilloscope; these effect are sometimes referred to as "stairstep" or "zipper" noise.
This is a very brief description of a complex subject, but I hope it gives you a little insight into the various effects that filters add to the sound.
Regards,
-BW
--
Bruce Wahler
Ashby Solutions™
http://music.ashbysolutions.com978.386.7389 voice/fax
bruce@...