(This article originally appeared in the November 1977 issue of Contemporary Keyboard.)
Back in the days when the only electronics on the bandstand was for the singer (and it wasn't that long ago!), equalizers were used primarily to compensate for deficiencies in the PA system and the acoustic environment. Today, equalizers are used more creatively, to achieve tonal balance in an ensemble of electronic signal sources, to give definition and presence to lead voices, and to tailor tone colors to create unique and novel sound qualities. Don't be misled by the literal meaning of the name "equalizer." Equalizers are actually complex, versatile filter arrays that enable a synthesizer player to greatly expand his or her filtering resources at modest cost. Graphic and parametric equalizers are two widely available types that are useful as synthesizer accessories.
The Basic Resonant Filter Section
Both graphics and parametrics consist of an array of resonant filter sections. The differences between them lie in the number of filter sections and in the type of panel controls associated with each section. The response of a typical single resonant filter is shown in the graph below. The response peak is described by three electrical quantities (parameters): height, center frequency, and width. The response peak shown has a height of 10dB and a center frequency of 1kHz. Note that the height of the peak is measured from a reference level, which is usually called "flat response gain." That is, flat frequency response is the same as a resonant peak of 0dB height. Heights greater than 0dB are called resonances, or peaks; heights less than 0dB are called anti-resonances, or dips.
The measurement of peak width gets a bit technical, so I'll simplify it somewhat. Imagine increasing the peak height as much as possible, then slicing 3dB off the top of the peak. The width of the plateau thus produced is pretty close to what engineers call the bandwidth of the resonance. The peak in the illustration has a width of about one octave.
Changing the height, center frequency, or width of a resonant filter section produces distinct changes in the quality of a harmonically rich sound, which is passing through it. Peaks wider than an octave or so have a bland quality, whereas narrow peaks (less than one-half octave) have a distinctive horn or vocal quality. Center frequency placement is especially important for accurate simulation of vocal or instrumental timbres. A one-octave drop in center frequency (say from 1,600 to 800Hz) can make the difference between a trumpet and a trombone timbre. The height of a resonant peak determines the amount of sound coloration. Furthermore, a negative peak (dip) produces a sound change that is often unrelated to that produced by a positive peak.
A graphic equalizer consists of several resonant filter sections (at least six, but usually ten) whose center frequencies and widths are fixed but whose heights are variable by means of front-panel sliders. The positions of the sliders on the panel roughly replicate visually the composite frequency response of the filter sections, hence the name "graphic." Since the resonance widths are all broad (typically one octave per peak), a graphic equalizer cannot be set up to give highly distinctive frequency responses. It is most useful for broad tailoring of the overall spectrum of a sound source, for instance to fatten up the bass (by raising the peaks in the 100-300Hz region), or to give the sound more definition (by lowering the 500-1,000Hz peaks).
Stereographic equalizers have another potent capability when used with a monophonic signal source. Stereo graphics consist of two identical arrays of resonant filters. Each array has its own input, output, and panel sliders. Feed a sound source into both inputs. Connect the output of one graphic channel to one speaker system, and the output of the other channel to another speaker system. Raise the sliders of the odd-numbered bands and lower the sliders of the even-numbered bands on one channel. Do the opposite on the other channel. Each graphic channel has a "comb" response consisting of several widely spaced, broad peaks. The two channels add up to a flat frequency response. Since successive peaks are fed to alternate speakers, the total sound spectrum is distributed in space. This gives a monster stereo effect. The location of the sound source image will lie between the two speakers. You can actually move the image around by changing which bands are fed to which speakers. The only rule to remember is: whichever bands are up in one channel should be down in the other.
Parametric equalizers contain fewer resonant filter sections than graphics (one to four, as opposed to six to ten), but each section has three controls, for peak height, width, and center frequency. In a well designed parametric, the controls are independent of one another: the center frequency control does not affect peak height or width, etc. Also, the filter sections themselves are independent of one another. This gives a high degree of flexibility in setting the equalizer's composite frequency response. However, the clear visual indication of the graphic equalizer's composite frequency response is missing in the parametric equalizer. When using a graphic, you can think directly in terms of the frequency response curve. When you use a parametric, you have to think in terms of the resonant peaks and dips that give rise to the composite frequency response. Because of this, operating a graphic is a visual activity, whereas setting up a parametric is done strictly by ear.
Parametric equalizers really become useful when you set out to simulate instrumental timbres. Most brass and woodwind timbres are characterized by one dominant resonance. Center frequencies of such resonances range from a couple of hundred Hz (for large instruments such as the tuba) to 2kHz or so (for a high trumpet). Widths range from 1/8 to 1/2 octave; heights are typically 15 to 20dB. The exact settings needed to simulate a given instrument depend on the nature of the input signal, your playing style, and how your ears hear things. To find a good instrumental timbre, start with an unfiltered sawtooth or rectangular wave and set the parametric width to 1/4 octave and height to +15dB. While playing some easy lick, vary the center frequency control until the tone color begins to approach what you're looking for. Next, vary the width and height controls, then go back and fine-tune the center frequency control. You will be able to quickly zero in on the sound that you are looking for.
You may wonder why additional filtering is needed, since nearly all synthesizers already have a voltage-controlled lowpass/resonant or multimode filter. Well, there are two reasons. First, accurate simulation of instrumental timbres requires at least two filters, a fixed filter to simulate the fixed resonance of the instrument, and a dynamic (voltage-controlled) filter to simulate the overtone buildup at the beginning of each note. For instance, to accurately simulate a clarinet attack transient, start off with a square wave, set the synthesizer voltage-controlled lowpass filter to open up (attack) in 0.1 seconds or so, feed the synthesizer output through a parametric filter section with a center frequency of 1,500-1,800Hz—and then play like a clarinet player! Of course, it's not that simple, but it is worth getting into if you want to understand how to simulate acoustic instrument timbres.
The other reason for supplemental filters for synthesizers is that some timbres are characterized by more than one fixed resonance. As an extreme example, you can do a dandy violin simulation with a couple of dozen correctly placed resonances! Vocal sounds are characterized by two or three resonances. So a parametric equalizer allows you to explore timbres that cannot be achieved with a single filter, while freeing your synthesizer filter to perform dynamic spectrum changes.
For more articles by Bob Moog, please visit http://www.keyboardmag.com/analog/1318/moog-monday-hub-page/57761