Recording Techniques (Part 12)
Part 12: Equalisation, there are equalisers, and there are equalisers. David Mellor wonders whether some are more equal than others.
Language is such a strange thing that I sometimes wonder how we manage to communicate at all. For instance, according to a learned volume on my bookshelf the word 'nice' didn't used to mean 'nice' at all. It used to mean 'precise'. Similarly, 'quite' didn't mean 'almost', it meant 'absolutely'. 'Equalisation' is a word that my book doesn't mention, but when used in the context of making recordings one might wonder what on earth are we meant to be making equal, and equal to what? The answer to this question (and to many of life's other great mysteries) lies in historical reasons.
Before mixing consoles were invented — indeed before any form of recording other than direct to 78rpm shellac disc was invented — telephone engineers were working on a problem concerning the sound quality of their lines. As you might expect, if you send a telephone signal down a very long cable, some of the sound gets lost. (The first transatlantic cable, by the way, was laid well before the beginning of this century, and therefore before recording of any kind). Different frequencies are lost in different amounts according to the characteristics of the cable, which means that the signal coming out of the far end of the cable is not the same as the signal that went in, either in level or frequency balance. In other words it is unequal to it. So, what the telephone engineers came up with was a process called equalisation, using circuitry which changes the balance of frequencies of the signal. Today we use similar processes to change the frequency balance of our musical signals, and so we use the same term — equalisation, or EQ for short.
If you look at the circuitry of any equaliser, you will see the same types of components. The capacitor is a component which passes higher frequencies easily, whilst making it difficult for low frequencies to get through. The inductor works in the reverse manner, letting only low frequencies pass through. (Clever circuitry can allow capacitors, which are easier to obtain in precise values, to simulate the effects of inductors). Variable and fixed value resistors control how much of the signal is allowed to pass along any signal path, for example through a capacitor to earth where it will be taken to that great last resting place of audio signals everywhere.
As well as these three passive components, small amplifiers are used to accentuate the effects of the frequency-sensitive components and to make up signal losses. All equalisers consist of these four types of component only, but designers have come up with different ways of making them useful to the operator. The three main types of equaliser you will come across are mixing console EQ, graphic EQ and parametric EQ. There are some other weird and wonderful devices which operate in ways of their own, but still use the same four basic types of component internally.
I mentioned console EQ earlier in this series, but now is the time to go into a bit more detail. This is an area where designers and manufacturers need to look seriously at their products and ask themselves whether they are doing as well as they could. Many are most definitely not. To get subjective for a moment there are two types of EQ, which may look exactly the same from the operator's point of view but sound totally different. I would like to use an analogy with woodworking to make my point clear: the chisel is a very useful tool for removing precise quantities of wood for making simple slots or complex joints, just as EQ is a useful tool for shaping sound. A sharp chisel is a pleasure to use — just position it correctly, push it with exactly the right amount of force, and it will fashion the wood almost as though you were doing the work by mind power alone. A blunt chisel is completely the opposite. It will do the work, but grudgingly and with a poor end result.
A good EQ — and I don't necessarily mean a complicated EQ — will shape the sound just the way you want it, precisely and with a positive effect. A poor or mediocre EQ will have an effect, but it's unlikely to achieve what you want. Advertisements may claim "smooth, musical equalisation" or similar, but don't believe anything other than your own ears. I must add at this point that there still isn't any consensus on what actually makes a good EQ, apart from the fact that you know it when you hear it. Most engineers who have any knowledge of what goes on inside an equaliser will have their opinions, and I'll be passing on mine, but there is still plenty of scope for some solid research to be done to find out exactly what makes a good equaliser tick.
The most basic form of EQ is the filter. Figure 1 shows the ways in which filters can affect a flat signal. There are two basic types of filter: high pass and low pass. People tend to find these terms confusing, because a high pass filter cuts out low frequencies and a low pass filter cuts out high frequency, so the range of frequencies where a filter has its effect is opposite the range described in its name. There is a third type, the bandpass filter which is a combination of high and low pass in series. Both principal types of filter have two main parameters: cut-off frequency and slope. The cut-off frequency of a filter is defined as the frequency at which the level is 3 decibels below the pass band level (ie the level of that part of the original signal that passes through unaffected). The slope is the maximum rate at which the level drops off above or below the cut-off frequency.
Slopes, measured in dB/octave, nearly always come in multiples of six: 6dB/octave, 12dB/octave, 18dB/octave and 24dB/octave. This is because filters are very easy to make with these slopes, but very difficult with in-between values (or with slopes greater than 24dB/octave). A 6dB/octave filter can be made with just two passive components, and has a very smooth roll off. At the other extreme, a 24dB/octave filter has a strong effect and really does things to the sound. Given the choice, I would specify variable frequency 24dB/octave high and low pass filters for the EQ section of my console, because they would give me the power I need, and I like to think that I am able to use that power sensibly. In this case, the variable frequency would probably have to be switched rather than continuously variable. You can't go wrong with a 6dB/octave filter, or even 12dB/octave, but you can't really do very much either.
The high and low frequency EQ controls on a console, which are separate from the filters, are very important too. The two types you are likely to find are shown in Figure 2. The bell curve cuts or boosts a range of frequencies around its centre point, the shelf cuts or boosts all frequencies above or below the stated frequency. (In practice, the shelf will roll off so that frequencies at the very edges of the audio band and beyond are not boosted unnecessarily). Both of these curves have their uses, and upmarket consoles will provide a switch to give you the choice.
When it comes to actually using the high and low frequency EQ controls, you will find that there can be an immense difference between the EQ on different consoles, even when the controls look and are labelled the same. I don't mean that the more complex facilities of upmarket consoles necessarily provide any extra benefit in themselves — what I am talking about is the difference between the sharp chisel and the blunt chisel. It's possible to have a sophisticated EQ that isn't particularly impressive, and a simple EQ that sounds great.
The mid frequency EQ sections on most consoles sound a lot more similar to each other than do the high and low frequency controls, so I'll refer you to the section that follows later on parametric EQ. The mid frequency controls on most consoles are simple parametrics without the Q control.
Graphics are a great invention, there's no doubt about that. But there is also no doubt that a graphic equaliser doesn't provide the answer to all EQ needs. The reason graphics came into being is that it's fairly simple to design a variable-gain bandpass filter section — it's not a great step to the idea that it might be nice to have a whole set of them to adjust different frequency bands covering the audio range. Add linear controls in place of rotary knobs and the graphic is born.
A typical graphic equaliser will have around 30 bands to cover the full audio spectrum from 20Hz to 20kHz. Some graphics have fewer, but they are not nearly as useful. Graphic equalisers may be described as being '1/3 octave' types. This means that each band covers a frequency range which is simply a third of an octave, and you can cut or boost all frequencies in that band by up to 12 or maybe 18dB. In reality, the response of each band extends outside its stated range, simply because it's impossible to create a filter with an infinite slope, and even if you could it probably wouldn't sound too brilliant.
This is where the 'graphic' concept falls down. It's easy to believe that, because you have sliders running up and down showing the level of each band in the vertical direction and the frequency of the bands horizontally, the slider knobs trace out a graph of the frequency response. They don't. They do give a rough guide but, for example, if two adjacent sliders are set to +6dB, then the degree of boost at frequencies in between would be rather more than this. Figure 3 shows what happens. Of course, this isn't a problem as long as you are aware of it.
Graphic equalisers are great for picking out problem frequencies or ranges of frequencies and dealing with them. As I shall explain later, they are also good for shaping the response over the whole frequency range. As far as changing the sound of individual instruments is concerned however, they are blunt chisels compared to a good console or parametric EQ — but you wouldn't be using a chisel when you really need sandpaper, would you?
Parametric EQ is so called because nobody could think of a better name for it. At least that's my theory. I suppose that at the time it was invented, it offered control over more of the parameters of the signal than any other EQ, so 'parametric EQ' was as good a name as any. Parametric equalisers have three controls for each frequency band, and a typical unit will have perhaps five bands altogether. The three controls are frequency, gain and Q. The Q control is sometimes called 'bandwidth' — the knob does exactly the same thing. Figure 4 shows the effects of the controls. The frequency control sets the centre frequency around which the boost (or cut) will take place. Gain sets the amount of cut or boost at the centre frequency. Q sets the sharpness of the bell shaped curve.
Doubtless there will be engineers who will be content just to use the Q control without understanding how the values are worked out, but since the Sound On Sound reader is always more demanding of explanations, here is the story of Q. In the early days of radio, circuit designers were trying to produce a device which would react strongly to the signal from one radio station and reject signals from others. Circuits were built using capacitors and inductors (exactly the same components found in equalisers), and some were found to be better than others. A radio station transmits its signal on a very narrow band of frequencies and the more precise the response of the detector, the more it will react to those frequencies. By reducing the resistance of the coil, engineers found that the sharpness of the response, or the sharpness of resonance as they would say, improved. To compare different circuits, it was necessary to measure this sharpness, referred to as Q (after Quality of resonance, some people say).
In this context, Q was defined as the arithmetical difference between the two frequencies where the response had dropped by 3dB, divided by the centre frequency of the resonance. Q can be measured in other ways, but they all amount to the same thing. For audio purposes, Q is the bandwidth between the -3dB points divided by the centre frequency. As it is just a simple ratio, it has no units. Bandwidth, as you can see from the last sentence, refers to the range of frequencies between the two points where the level is 3dB down from the level at the centre frequency. On a graphic equaliser, it would be measured in octaves or fractions of an octave.
Let's be clear about one thing: you don't have to use EQ. If whatever you are recording sounds good with no equalisation, then you don't have to use it just because you have EQ facilities in your rack or on your mixer. But there may be problems with the sound that EQ can correct, or alternatively EQ can be used to enhance a sound and make it better than real life. I'll examine the corrective uses of EQ first.
If you're making a live recording there will be no shortage of unwanted noise. Some of it you'll just have to live with. (For example, I recently had to put up with helicopters flying around the church where I was recording a concert of 13th century music.) Other noises can be reduced by EQ or filtering. Excess bass is often a problem in live work, simply because bass frequencies find it easier to propagate in large auditoria. If microphones are suspended on wires or ropes, then they may be subject to convection currents in the air which will generate low frequency noise (sometimes so low that you can't hear it, but you see the meters dancing up and down and wonder what's going on).
Another source of low frequency noise is the stand-mounted microphone on a wooden stage. Vibrations from stomping feet will work their way up the stand and into the mic no matter how well it is suspended. The solution is to flick the mixing console's high pass filter into circuit (remembering that 'high pass' means 'low cut'). If you have a choice of frequencies, 80 or 90Hz is good and unnoticeable, and 150Hz combats stronger low frequency noise but does remove some bass. I am always inclined to use the high pass filter on any instrument that doesn't produce any bass, just leaving it out for male vocalists with deep voices, bass guitars, double basses etc.
Coming back into the studio, every so often someone is going to bring you a tape with mains hum on it and ask you to fix it — mains hum has a habit of creeping into the most orderly of recording setups. In a fixed installation it can normally be dealt with, but elsewhere it can be more of a problem, and it is often not spotted until after the event (which is why you have to deal with it in the studio). Fortunately, the graphic equaliser will be able to help. It can't cure the problem entirely, but if you remember that mains hum has a frequency of 50Hz in this country, and that it will have harmonics at 100Hz and 150Hz etc, then you can reduce its effects simply by pulling down the sliders closest to those frequencies. The program material will sound hardly any different. A more offensive form of mains hum is dimmer noise, produced by thyristor controlled lighting dimmers close to audio equipment and cables. This too has a fundamental frequency of 50Hz, but it has harmonics all the way up the audio band and beyond. A conventional graphic EQ can make things slightly better, but don't expect miracles.
Equalisation has countless creative uses. As I said before, you don't have to use it if the sound is OK to start with, but correctly applied EQ can make the difference between a rough amateurish recording and slick professionalism.
The first creative use of EQ is to improve the basic sounds of instruments. You may choose to do this as you record them on to multitrack tape. Console EQ and parametrics are best for this purpose — we'll save the graphics for later. If you are EQing an instrument as you are recording it, then you'll be doing one of two things. You will either be enhancing the instrument while still retaining its characteristics, or you'll be EQing so heavily that you are producing a new sound entirely. In the second case, go ahead and be as radical with the EQ as you like, because you are using your own artistic judgement — you don't need advice from anyone else. But if you are simply enhancing a sound, it's important to be careful. Usually you will want to bring out an instrument's essential characteristics and reduce any undesirable by-products. Do this sparingly and you will make subsequent overdubs and mixing easier and more pleasant. Overdo it and you'll make life more difficult later on. Remember that what you take away, you can't always put back.
During the mix, EQ is used to make the various sounds come together in the right way. Basically, you can mix for clarity or for 'thickness'. I'll explain in more detail when I come to the subject of mixdown later in the series, but essentially you can mix for clarity by finding the characteristic frequency range of each instrument and lifting it a little. Also, reduce the frequencies where each instrument doesn't contribute very much. This will produce a mix where the instruments stand out from each other.
To mix for thickness, reduce the level of each instrument's characteristic frequency range slightly. This will help the instruments blend together. Usually, if you are working with 'real' instruments then you will be trying to get more clarity. If you are working with synthesised sounds, then they are probably too clear to start with.
The graphic equaliser is also useful when you are mixing, to shape the overall sound. Plug the console's stereo output into the graphic and go through that onto tape (remembering to make sure that you are monitoring the output of the graphic — you'll do this in different ways on different consoles). Once again, I'll explain more when I cover mixdown itself, but for now just let me say that using a graphic in this way gives you an opportunity to colour the sound in a way that can't be achieved using the console EQ on a channel-by-channel basis.
In the end, correct application of EQ comes down to using your ears — if it sounds good (and if you can rely on your monitor speakers to give you the true picture) then it is good. Don't use too much EQ, think about what you are doing and why you are doing it, and the various types of equaliser will become your indispensable companions in your voyage through the wonderful world of recording techniques.
Feature by David Mellor
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!