Signal Processing

Mixers A mixer can be thought of as the "heart" of a sound system. It is here that the operator controls levels, adds equalization and effects, and routes the audio signal to speakers. It is very very important.

A mixer, or console, or mixing desk, or mixing console, or sound board* generally does the following: it takes a bunch of inputs, through input channels, screws around with them (i.e. levels, eq, effects, etc), and sends them to any number of outputs, referred to, usually, as buses. [*: Darrell's note: "sound board" is a good term to use when talking to non-sound people, like lighting designers and carpenters.]

INPUT CHANNELS

There are generally two types of physical input connections-- balanced and unbalanced connections, with two types of electrical connections-- line-level, and mic-level. In pro-audio / sound reinforcement, mic-level connections will be balanced connections using, typically, XLR connectors. Line-level connections may be either balanced or unbalanced, using either XLRs or 1/4" input connectors. On some mixers, you may find the designation "Lo-Z" and "Hi-Z." These terms refer to impedance. Pro-audio microphones are "Lo-Z"-- low impedance, which means, essentially, that you can have long lengths of cable, provided it's balanced, without any detrimental effects. Cheapo Radio Shack microphones and things you buy for your everyday classroom tape deck are "Hi-Z"-- high impedance, which means, essentially, that if your cable is longer than about twenty-five feet, you're gonna get hum and some other nasty side effects. Generally, Hi-Z will mean unbalanced, and Lo-Z will mean balanced. Avoid using unbalanced microphones at all costs; unbalanced line-level gear will be fine. 

PREAMPLIFIERS ACCORDING TO YAMAHA

Preamplifiers are used to boost the weak output levels of microphones to levels those that are about line-level. The preamplifier is the first active stage, the first electronic circuit that processes the microphone signal connected to a mixer. Preamplifiers generally are designed to operate within a certain gain range. When you operate the trim / gain control on a mixer's input channel, you generally are adjusting the gain of the preamplifier. If operated at unity gain (no amplification), many preamplifiers will become unstable and may exhibit increased distortion or a tendency to oscillate. Therefore, design engineers will generally provide attenuator pads before and/or after the preamplifier. This enables the signal to be knocked down so that the preamp can always be operated with some gain. 

PREAMPLIFERS SIMPLIFIED

Each input channel usually has the following: a gain control, which controls the level of the incoming audio signal into the channel's preamplifier. If the channel is accepting a microphone, you will need the preamp. If it is accepting a tape deck / other sound source, you will still need it but not to the same extent. For instance: a microphone may put out an audio signal level of -60 dBu. This is not "strong" at all-- consider again how a microphone works-- the diaphragm moves a little bit and a small current is produced. The mixer will not be able to work with this signal without producing a tremendous amount of noise in the signal path. Besides, to drive amplifiers or effects, the signal needs to be a certain level-- or at least within a certain threshold. A pro DAT/CD deck outputting "pro" output levels will produce a signal level of +4 dBu. That's quite different from -60 dBu. The preamp will most likely not be needed at all; if it is used to too great an extent, one will overload the preamplifier and clipping will result. Clipping happens when the input signal is too strong for the device it is driving; the device will "clip" the audio signal above this threshold and distortion will occur. When using microphones, if the gain control is not properly set, loud SPL levels into the microphone will also cause the preamp to clip. When the gain control is not enough, we use an attenuator pad. An attenuator pad, or just "pad," will dampen the audio signal even further than the gain control can, useful for very loud line-level signals. From the preamp, the audio signal may be routed into a couple of different places: the equalizer bus, and a pre-fader auxiliary bus. We will discuss the aux buses later. 

INPUT CHANNELS, CONTINUED

Equalizers were developed back in the days of ancient telephone systems, when long-distance transmission would result in some frequencies being lost. Equalizers took this signal, and boosted the offending frequencies-- hence the name. Nowadays, we use equalizers to equalize sound systems to produce a "flat" response in auditorium spaces that do not have flat responses. (Flat response implies that no frequencies are boosted or cut. Any space will have a given frequency response that may result in boosted or cut frequencies. A space that boosts certain frequencies will result in those frequencies feeding back when an open mic is used-- more on this in Section Thirteen). We also use equalizers (or at least I do) to equalize people's voices when they are too annoying-- when voices are too screechy, use the channel EQ to cut the offending frequencies; or to equalize tonal loss in, say, cassette tapes. Basically, an equalizer is a fancy tone control. 

EQs come in very many formats-- from outboard equipment with thirty-one bands to a simple tone control on your car radio. Mixers generally have two-to-four band equalization. Most mixers have a high-frequency eq, which boosts or cuts around 1 kHz; a low-frequency eq, which boosts or cuts around 100 Hz; and occasionally, in nice mixers, some sort of midrange control. Many sound reinforcement mixers have two pseudo-parametric equalizers for the midrange frequencies. With these equalizers, one can select approximately what frequency one wants to boost/cut using one knob, and can select how much boost/cut with another. Generally there is one for higher-midrange frequencies (250 Hz - 12 kHz), and one for lower-midrange frequencies (30 Hz - 300 Hz). Learn how to use these. These are very very handy.

From the equalizer, the audio signal generally goes to the channel fader. The channel fader controls the level of the channel as it gets routed to different outputs-- left and right, for instance. A pan control is also present. The pan pot (short for "panoramic potentiometer") spreads the signal over left-right, or alternately, different subgroups, which are discussed later on. Occasionally adorning the channel fader area will be a channel mute switch, and the very important PFL switch. The Pre-Fader Listen switch enables the sound engineer to preview the channel through a monitor bus (headphones, wedge monitor) before he/she turns the fader on. This is very important for cueing tapes or wireless microphones-- always make sure the wireless mics are on, etc. Also, as an aside, the PFL switch is a great way to listen in on people who are wearing wireless microphones before the show, or during rehearsal, when they are offstage. Actors rarely remember that they are wearing mics and that the sound engineer can ostensibly listen to every word they are saying. This is called spying ("So what did you do with him?" "Oh, not much, we sorta sat around." "Did you kiss him?" etc., etc.). We like this switch. We are not obsessive. We do not have a one-track mind.

OUTPUT BUSES

There are several outputs on any given mixer. They have a variety of uses, all dictated by the sound engineer's needs. First and foremost are the program outs-- the main outs-- left and right. For small installations and reinforcement, these will be the main outputs from which the signal goes to the amplifiers. 

Some mixers also have separate output groups-- known as subgroups, groups, or buses. The most popular configurations are four-bus mixers and eight-bus mixers. Some consoles used for live concert reinforcement may have more. Switches on each channel can route the audio signal to any combination of these groups and to the left-right outputs. Groups can be used in a variety of ways. For multi-track recording, one can route, say, the vocals to group 1, which is routed to the first track on the tape; keyboards to group 2; guitars to group 3; and drums to group 4-- all corresponding to a track on the tape. This enables one to record a live band onto different tracks of a multitrack tape for mixdown (mixing of the individual tracks, editing, adding of effects) at a later date. In sound reinforcement, we can route wireless lavaliers to group 1, wireless handhelds to group 2, ambient mics to group 3, band to groups 4 and 5, and effects to groups 6 and 7. In this way we can bring literally groups of microphones down and control them in a less unwieldy manner than trying to control fifteen faders with ten fingers. We can also, in smaller rein-forcement systems, route a separate set of speakers from each bus. For instance, if we had a small system comprised of two stereo house speakers and two stereo subwoofer units, we could route the bass guitar to the two subwoofers (groups 3 and 4), the keyboards to groups 1, 2, 3, and 4, the vocals to groups 1 and 2, etc., etc. Ostensibly the left and right outputs could then be used to record the show live to tape. Groups are a very versatile function in a mixer.

In addition, there is the auxiliary send loop. The aux send / return loop is intended to interface outboard effects processors into the signal chain. A mixer may have from one to ten different aux sends (or efx sends, echo out, monitor out, etc., etc.). Each input channel contains individual level controls for each aux send. Ostensibly, one can create an entirely different mix using the aux sends. The intent is this: say the sound engineer wanted to apply some reverb to the lead vocalist's microphone. If he/she wired the reverb unit through the left and right outputs, he/she'd reverb the entire mix (yuck). But, if he/she brought up aux send 1 on the lead vocalist's microphone's input, and patched the reverb unit out of aux send 1 and then routed the output of the reverb unit back into another channel / an aux return, then the sound engineer has applied reverb to the lead vocalist's channel and that channel only. If he/she wanted to apply reverb to the backup vocalist's channel as well, all he/she would have to do is bring up the aux send level on that particular channel. Such was the original intent of the auxiliary send.

One characteristic of the aux send is this: is the aux send pre-fader or post-fader? If the aux send is pre-fader, it means that the signal that goes to the aux send will not depend on the fader level-- the signal is taken pre fader. Ofttimes the signal will also not be affected by the channel equalizer. If the aux send is post-fader, it means that the signal that goes to the aux send will depend on the fader level. If the channel fader is all the way down (off), no signal will be present at the aux send. This is a very important feature. If you want to apply reverb to someone, you want the aux send to be post-fader, so if his mic is down, the reverb will be quiet. If you applied the reverb through a pre-fader aux send, and the channel fader is down, the mic of course is still on and is still delivering a signal and it will go to the aux send, and some reverb will still be coming out. On the other hand, when using aux sends for stage monitors for the per-formers, you generally want the send to be pre-fader, so no matter what you set the channel fader at, the performer will have a constant monitor. Believe you me, you want them to have monitors to hear each other. Many mixers nowadays have a pre/post switch on at least two of the aux sends, enabling you to customize the mixer to its maximum.

To complement the aux send bus, there is the aux return section, which are usually separate line-level stereo inputs where the effects processors will supposedly return their signal. You can use them for whatever you want-- extra line-level inputs, a place to patch in your CD player, whatever.

MIX MATRICES

Matrix outputs are cool. Not every mixer will have them. Generally only very heavy and expensive sound reinforcement mixers will have them. Matrix outs come after the subgroups and are independent of the main left-right outputs. Mix matrices work as follows: imagine a mixer with eight subgroups, with an eight-by-eight matrix. Each subgroup will have eight controls. Each of these controls is one matrix, which, essentially, is an additional output with a customized mix. If, say, you have set up your system so that a specific group of inputs is routed to one bus (i.e. vocals wireless to group one, vocals wired to group two, ambient mics to group three, et cetera, et cetera), and you have a bunch of different speakers and locations (i.e. house center cluster, house left fill, house right fill, house under balcony delay, house surround), you will need to use the matrix outs. Each matrix output will correspond to a different speaker or set of speakers. By adjusting the level of each subgroup into one matrix, you can route the audio signal in a variety of ways. If, say, you wanted only the vocals to come out of house center cluster, then you would bring up the level on the vocal groups to the matrix containing the house center cluster. If you wanted only the band, ambient mics, and reverb to come out of the house left and right fills, you would bring up the level on the band, ambient, and reverb groups to the matrices containing the house left and right fills. If you want a nice mix of everything to go to the house under balcony speakers, you would route all the subgroups (adjusting their levels individually) to the under balc matrix. If you brought down the subgroup fader containing the vocal mics, it would correspondingly lower the level to each matrix. Mix matrices are cool. 

CONCLUSION

There are an absurd number of mixers out there on the market today. Each one has its own separate features and odd quirks, and there's no way to document each one. Also, technological advances are allowing mixers to become more sophisticated as technology improves... VCA automation, introduced in the late '70s, has made its way out of the pro studio into the home studio and into reinforcement. Computer controlled sound systems, such as that offered by Level Control Systems, are starting to become popular. We'll try to keep current, but it's difficult... 

Outboard Signal Processors

INTRODUCTION

Signal processing devices are technically defined as devices which alter that audio signal in a non-linear fashion. By that definition, a simple fader, level control, or amplifier is not a signal processor. Regardless, though, we'll discuss amplifiers and all that sort of good stuff here. 

A mixer can technically be thought of as a signal processor, since it manipulates the electrical signal: it mixes it, controls the amplitude (level), boosts or cuts frequencies (equalization), and possibly can do many other things... or at least can be linked to other equipment that can do many other things.

Such equipment may include outboard equalizers, effects processors, acoustic "enhancers," compressors, limiters, noise gates, noise reduction systems, flangers, and phase-shifters, among others.

We will examine as many signal processors as we can think of, and have time to write about.

EQUALIZERS-- A HISTORY

In the early days of the telephone industry, when long cables were used to transmit voice, a lot of signal loss (attenuation) occurred. Amplification could be used to make up for that loss, but it turned out that the loss was frequency dependent, with some frequencies suffering greater attenuation than others. Special circuitry was developed to differentially boost the frequencies that suffered the greater attenuation. Since these circuits made all frequencies equal in level, the circuits were called equalizers. Originally the term equalization referred only to circuits that boosted certain areas of the audio frequency spectrum. 

You may recognize that a circuit which acts on a certain portion of the frequency spectrum is a filter. Filters generally cut certain frequencies. If you cut most frequencies and allow certain frequencies to pass without being cut (a band pass filter), the net result is similar to having boosted those frequencies that pass through unaltered, especially when amplification (gain) is added to make up for the attenuated frequencies. Filters can thus be used, in a sense, to produce boost. Filters that act to cut frequencies were eventually combined in a single unit with equalization circuits that act to boost certain frequencies, creating the reciprocal boost/cut devices that are widely used today.

While it is historically and technically accurate to use the term equalization only when referring to boost, common usage today applies the term to boost and cut circuits. You will also see the term filter set in some contexts, such as 1/3 octave filter set where the term equalizer might or might not be equivalent (some filter sets provide cut only, and hence are really not equalizers at all). We are not too concerned with precise terminology, but we did think you should know why some people draw careful distinctions in this area.

TONE CONTROLS

The typical tone control on a hi-fi amplifier or car stereo is a form of equalizer. It generally operates in just two bands: low frequency (or bass) and high frequency (or treble). 

When you turn up the bass tone control, you increase the level of lower frequency sounds (boost them) relative to the rest of the program. This results in a richer or fuller sound or, in the extreme, in a boomy sound. Conversely, when you turn down the bass tone control, you decrease the level of these same frequencies (cut them), resulting in a thinner or tinny sound.

Let's examine the shelving type EQ created by the bass control. The graph indicates that the boost or cut gradually builds below the 1000Hz hinge point. With the control set at one extreme or the other, the circuit produces 10dB of boost or cut at 100Hz, with less and less effect above that frequency. Below 100Hz the amount of boost or cut remains constant, as shown by the response plot that has ceased to slope and is again level . This new boosted or cut level portion of the curve looks like a shelf, hence the term shelving.

The treble tone control operates like a mirror image of the bass control, boosting or cutting frequencies above the 1000Hz hinge point, but providing no more than the maximum set boost or cut above 10kHz (in the shelving region). The hinge point of such tone controls varies, and may not be the same for bass and treble circuits. The bass control might hinge at 500Hz, and the treble control at 1.5kHz, causing no change whatsoever between 500Hz and 1500Hz when either control is adjusted. Also, the point where maximum EQ occurs may vary in the real world, with bass controls reaching maximum between 50Hz and 150Hz, and treble controls between 5kHz and 12kHz.

If one were to classify this particular set of tone controls as an equalizer (which it is), it would be specified as a two-band, fixed frequency range equalizer having a shelving characteristic, and up to 10dB of boost or cut at 100Hz and 10kHz.

MULTI-BAND CONVENTIONAL EQUALIZERS

Each input channel on a typical mixer may have a two band equalizer, similar to the hi-fi tone controls described above, but it is more likely to have a somewhat more elaborate equalizer that affords separate, simultaneous control of at least three frequency bands. In the case of a three band equalizer, the middle frequency band (midrange) will always exhibit what is known as a peaking characteristic, as shown below. 

Another term for peaking is peak/dip, which reflects the fact that the peak amount of equalization can be an increase in level due to boost ( a peak in the frequency response curve) or a decrease in level due to cut (a dip in the response curve). All peaking equalizers have some center frequency at which maximum peak or dip occurs, and below or above which there is less and less effect until, at some distance from the center frequency (along the frequency axis) there is no effect. Contrast this with the shelving EQ, above (or below) whose effective frequency the amount of boost or cut remains constant.

Many mixing console channel equalizers provide two or more midband peaking equalization controls between a pair of shelving high and low frequency equalization controls, thus affording a greater degree of control of those frequencies where most of the music energy exists and where our ears are most sensitive (500Hz to 4kHz). Unfortunately, the selection of just a few EQ frequencies that are supposed to be good for everything seldom produces exactly the sound that someone wants for a very specific mixing job. For the reason, some manufacturers provide a way to alter the actual center frequencies of the peaking EQ (or the knee frequencies of a shelving EQ). Such a scheme, with a simple choice of two frequencies in each of four bands, is shown below. Observe that the low and high bands have shelving type curves (still, with switchable frequencies) while the low mid and high mid bands have peaking type EQ.

Where permitted by cost considerations and available panel space, it is generally desirable to be able to simultaneously control more frequency bands. This means that different aspects of the sound can be manipulated. Suppose an electric guitar is the input to a given mixing channel. With a two band EQ (tone controls), all one can do with the boost is the left the bass for a richer sound, which unfortunately also adds a boomy quality to certain bass notes... or lift the treble for more brightness, which, unfortunately, also emphasizes the sound of fingers sliding on the strings. Processing this same guitar with a four band EQ is an entirely different story. Now the low frequency shelving EQ control can be rolled off to cut unwanted boominess, while a lower mid peaking EQ can apply some boost at around 200Hz for a thick sound, the upper mid peaking EQ can apply some boost at 2.5kHz to increase the punch, while the high frequency shelving EQ can roll off frequencies above 8kHz to reduce extraneous noise. These selected EQ frequencies, the choice of how much boost or cut to apply, and whether the curve is peaking or shelving will depend on many factors: individual taste, the instrument or mic used, the acoustics of the environment, the sound system quality, the availability of specific EQ options, and more.

Some mixing consoles have been built with a choice of twenty or more discrete EQ frequencies, in four or five bands, on the input channel equalizers. Even this may not be adequate for pinpointing the desired sound, which is why other types of equalizers were developed, as explained in the following paragraphs.

SWEEP TYPE EQUALIZERS

For years it was recognized that if one could sweep the center or knee frequency of an equalizer, this would provide much more precise control of the sound. The technique was very costly due to the nature of the electronic circuitry in early equalizers. The coils (inductors) were either fixed in value or very difficult to alter. Newer circuits utilize relatively less costly integrated circuit operational amplifiers, plus relatively inexpensive capacitors and resistors, to emulate the function of the inductor, with the added advantage of easily changed circuit values. This has made it practical to build stable, cost effective equalizers with sweepable frequency controls. The sweep type equalizer is much like the multi-frequency conventional EQ discussed above, except that instead of switching the center of knee frequency, one can continuously adjust it. 

PARAMETRIC EQUALIZERS

In all the equalizers discussed thus far, the steepness of the EQ curve has been fixed. At a given value of boost or cut, the bandwidth of the peaking curve (the amount of the audio spectrum affected) has been set by the manufacturer and is not adjustable. Sometimes one wishes to have a very broad EQ curve, with a gentle onset and a very gradual buildup to maximum peak or cut (or to the shelving value) with respect to frequency. For example, to bring out a bit of presence in the overall mix of several vocalists, a broad peak at around 6 to 8kHz may be called for. On the other hand, a certain note or a noise can be either accented or diminished in strength with minimal effect on adjacent frequencies. This aspect of the equalizer-- the broadness or sharpness of the curve-- is described by a specification called Q. The higher the Q, the sharper the curve. 

A few equalizers a provided with switchable Q, but the majority of equalizers that provide any control of Q offer continuously variable Q between a broad and a narrow characteristics (typically Q of 0.5 through Q of 3 to 5). A very narrow notch filter, with only a few Hz bandwidth, may have a considerably higher Q. Such filters are not normally found on a mixing console channel equalizer, but are restricted to specialized uses, such as notching out harmonics of motion picture camera noise, or reducing the strong 120Hz second harmonic of 60Hz power line hum. Equalizers that provide both sweepable center frequencies and adjustable Q, as well as boost/cut controls, are known as parametric equalizers (because they allow you to adjust all the parameters of the equalization).

Usually there are several filters in a parametric EQ, and some outboard parametrics are set up for stereo operation so that adjusting one control affects two channels (which is desirable for keeping a stereo image in proper perspective). Each filter section in the parametric equalizer can either cut or boost frequencies within its band, and the range of center frequencies available from adjacent filters usually overlaps.

Some so called parametric equalizers do not have adjustable Q, and are really sweep type equalizers. Some offer parametric EQ in one or more bands (i.e. just the midrange band), but switchable or fixed frequency EQ in the other bands. These are not fully parametric. In reality, just about any conceivable combination of fixed frequency or sweep type or parametric EQ, with shelving and/or peaking curves has appeared at one time or another, so be sure to closely examine any equipment to determine how it actually works.

One of the alleged advantages of the parametric type EQ is that it enables the frequency needing help to be precisely selected, and the Q to be adjusted, so that a minimal amount of boost or cut can be applied, with correspondingly fewer ill effects on adjacent frequencies where the correction is not needed. By adjusting a filter for wide band rejection characteristics (low Q), it can perform room equalization in a similar manner to a graphic equalizer, or it can act as a variable frequency cut or boost tone control. In a narrow band reject mode (high Q), a parametric equalizer can be used for feedback control, or (as previously explained) to notch out hum frequencies without subtracting much of the adjacent program material.

Since all EQ causes phase shift, boost can reduce headroom and cut can eliminate desired portions of the program. The ability to use only the minimum amount of equalization required is thus a genuine advantage. Some people dislike parametric EQ because there are so many parameters that MUST be adjusted, and because it is difficult to make note of specific settings so they can later be duplicated in other mixing situations. If inexperienced operators will be using a mixing console, with minimal time to become familiar, it may be better to have simpler EQ. But a good parametric EQ in the hands of an experienced professional is quite a tool.

The debate over which type of EQ is best is complicated by the actual sound quality of some equalizer circuitry. For a higher quality fixed-frequency equalizer may sound much better, even if the correction cannot be as precise, than a mediocre quality parametric EQ. High quality equalizers exhibit less distortion and/or noise than lower quality units, and may give longer service without maintenance where better quality controls are employed. Some units exhibit somewhat lower phase shift, though this is more a function of the amount of boost or cut selected. As with all sound equipment (indeed, any technical equipment), the way a feature is provided is as important as the feature itself.

When applying parametric EQ to the program as a whole, you should remember that excessive boost may reduce system headroom, create clipping and make extreme power demands on amplifier and loudspeakers. In addition, a parametric equalizer may ring considerably at high Q (narrow) boost settings. Ringing is a problem caused when a filter begins to act like an oscillator. (Ringing is the tendency of a filter to resonate at its natural frequency when excited by a sine wave pulse at that frequency.) Ringing is present to some extent in all equalizers, but is usually masked by the reverberance in a sound system. High Q filters, though, can generate excessive ringing or resonance. Such ringing may be useful as an effect on a particular input source, but is generally not desirable when it affects the overall sound system. Used carefully, a parametric equalizer can be an extremely useful tool for sound reinforcement or for recording.

GRAPHIC EQUALIZERS

A graphic equalizer is a multi-frequency, band reject filter, or a bandpass/reject filter. Unlike typical three or four band input channel equalizers, a graphic equalizer can simultaneously operate on eight or more frequency bands, typically chosen to have one octave or one-third octave band centers. Most graphic equalizers use ISO standardized band center frequencies. Less common, but sometimes found are graphic equalizers with two-third octave, one-eighth octave, one-sixth octave, and, on rare occassion, one-twelfth band centers. 

The units are called graphic because most have linear slide controls. When they are set they create a visual image that resembles of the overall frequency response curve of the EQ (not the response of the sound system!). A graphic equalizer may provide attenuation only (band reject), or, more commonly, attenuation and boost (band pass/band reject).

One octave, two-thirds octave and one-half octave graphic equalizers are considered to be broadband devices, useful for general corrections or alterations in the frequency response of a system. One third, one-sixth and one-twelfth octave equalizers may be considered narrowband devices although technically they are still broadband. Truly narrowband filters have a bandwidth on the order of 4 to 10Hz rather than one-twelfth of an octave. Why are we concerned about relatively broad or narrow band filters in the equalizers? It turns out that things like AC hum or motor generated noise occur in very narrow bands, and many room resonances are very narrow. Correcting them with broader filters means that some non-problem frequencies will be affected, which have unwated audible side effects.

There are a number of reasons why few graphic equalizers are one-sixth or one-twelfth octave devices, however. For one thing, what can be covered in 27 to 31 one-third octave bands requires about 60 one-sixth octave bands, or over 100 one-twelfth octave bands. That becomes a very expensive device, a very large device, and one which is very, very time consuming to use when tuning a room. Greater phase shift occurs with narrow filters, which can create unpleasant swishing sounds as program frequencies sweep through the equalized band. Technology and the marketplace have, so far, determined that one octave graphic equalizers are useful for general tonal corrections, and one-third octave graphic equalizers are sufficient for most room tuning and feedback avoidance.

Graphic equalization reduces the effect of resonant peaks and dips in loudspeaker response and, to a lesser degree, in the acoustic environment, reducing the tendency for acoustic feedback to occur. As the overall gain of the sound system is turned up, feedback will first occur at that frequency (or frequencies) where the system has a peak. It typically begins as a slight ringing, and then becomes a loud howl. By using a graphic equalizer to attenuate the first peak, the overall system gain can be increased until the next (formerly lower) peak begins to feed back. That peak is then attenuated using another graphic EQ band, and the system gain can be further increased. When the peaks have all been leveld to the extent possible with the EQ, the overall system gain may increase from 6 dB to 10 dB above the initial gain before feedback commences.

Another use of graphic equalization is to contour the frequency response of the mixing console's output to obtain the most pleasing sound quality or improved intelligibility. Flat response is seldom desired, and almost never realized in sound reinforcement applications. Audio may be reasonably flat over the middle of the audio spectrum, but the bottom end is sometimes boosted for effect or rolled off for power handling and reverberant considerations, while the top end is usually rolled off somewhat due to typical listening preferences. Sometimes the middle portion of the spectrum (1 kHz to 5 kHz) must be boosted to improve the recognition of vocal consonants and sibilants, particularly when these sounds are masked by other sounds in nearby frequency bands.

The graphic equalizer is a very useful tool, but it cannot substitute for good acoustics or for well designed amplifier/loudspeaker systems. Excessive boost, especially at lower frequencies, drains much of the available amplifier power, overstresses the drivers in the loudspeaker system, and reduces overall system headroom. Excessive cut takes out noticeable portions of the program along with a desired response peak or noise component.

The signal driving each loudspeaker (each main cluster or each monitor mix) usually requires its own channel of graphic equalization, which is installed just after the mixing console output, before any electronic crossover or the power amplifier. Stage monitor feeds, for example, may require very different equalization than house feeds. In recording and broadcast applications, the graphic equalization applied to the recording is usually for tonal considerations, and to avoid exceeding the frequency response limits of the medium. The studio monitors or audience foldback system might require graphic equalization to suit very different needs.

REVERBERATION AND OTHER EFFECTS

Reveberation-- the phenomenon that occurs when sound is reflected and reinforced-- occurs naturally in most enclosed spaces. Each venue will have different reverberation characteristics-- size, shape, obstacle locations, etc., etc. will all play a part in how the sound acts once it is let loose in the house. Reverberation is defined as consisting of multiple, blended sound images caused by reflections from walls, floors, ceilings, and other obstacles that do not otherwise absorb the sound. Reverb processors are electronic devices which emulate this effect. Designed for the recording industry to simulate different types of sound locations, they have also found a home in the reinforcement industry to give more "fullness" to program material. Reverb units increase "depth" of the sound (most noticeably on the last syllable); reverberation is often confused with delay or echo, especially since most modern signal processors include both effects. Delay refers to one or more distinct sound images-- echoes. In fact, true reverberation normally begins with a few relatively closely spaced echoes known as early reflections. These are caused by the initial bounce back of sound from nearby surface. As the sound continues to bounce around, the increasing number of reflections blend, creating the more homogeneous sound field we call reverberation. 

In days of old, reverb unit designs included wiring a small loudspeaker to one end of a garden hose and micing the output. The spring type reverb was one of the more common units. A transducer was attached to a metal coil-- a spring. The speaker twisted the spring, and the sound then traveled up and down the spring. Another transducer attached to the other end of the spring converted these mechanical reflections back into an electrical signal. There were many mechanical problems that resulted in bad quality when used improperly. A similar design using a large metal plate worked slightly better, but the inordinate size and weight of the unit restricted much of its use to studio recording. Delay units consisted of a tape loop which recorded the signal at a given point on the tape; subsequent playback heads set at different intervals after the record head played back the recorded signal and thus delayed the signal. The main problem with these units was tape degradation-- thus, signal degradation.

In the days of digital equipment, producing such effects has become more cost-effective with better quality. The input section of the unit will sample the analog signal and convert it to a binary digital form, using often the same method as a compact-disc player or DAT deck. Digital effects units use complex algorithms to manipulate the digital signal; at the end of it all, the signal is re-converted into an analog audio signal. There's a lot of mathematical manipulation going on inside the digital reverb. Most digital effects units also include some memory function for storing user-preset effects.

Modern delay units work in much the same way. An analog-to-digital converter converts the incoming analog signal into binary data, which is then fed into RAM registers. A clock (crystal oscillator) generates sync pulses that cause the memory registers to expel the data, which is then converted back into an analog audio signal. The amount of memory and the clock speed with determine the audio quality and the maximum amount of delay time. The use of delay units is covered in the Sound Reinforcement section.

COMPRESSORS AND LIMITERS

Compressors and limiters are signal processors that reduce the dynamic range of the signal. The compressor and limiter are essentially the same thing; the difference in nomenclature has to do with the actual use of the unit. The compressor/limiter is designed to prevent signals from exceeding a given (adjustable) threshold level. The ratio of the change in output level to the change in input level (in dB) is known as the compression ratio. Most limiters will have a compression ratio of from 8:1 to 20:1. If a unit is set to 8:1 compression, then an increase in input level of 8 dB (assuming the input is set above the threshold level) will result in a 1 dB increase in the output level. Some units offer infinite compression, where no amount of increase in input level above the threshold will cause an increase in output level. 

Limiters are generally used to process only program peaks, which is why they are also known as peak limiters. In sound reinforcement, they can be used to protect loudspeakers from mechanical destruction in the event of a dropped microphone by limiting the peak level that will be fed to the amps and speakers.

If the threshold level is reduced so that most or all of the program is subject to compression, then the device functions as a compressor. Compressors generally use lower compression ratios than limiters-- typically 1.5:1 to 4:1. Compression has a number of uses. In tape recording, broadcast, or reinforcement, compression is sometimes used to squeeze the dynamic range of a program to suit the storage or reproduction medium. It is used similarly to a limiter in sound reinforcement situations to increase gain but also protect against peaks and other destructive transient sounds.

Compressors will usually have some of the following controls: attack time control, which regulates the speed at which the gain is reduced in response to an increase in input signal level; release time, which regulates the speed at which the gain is restored to the original value after the input stimulus is removed; a side-chain circuit allows compressors to be used in response to certain frequency signals--if you want more compression in response to high frequency signals, you insert an equalizer in the chain with the high frequencies boosted. This setup is often used for de-essing, where vocal sibilance is removed by differential compression (that is, since the high frequencies are boosted and thus have a higher dynamic range, they will be compressed before low frequencies are affected). If low frequency equalizer cut is used, the compressor allows drum sounds to get through more or less unaltered, yet may clamp down on a relatively less powerful (but more threatening to tweeters) high-frequency synthesizer note.

The following is from the Yamaha Sound Reinforcement Handbook, page 273:

With a given input signal, adjust the input level control (if any) so the input is well above the noise floor, but does not clip the input stage. Then set the threshold to whatever point, and set the compression to whatever ratio that may be appropriate for the situation. For speaker protection, for example, the threshold should be set to a point that prevents the power amplifiers from delivering whatever power level is established as the mechanical limit for the speakers. Suppose a loudspeaker is rated at 100 watts continuous and 200 watts peak, and the power amplifer is rated at 200 watts output to that speaker's rated load impedance, given a +4dBu input. Let's also suppose that the power amp's input attenuator is turned down 10 dB. (For simplicity, we'll assume that the compressor's input and output level controls are adjusted for unity gain through the device when there is no compression). In this case, a +14dBu signal applied to the amp causes it deliver 200 watts to the speakers. The threshold and compression ratio of the compressor should therefore be set to avoid exceed +14dBu. If you want to preserve as much as possible of the natural program dynamics, set the threshold to +10dBu. Our criteria require that any input signal, no matter how loud, should not cause the output to increase by more than 4dB beyond that value. We assume that due to the capabilities of the equipment feeding the compressor, no input signal will exceed +26dBu. We subtract +10 from +26 and see that a 16dB dynamic range must be compressed to 4dB, and simple math shows us that a 4:1 compression ratio should do the job. 

NOISE GATES AND EXPANDERS

A noise gate is a signal processor that turns off or significantly attenuates the audio signal passing through it when the signal level falls below a user-adjustable threshold. The idea is that the desired program will pass through unaltered, but low-level hiss and noise (or leakage from other sound sources) will not be heard when the primary program is not present (presumably when the level is below the set threshold). 

Those noise gates that literally shut off the signal flow when the program is below the threshold level will tend to have an audible effect as they cut in and out. The sudden change in background noise level may be disturbing. This is why some noise gates are designed to merely reduce the signal level by a finite amount (to lower the gain) when the level falls below the threshold. The effect is to reduce noise, but not to have a drastic, sudden change. To further avoid the audible modulation of background noise, these units may have automatic or adjustable time constants where after the level drops below the threshold, it takes so many milliseconds for the gain to be reduced.

The circuit that reduces the gain is an expander, although it is not known as such in this case. What is happening is that the noise floor of the program is being reduced, and hence the dynamic range of the program is being expanded.

When the expansion circuit works only below a set threshold, we call the device a noise gate. There are also signal processors that expand the entire program. In this case, the threshold is set to be any convenient zero point, typically at the nominal program level. Any signals falling below that threshold are expanded downward in level so they become even quieter than they already are, and signals above the threshold are expanded upward in level. The net result is a program with greater dynamic range. In this case, the device is called an expander.

Noise gates are useful for automatically muting temporarily unused mics in a recording or sound reinforcement system. The number of open mics reduces the available gain before feedback in a sound reinforcement system, and generally adds to the background noise in a recording. Particularly in complex, multichannel setups, the use of a noise gate can improve the sound without increasing the workload for the mixing engineer. In order to be effective, with minimum audible side effects, each subgroup, or perhaps nearly each input to the mixing console, should be processed by its own noise gate.

Expanders are a component in most tape noise reduction systems. They do the decoding of the encoded (compressed) audio tape, simultaneously restoring the original dynamic range of the program and pushing down any added tape hiss or noise below the inherent program noise floor. Expansion can also restore (or create) the missing punch of a complete program mix or an individual signal in that mix.

OTHER EFFECTS

Other effects processors include phasers, flangers, and exciters. Phasers and flangers essentially produce the same effect through very different means. Flangers originated from the use of two tape recorders playing the same program. By mixing the outputs of the two and alternately slowing down one machine, then the other, different phase cancellations occurred. The slowing down was achieved by using hand pressure against the flanges of the tape supply reels, hence the name. Modern flangers use electronics to simulate this effect. If a given signal is delayed, then mixed back with the original signal, the result is cancellation at a frequency whose period is twice the delay time. This cancellation also occurs at odd harmonics of the signal frequency. 

Phasers, or phase shifters are devices that contain one or more deep, high Q filters. The input signal is split, with some of it going to the filter circuit, and some bypassing the filter. A lot of phase shift is created at frequencies on either side of the filter notch. By sweeping that notch up and down the frequency spectrum, and mixing the resulting signal back with the direct signal, a series of ever-changing phase cancellations results.

Phasers and flangers are usually used in-line on guitars, bass guitars, and keyboards. Flanging effects, since they are dependent on delay, can also sometimes be found on delay units.

In 1975, Aphex introduced the Aural Exciter. The exciter added more punch to the overall program material without appreciably changing overall system gain. The input signal was split-- one side directly to the output, and one side to a system of filters and equalization circuitry. The resulting output was mixed at the output and created a punchier, more intelligible sound. The unit became popular in broadcasting, where greater penetration could be obtained without overmodulating the signal; in sound reinforcement, where feedback and headroom were not sacrificed; and in dance-club sound systems, where listeners benefitted from an apparent change in level without distortion.

Aphex still manufactures the Exciter, and other companies, such as BBE, Behringer, Furman, and dbx all manufacture some sort of program exciter.

Amplifiers

Uh-oh. Looks like someone hasn't written the "Amplifiers" section. Yet. 

I'm sorry.

• Back to the top