Equalisation

Discotheque equalisation ('EQ') falls into two discrete categories. Microphone EQ can be used to project your voice with clarity or imbue it with character - such as presence or breathiness. EQ can also make loud vocals a reality without risk of feedback whereas for general music, reproduction is usually of a totally different nature. For a purist, creative EQ isn't an inherent requirement here, because if a record is worth playing, it will 'come across' of its own accord; Black Sabbath live and music that is often 'out of balance' thrive without EQ provided your audience are attuned. In this instance, music EQ is as philistine as touching upan impressionist painting because 'it looks untidy!'

An alternative and holistic viewpoint is that it's valid for the DJ to interpret the records he plays by EQ'ing them and even by adding sound effects. But this approach demands a sensitive appraisal of the nature of the record and your audience; a souped-up version of 'Foxy Lady' would not endear you to those who hold Hendrix in reverence. From a different angle, EQ is required to repair the unwholesome sound of many records. Limitations of equipment often conspire to ensure that electro-music is rarely heard with fidelity - either recorded or live, whilst, ironically, the public's awareness of sound quality goes from strength to strength. This problem is most acute in the realms of Heavy Metal discs, which frequently emulate the ear-ripping characteristics of certain Rock PA systems, rather than the sweet (if cacophonous) sound of the unadulterated music. Conversely, clarity on record frequently goes hand-in-hand with a good live sound. Even though Reggae and Funk are often excellently recorded, the startling clarity and vitality of recordings by Aswad, Level 42 and Earth, Wind and Fire violently underline the paucity of microphone technique, mixing and cutting skills, and pressing quality in a lot of vinyl.

Equalisation is also a means of correcting deficiencies in the system, particularly those occurring in the loudspeakers, and it can compensate for bad acoustics to an extent. For the time being, however, we'll concentrate on creative EQ - which isn't really equalisation at all, of course!

The Tools

The ubiquitous Baxandall tone control (first described by Peter Baxandall in 1952), owes its popularity to its ability to provide fair compensation for the deficiencies at the extremes of the audio spectrum commonly encountered in low-cost audio equipment. But as a creative equaliser, it's not especially useful in the classic 'bass and treble' configuration, your control over the nature of the sound being limited by the most arbitrary dichotomy. And the regular circuit is inherently limited to providing shallow response curves which peak only at the extremes of the audio spectrum. Nonetheless, the standard configuration can be usefully tailored, notably to home in on the two key regions in vocal applications. The fundamentals of male vocals lie between 70Hz and 150Hz, whilst the upper harmonics for both male and female lie in the 5kHz to 12kHz region. The component values in Figure 1 are attuned to providing maximum effect about these frequency bands as opposed to circa 30Hz and 18kHz, as in Baxandall circuits a la domestic 'Hi-Fi'. Apart from making the standard Baxandall tone control circuit better suited to adding character to your vocals, it's also possible to enhance its ability to add 'sparkle' and 'Tsst' to the music viz, boosting the high treble (10-18kHz) region. The usual problem here is that reasonable accentuation of these frequencies incurs a degree of overkill in the 2-7kHz region (where the ear is extremely sensitive, the pain threshold being a mere 108-114dB hereabouts) which creates a 'bitey', harsh, fatiguing sound, quite apart from being cruel to bats and dogs! In Figure 1, this problem has been overcome by adding a switched shelving control. When the control is in position 'A', the treble boost grows with increasing frequency starting around 1kHz, in the normal fashion, only flattening out (or shelving) at the uppermost end of the spectrum. In positions B and C however, the boost curve begins at higher frequencies. Thus painful 3kHz won't predominate as a consequence of gaining an audible effect at 16kHz for instance.

One major shortcoming of the original Baxandall control is its inability to provide any useful control over the midrange frequencies. Whilst a midrange section can readily be added to yield useful results, it suffers from interaction. This is an inevitable consequence of placing three passive networks in parallel, with close seated frequency ranges. Indeed, it's quite common to find interaction in the two-band version - the high treble frequently evaporates when maximum bass boost is applied. Apart from this inexpedience, a three band equaliser is still one elusive step from providing useful control over the four essential elements of music. In other words, equalisation which provides control over the bass (the soft, physical sounds), the low midrange (the warm, muddy sounds), the high midrange (the hard, delicate sounds) and the treble (the cold and transparent sounds) is the pre-requisite for creativity in sound. Like the north and south poles of a magnetised material, all music can be split into a series of alternately soft/warm/hard/cold sounds.

Figure 2 shows a practical four-band equaliser. In this circuit, the problem of interaction has been elegantly sidestepped by using a pair of two-band networks in series.

If you use an active crossover, an unusual brand of EQ can be achieved by tweaking the attenuators in each band. This yields a flat, broadband adjustment that is quite unique, although not necessarily useful. However, if you only sport bass and treble controls on your desk, the midrange attenuator(s) on the crossover can be useful for restraining the brash midrange sound on certain records, or, alternatively, the bass and treble attenuators can be adjusted in conjunction with the desk's tone controls to yield a balance between overall bass boost and bass boost centered in the high or low regions. For instance, by cutting the bass on the Baxandall circuit, and boosting it on the crossover attenuator, the high bass (circa 200Hz) can be boosted at the expense of the low bass - a reversal of the normal state of affairs! The key disadvantage of using the crossover in your equalisation schemes is the impossibility of differentiating between microphone and music EQ; if the crossover is over-enthusiastically adjusted in pursuit of the latter, protestations a la feedback invariably occur from the microphone...

The success of four-band EQ depends greatly upon a judicious choice of centre frequencies - the point about which the equalisation in each band has most effect. The SWEEP EQUALISER overcomes any compromises in this respect, as the centre frequencies are continuously variable over a range of several octaves. This gives great flexibility, but its versatility can overawe, and considerable experience is necessary before one can instinctively 'feel' a 4-band sweep equaliser and adjust it successfully whilst concentrating on your audience. Much the same goes for the PARAMETRIC EQUALISER in which the bandwidth (or 'Q') of the equaliser's effect can be either sharpened up or spread out-together with a continuously variable centre frequency. It's worth noting that some sweep equalisers on the market are sold as 'parametric' - naughty and deceptive! The true parametric is particularly valuable for microphone EQ; its selectivity allows you to accentuate the interesting harmonics in your vocals. The graphic or multiband equaliser is an extension of the four-band equaliser, but it's not necessarily advantageous, because after spending a lot of money, you're stuck with a finite number of centre frequencies. An attempt to save money can lead you to problems of interaction once again. In general, for creative applications, choosing a top-notch 5 to 10 band unit (such as the Klark Teknik DN22) is preferable to skimping (willy-nilly) for the sake of more bands.

Although their creative applications are legion, these instruments can yield rather vulgar sounds that forbid their trial and error adjustment mid-performance, though of the three, the sweep equaliser is the easiest to 'throw-in' surreptitiously. Instead, useful settings for vocals and/or certain records are best noted before the performance; the equaliser is then switched deftly in and out between records or announcements. Multiband and parametric equalisers will be of greater practical use to the DJ when enterprise ties them to a microprocessor to provide a library of instantly accessible preset equalisation settings.

Metering

Meters can help you to equalise (literally this time!) the level of the forthcoming disc or tape, prior to crossfading. Although it's often not possible to pre-adjust the level of disc sources, a meter will at least forewarn you of impending changes in the modulation. On the line inputs (e.g. tape sources) a gain control is usually present, and the meter can be used before the gig to roughly match tape and disc levels.

Metering can also provide you with an objective reference as regards sound levels; a clean sound system can 'feel' deceptively quiet, and whilst it's tempting to raise the level (good music is never loud enough!), restraint is often called for, to appease either your client or a nasty sound level meter/cutout. Metering of this nature is best placed across the output lines, immediately prior to the power amplifiers; but for the metering to be meaningful, it's necessary to abstain from adjusting the amplifier's gain controls - which lie beyond the meters - during the performance. These meters can also serve as an invaluable guide to the vulnerability of your speakers when you have to push the sound system to its limits, provided they are accurate, and the input sensitivity and output power of each amplifier is sensibly aligned.

It's quite common to place the output metering in the desk - or across the line level outputs - if the console features internal power amplifiers. However, unless it's guaranteed to always drive amplifiers with identical sensitivities, which match the 0dBu/0VU reading on the meters, this arrangement is fraught with error. However, these meters can be of use in large rigs as a guide to the output level of the desk, as opposed to other line levels, as the signal passes through equalisers and crossovers en route to the power amplifiers. They're also a reassurance when nothing works five minutes before the performance that an output signal is present, and that the desk isn't being grossly overdriven, though no competently designed desk should approach clipping under normal conditions. Metering of this kind is also useful across the inputs, and whilst it needn't be precise, it must indicate over a wide dynamic range. Given this, it's a godsend and a source of sobriety when leads fail, or connections and switches are attacked in a wild and frantic effort to find the missing signal! As you won't usually want to know if all inputs and outputs on the desk are working simultaneously, a switched metering arrangement is adequate (Figure 3). Another vexing question, 'Is this *?*!*??* power amplifier working?' cannot reasonably be answered with switched metering, and short of going to the expense of providing a full range meter for each amplifier, a useful indication of an amplifier's utility can be provided with two LEDs, indicating outputs of approximately 10% and 100% of full power respectively (Figure 4).

The VU and its shortcomings

Affordable meters come in two varieties. The VU meter, dating from 1939, is cheap but has numerous shortcomings. Its VU (or 'Volume unit') scale (which corresponds to dBm, albeit with a zero error of +4dBm) only applies if the meter is connected across a 600 ohm line, which today is largely obsolete in Europe. Because the VU standard specifies a meter impedance of 3k9, which causes significant loading and even distortion in a 600 ohm system, a series resistor has to be added, conventionally to make the impedance up to 7k5. With this attenuation, 0VU becomes equal to +4dBm (0dBm is 775mV on a 600 ohm line) with a sinusoidal or musical waveform - but only if you're measuring across a 600 ohm line!

This pedantry is completely out of step with modern UK equipment practice, where low impedance (<100 ohm lines) are standard. This, coupled with the zero error makes the VU an unhelpful standard, to say the least. The problem is compounded by the widespread use of nondescript oriental meters, which are calibrated in 'VUs', but conform to no recognisable standard. These scruples can be transcended by driving the meter actively (Figure 5) rather than just dunking it across the output, and by regarding it as a guide rather than as the ultimate arbiter. This attitude is important, because the VU meter movement is sluggish; it takes time to respond. Thus it can only indicate the average (or long term) levels in a signal. When instruments giving a substantially continuous tone e.g. woodwind, brass and strings predominate in the signal, the meter will read fairly accurately. But signals bearing percussive sounds, viz: vocals, drums and keyboard instruments will under-read by some 8 to 14dB; hardly a small error!

As Rock, Funk and Reggae abound with - indeed, are built out of percussive sounds, VU meters are largely useless for assessing the peak level of the signals in a discotheque system and the onset of overload in power amplifiers. Now because it's often dipped waveforms, rather than a small excess of sinusoidal power, that blows speakers, a lone VU meter isn't a reliable guide to driving your system to its limits without damage. However, it's still quite valid as a guide to perceived loudness. Thus the VU is acceptable as a means of equalising levels, and indicating the presence of a signal, but there its usefulness ends.

A further problem with the VU meter is its linear scale; this is fine for judging modulation levels percentage-wise but as a 'dB' meter, around two-thirds of the scale length is taken up with the 6dB's centred around 0VU. Hence the meter needle either registers vaguely at the bottom and or flickers confusingly all over the top end of the scale.

Peak programme metering

Whilst transient overloads are unforgiveable if they occur on a master tape costing £10,000 to produce - or if they blow up a broadcasting transmitter, they're apparently harmless in a live performance, in that they are quickly forgotten, to an extent, this viewpoint is reasonable.

However, the crux of the problem is the knife-edge between dirty, clipped sound and clarity in conventional transistor amplifiers. Occasional clipping - say half a dozen mildly clipped cycles every few minutes is harmless, but minute changes in the voltage of the mains supply, or a tweak of the gain or EQ controls can send the system into prolonged clipping; the difference between live speakers and dead ones can be traced back to an excess of a mere 2 or 3dB's over several hours... Apart from the immediate upsurge in speaker voice coil temperatures and the precipitous cone accelerations called for by the substantially square shape of clipped waveforms, there's ample evidence that it's clipped percussive sounds rather than loud music per se that is primarily responsible for hearing loss. In other words, accurate metering which will enable you to steer clear of long term clipping will not only save you the cost of damaged speakers but will also prevent you alienating and offending your clientele, through the nausea and ringing ears that come hand-in-hand with distorted sound.

So, the second variety of meter is the PPM (or peak programme meter). Originally spawned by the BBC, the PPM features an agile meter movement with a log scale, and a driver circuit. It remains the Rolls Royce of audio metering on account of its rigorous specification and thoughtful design. But true (i.e. BBC spec.) PPMs are expensive enough to be restricted to broadcasting and recording studios. With the benefits of LEDs, however, we can synthesise the most useful aspects of the BBC PPM - and even improve on them. The prime requirement for indicating peak levels is not only for the meter to respond quickly, but also to present the message 'Oi, you're overloading me!' for a reasonable length of time. To this end, the BBC PPM provides full deflection in 2.5ms (although not exceptionally fast, the ear is unlikely to hear distortion occurring over such a short duration). The needle then takes one second to approach zero level again. Any well designed LED PPM should iterate this decay characteristic (approx 8dB/second); unfortunately, the LED meters built into many consoles fail in this respect, having a 'liquid' response that makes the peak readings hard to see, let alone attention-grabbing. In the latter respect, the LED meter has the advantage that green, yellow and red can be used to shout 'OK', 'You are approaching the limits' and 'Ouch!' In this case, you only have to glance at the meter to know roughly what's going on. The BBC PPM has a 1-7 numerical scale, but a discotheque version using LEDs is more useful if it's calibrated in dBU or watts (Figure 6). Finally, by adjusting the time constants of the attack and decay characteristics an LED peak meter can also yield average characteristics - in other words, an LED 'VU' meter! A single LED meter with a 'peak/average' switch can then meet all the metering requirements in a console. The scale needn't be regular, it's useful to provide 1 or 2dB steps around the overload point, with 3 to 6dB steps at the lower end of the spectrum.