The Science of Synthesis - Part 6 - Voltage Controlled Filters

A voltage-controlled filter (VCF) is a processor, a filter whose operating characteristics (primarily cutoff frequency) can be controlled by means of a control voltage applied to control inputs. It can be considered to be a frequency-dependent amplifier. Although popularly known for their use in analog music synthesizers, in general, they have other applications in military and industrial electronics.

Following the oscillator's mixer section are the filters for sculpting the previously created signal. In the synthesizer world, if the oscillator's signal is thought of as a piece of wood that is yet to be carved, the filters are the hammer and chisels that are used to shape it. Filters are used to chip away pieces of the original signal until a rough image of the required sound remains.

This makes filters the most vital element of any subtractive synthesizer because if the available filters, are of poor quality, few sound sculpting options will be available and it will be impossible to create the sound you require. Indeed the choice of filters combined with the oscillators waveforms is often the reason why specific synthesizers must be used to recreate certain 'classic' dance timbres.

The most common filter used in basic subtractive synthesizers is a low-pass filter. This is used to remove frequencies above a defined cut-off point. The effect is progressive, meaning that more frequencies are removed from a sound, the further the control is reduced, starting with the higher harmonics and gradually moving to the lowest. If this filter cut-off point is recused far enough, all harmonics above the fundamental can be removed leaving just the fundamental frequency. While it may appear senseless to create a bright sound with oscillators only to remove them later with a filter, there are several reasons why you may wish to do this.

• Using a variable filter on a bright sounds allows you to determine the color of the sound much more precisely than if you tried to create the same effect using oscillators alone.
• This method enables you to employ real-time movement of a sound.

This latter movement is an essential aspect of sound design because we naturally expect dynamic movement of sound throughout the length of the note.

Using our previous example of a piano string being struck, the initial sound is very bright, becoming duller as it dies away. This effect can be simulated by opening the filter as the note starts and then gradually sweeping the cut-off frequency down to create the effect of the note dying away.

Notably, when using this effect, frequencies that lie above the cut-off point are not attenuated at right angles to the cut-off frequency; therefore, the rate at which they die away will depend on the transition period. This is why different filters that essentially perform the same function can make beautiful sweeps, whist others can produce quite uneventful results.

Action of a low-pass filter

When a cut-off point is designated, small quantities of the harmonics that lie above this point are not removed completely and are instead attenuated by a certain degree. The degree of attenuation is dependent on the transition band of the filter being used. The gradient of this transition is important because it defines the sound of any one particular filter. If the slope is steep, the filter is said to be 'sharp0 and if the slope is more gradual the filter is said to be 'soft.' To fully understand the action of this transition, some prior knowledge of the electronics involved in analogue synthesizer is required.

When the first analogue-synthesizers appeared in the 1960s, different voltages were used to control both the oscillators and the filters. Any harmonics produced by the oscillators could be removed gradually by physically manipulating the electrical current. This was achieved using a resistor (to reduce the voltage) and a capacitor (RC) circuit. Because a single RC circuit produces a 6dB transition, the attenuation increases by 6dB every time a frequency is doubled.

One RC element creates a 6dB per octave 1-pole filter that is very similar to the gentle slope created by a mixing desks EQ. Consequently, manufacturers soon implemented additional RC elements into their designs to create 2-pole filters, which attenuated 12dB per octave, and 4-pole filters, to provide 24dB per octave attenuation. Because 4-poles filters attenuate 24dB per octave making substantial changes to the sound, they tend to sound more synthesized than sounds create by a 2-pole filter; so it's important to decide which transition period is best suited for the sound. For example, if a 24 dB filter is used to sweep a pad, it will result in strong attenuation throughout the sweep, while a 12 dB will create a more natural flowing movement.

The difference between 12dB and 24dB slopes

Although low-pass filters are the most commonly used type, there are numerous variations including high pass, band pass, notch and comb. These utilize the same transition periods as the low-pass filter, but each has a wide different effect on the sound.

Action of the high pass filter

A high-pass filter has the opposite effect to a low-pass filter, first removing the low frequencies from the sound and gradually moving towards the highest. This is less useful than the low-pass filter because it effectively removes the fundamental frequency of the sound, leaving only the fizzy harmonic overtones. Because of this, high-pass filters are rarely used in the create of instruments and are predominantly used to create effervescent sound effects or bright timbres that can be laid over the top of another lowness sound to increase the harmonic content.

Action of the Band select filter

The typical euphoric trance leads are a good example of this, as they are often created from a tone with the fundamental overlaid with numerous other tones that have been created using a high-pass filter. This prevents the timbre from becoming too muddy as a consequence of stacking together fundamental frequencies. In both remixing and dance music, it's commonplace to run a high.pass filter over an entire mix to eliminate the lower frequencies, creating an effect similar to a transistor radio or a telephone. By reducing the cut-off control, gradually or immediately, the track morphs from a thin sound to a fatter one, which can produce a dramatic effect in the right context.

If high- and low-pass filters are connected in series, then it's possible to create a band-pass, or band-select filter. These permit a set of frequencies to pass unaltered through the filter while the frequencies either side of the two filters are attenuated. The frequencies that pass through unaltered are known as the 'bandwidth' or the 'band pass' of the filter, and clearly, if the low pass is set to attenuate a range of frequencies that are above the current high-pass setting, no frequencies will pass through and no sound is produced.

Band-pass filters, like high-pass filters, are often used to create timbres consisting of fizzy harmonics (Figure 1.15). They can also be used to determine the frequency content of a waveform, as by sweeping through the frequencies each individual harmonic can be heard. Because this type of filter frequently removes the fundamental, it is often used as the basis of sound effects or lo-fi and trip-hop timbres or to create very thin sounds that will form the basis of sound effects.

Although band-pass filters can be used to thin a sound, they should not be confused with band-reset filters, which can be used for a similar purpose. Band-reject filters, often referred to as notch filters, attenuate a selected range of frequencies effectively creating a notch in the sound - hence the name - and usually leave the fundamentals unaffected. This type of filter is handy for scooping out frequencies, thinning out a sound while leaving the fundamentals intact, making them useful for creating timbres that contain a discernible pitch but do not have a high level of harmonic content.

Action of the notch filter

One final form of the filter is the comb filter. With these, some of the samples entering the filter are delayed in time and the output is the fed back in the filter to be reprocessed to produces the results, effectively creating a comb appearance, hence the name. Using this method, sounds can be tuned, to amplify or reduce harmonics based on the length of the delay and the sample rate, making it useful for creating complex sounding timbres that cannot be accomplished any other way. Because of the way they operate, however is rare to find these featured on a synthesizer and are usually available only as a third-party effect.

Action of the comb filter

As an example, if a 1kHZ signal is put through the filter with a 1ms delay, the signal will result in phase because 1ms is coincident with the inputted signal, equalling one. However, if a 500 Hz signal with a 1ms delay were used instead it would be half of the period length and so it would be shifted out of phase by 180°, resulting in a zero. It's this constructive and deconstructive period that creates the continual bump then dip in harmonics, resulting in a comb like appearance when represented graphically, as in Figure 1.17. This method applies to all frequencies, with integer multiples of 1 kHZ producing ones and odd multiples of 500 Hz (1.5, 2.5, 3.5 kHz etc.) producing zeros. The effect of using this filter can at best be described as highly resonant, and forms the basis of flanger effects; therefore, it's use is commonly limited to sound design rather than the more basic sound sculpting.

One final element of sound manipulation in a synthesizer's filter section is the resonance control. Also referred to as peak, this refers to the amount of the output of the filter that is fed back directly into the input, emphasizing any frequencies that are situated around the cut-off frequency. This has a similar effect to employing a band-pass filter at the cut-off point, effectively creating a peak. Although this also affects the filter's transition period, it is more noticeable at the actual cut-off frequency than anywhere else. Indeed, as you sweep through the cut-off range, the resonance follows the curve, continually peaking at the cut-off point. In terms of the final sound, increasing the resonance makes the filter sound more dramatic and is particularly effective when used in conjunction with low-pass filter sweeps.

The effects of resonance

On many analogue and DSP-analogue-modelled synthesizers, if the resonance is turned up high enough it will fed back on itself. As more and more of the signal is fed back, the signal is exaggerated until the filter breaks into self-oscillation. This produces a sine wave with a frequency equal to that of the set cut-off point and is often a purer sine wave than that produced by the oscillators. Because of this, self-oscillating filters are commonly used to create deep, powerful sub-basses that are particularly suited to the drum 'n' bass and rap genres.

Notably, some filters may also feature a saturation parameter which essentially overdrives the filters. If applied heavily, this can be used to create distortion effects, but more often it's used to thicken out timbres and add even more harmonics and partials to the signal to create rich sounding leads or basses.

The keyboard's pitch can also be closely related to the action of the filters, using a method known as pitch tracking, keyboard scaling or more frequently 'key follow'. On many synthesizers the depth of this parameter is adjustable, allowing you to determine how much or how little the filter should follow the pitch.

The effect of filter key follow

When this parameter is set to it's neutral state (neither negative nor positive), as a note is played on the keyboard the cut-off frequency tracks the pitch and each note is subjected to the same level of filtering. If this is used on a low-pass filter, for example, the filter setting remains fixed, so as progressively higher notes are played fewer and fewer harmonics will be present in the sound, making the timbre of the higher notes mellower than that of the lower note. If they key follow parameter is set to positive, the higher notes will have a higher cut-off hand, the key follow parameter is set to negative, the higher notes will lower the cut-off frequency, making the high notes even mellower than when key follow is set to it's neutral state. Key follow is useful for recreating real instruments such as brass, where the higher notes are often mellower than the lower notes, and is also useful on complex bass lines that jump over an octave, adding further variation to a rhythm.

Here concludes the sixth part of this post, if you want to know more about acoustic science please read. Rick Snoman's Dance Music Manual (Second Edition) Tools, Toys and Techniques.