Substractive synthesis Part II – Filters

Today, I’ll talk about the one thing that no other synthesis method uses: the filter.
As you probably imagine, the filter, does filter something, in this case, the harmonic content produced by the oscillator.

So, basically, the filter shapes the tonal characteristics of the raw sound. In other words, it re shapes the waveforms produced by an oscillator. It can make the sound darker, or brighter. or thinner or fatter, etc.

There are various types of filters, being low pass filter and high pass filter, the most common ones in synthesisers. As their name states, low pass filter, let frequencies lower than the cutoff point pass, and gradually, frequencies higher than the cutoff point start to roll off; on the other hand, the high pass filter, does exactly the opposite: frequencies lower than the cutoff point are rolled off, and the higher frequencies are allowed to pass. In any synthesiser, the cutoff point is variable, and is the user who defines it. Another common filter type is bandpass filter, this, as it name suggest, it allows only a band of frequencies to pass, rolling off higher and lower frequencies, it’s like using a high pass and lowpass filter in tandem. Similar to the bandpass, is the notch filter, this, does the opposite: one band of frequencies is attenuated and higher and lower frequencies pass.

Most filter designs in synths also add an emphasis control. This, as it names states, boosts the cutoff point frequency, making this frequency highly audible and resonant. This is why in many synthesiser, this is labeled as resonance. In some other designs, it’s labeled as peak. Anyway, Emphasis (as in most Moog synthesisers) or Resonance (as in most Roland synths), or Peak (as in most Korg synths), they all mean the same, an extra boost right on the cutoff frequency.
So, lets say that we are dealing with a low pass filter with the cutoff point set at 8khz, we know for sure that all frequencies above 8khz will be rolled off until they disappear. Lets say we will add resonance (or peak, or emphasis), this will boost the 8khz frequency (and some of the near frequencies, too).
In most analog synths, the boost produced by the resonance can even produce a sine wave, transforming the filter in an oscillator. This is called self oscillation, and it’s quite beautiful when used in a musical way, in fact, as we will see later on, you can totally play melodies with a self oscillating filter.

Besides having different filter types, there are also different slopes, some low pass filters will roll off frequencies above the cutoff point for 6 dB per octave, some others for 12 dB per octave, some for 18 dB and some for 24 dB per octave. The most used slopes are 24 dB and 12 dB per octave. This are also called 4-pole filter (24 dB/oct) and 2-pole filter (12 dB/oct).

Almost every synth has a low pass filter. Some others also add a second filter, mostly a high pass filter. Some other synths uses what is called a multimode filter, having at least low pass, high pass and bandpass filter modes that are defined by the user. Every analog filter design has its own tonal qualities, and every major synth is recognisable for its filter. Think about a Roland Tb-303… well, it’s hard not to recognise its sound, and that is not because of the oscillator; the oscillator in the 303 can only produce standard sawtooth or pulse waves, something that pretty much any synth can do, but its filter design is unique, specially when you crank up the resonance. In the case of the 303, it’s a 3 pole 18 dB per octave low pass filter. Now, think of the Minimoog… that absolutely fat sound is hard to confuse with the sound from the 303, and that’s because of the filter; the famous Moog filter is a 4-pole 24 dB per octave ladder low pass filter, and it’s quite unique and musical. Another example of this, is the Oberheim filter, they sound brighter, more polite, more gentle than the Moog, some people say their sound is creamier… Well, they use a 2-pole 12 dB per octave filter. And then you’ve got the classic Yamaha CS-80… think of all those sounds from Vangelis, well, the Yamaha CS-80 used 2 different filters in cascade, a 2-pole 12 dB per octave low pass and, then another 2-pole 12 dB per octave High pass filter; none of them was able of self-oscillation, so, even though you crank up the resonance, filters would not produce sound on their own.
Quite an opposite case comes when talking about the Korg Ms-20, although it has a low pass and a high pass filter set in cascade, and despite they are just 1-pole 6 dB per octave filters, they will self oscillate, and they will start to scream a lot when you start cranking up resonances. And then, there are the Curtis chips used by SCI and nowadays by DSI, this filters are switchable between 4-pole and 2-pole, they are clean, even though in the 4-pole mode it will self-oscillate, it’s a filter somewhat creamy, very musical. Then the Steiner-Parker 2-pole multimode filter, used in many modular systems and recently in the Arturia Microbrute and Minibrute models, it’s also a very musical filter, with lots of character.

Thanks to Dr. Robert Moog, who came up in the late 60s with the Voltage Controlled Filter or VCF, cutoff points can track the keyboard voltages, so, for example, lower notes will sound darker than higher ones. VCF, also, let the user produce sweeps by closing and opening the filter with a knob, also, thanks to VCF, we can modulate the filter with an envelope generator or an LFO, or control the cutoff point by velocity, aftertouch or a mod wheel, or even a expression pedal.

Some designs can make the filter track the keyboard, as I’ve said before, in this designs, if the filter can self-oscillate, you can tune the filter to an specific note and play, like if it was an oscillator. Pretty cool, right?

In some other designs, like the DSI MoPho, the Oscillator can modulate the filter. This is called audio modulation, and it works in a pretty similar way than FM synthesis.

Most filters sound absolutely great if you overload the signal before the filter, this was a trick used pretty much in the Minimoog, where one would take the phones output, to the input of the minimoog, and then the signal gets overdriven by feedback before the filter. Nowadays, almost every modern analog synth has some means to achieve this feedback overdrive without using a cable.

Speaking of audio inputs, a lot of synths have an audio input for you to filter external sources, so, the VCF can be used as a processor, too. You should definitely try that out with drums, or guitars!

So, filter is what gives character to a synth. No analog filter design is perfect, and that’s the beauty of it. In digital synths, when you crank up the resonance, if you sweep the cutoff point you will hear some stepping, some unnatural quantising due to digital limitations… this is something you don’t hear in analog synths (unless of course the filter is digitally controlled or quantised by poor midi resolution). No filter is a bad filter, every filter has its own charm! Oh, BTW… you can definitely create some wah wah style effects using the filter!

See you in 15 days when we talk about the Amplifier!