The Rough Guide To Choosing Microphones
Dave Lockwood presents the ultimate guide to microphone basics — how they work, how to choose the right mic for a given application, and how to use them.
Microphone design and usage is an intriguing subject, for despite ever-advancing technology, which in other fields has resulted in far more predictable performance, microphone choice remains in many ways such an inexact science.
Results can never be guaranteed; sometimes a unique combination of instrument and acoustic will inexplicably suit one model more than another, but an apparently identical set-up on another occasion may produce the opposite result. Nor are the most expensive models necessarily always the best mics for the job. Even in the most esoteric recording circles, there are still situations in which a basic £100 model will give you exactly what you want, when an up-market £1,000 model cannot.
Not surprisingly, this makes microphone choice, whether you are a first-time buyer trying to identify the right one to use with your portastudio, or an engineer selecting the best one for the job from a well-filled studio stock-room, something of an art. It is often said that there is no substitute for experience in this field, but the fact is that a little basic understanding and a few guidelines will go a long way.
Although microphone specification sheets may appear, to the inexperienced eye, to contain a lot of rather obscure technical information, you can actually get away with knowing only four key facts about a microphone and still be able determine its suitability for usage in a particular application.
• OPERATING PRINCIPLE
Is the mic a dynamic or a condensor? Other types are merely variations of these basic sub-divisions; electrets are fundamentally condensors, ribbon mics are fundamentally dynamic. Even PZMs are just condensors with a specialised housing.
• DIRECTIONAL CHARACTERISTIC
Does the mic have a directional bias, like a 'cardioid', or is it an omni, equally sensitive to sounds from all sides? (See box).
• FREQUENCY RESPONSE
Is it a full-range, basically 'flat', model, or does it have a tailored response such as a presence peak or low-frequency roll-off?
Does it produce enough output to pick up signals at a distance, or is it optimised for close-miked working?
However limited your technical knowledge, an understanding of just these major points will tell you immediately what makes, for example, a Shure SM58 so suitable for hand-held stage vocals but rather less accomplished at recording acoustic guitar.
A superficial appreciation of operating principles, although not actually essential, is certainly worthwhile for anyone looking to get the best out of their microphones. Dynamic (moving-coil) models use the simplest possible principle for converting sound energy into an alternating voltage. A lightweight membrane, called the diaphragm, is suspended at the end of the mic housing; sound waves (consisting of periodic variations in air pressure) move the diaphragm when they strike it, and this in turn moves a coil of wire attached to the diaphragm. Magnets are placed around the coil, and as you may remember from physics at school, moving a coil within a magnetic field induces an alternating voltage in the coil.
The mechanical simplicity of dynamic mics, and the absence of any on-board electronics, allows them to be robust and reliable, hence their popularity in stage and hand-held applications.
Ribbon microphones, seldom now encountered in the mainstream of contemporary recording, also fall into the dynamic category. An extremely thin corrugated metal strip suspended under tension in a magnetic field acts as the diaphragm. With both sides of the ribbon equally exposed, figure-of-8 response is most common.
Although capable of excellent results, their inherently physical fragility and a marked susceptibility to unwanted low frequency vibration (wind and impact noise etc.) limits the usage of ribbon mics — it would be possible to damage the delicate ribbon simply through use in an inappropriate application! However, a smooth, extended frequency range and excellent transient response ensure that they continue to be used in specialised applications like 'purist' classical music recording.
An inherent factor governing the performance of dynamic mics is that the diaphragm assembly must be sufficiently substantial to be able to move the coil, therefore its ability to respond to the very low energy pressure variations at high frequencies is restricted by its own inertia. Consequently, whilst many dynamic models are highly regarded in specialised applications, and for use where their other attributes are at a premium, such as on-stage, they would not normally be first choice to make a full-range recording.
One company (AKG) tackled this by producing dual-diaphragm models, with a second, smaller, lighter diaphragm for high frequencies — rather like a dual-concentric speaker in reverse. Models in this category (D224, D202) can give a high quality extended-range output for a dynamic, but at moderate sensitivity, making a good low-noise mic amp essential.
The use of more powerful rare-earth magnets as in the Electrovoice 'N-Dym' series (using a Neodymium alloy) creates an inherently more efficient electro-magnetic transducer. A lighter diaphragm/coil assembly can then be used, achieving both higher output and extended frequency response. This is the principal difference between Shure's upgraded Beta 58/57 and the standard versions.
Diaphragms can be made lighter still however, by doing away with the coil altogether. In condensor microphones (also known as capacitor or electrostatic) the movable diaphragm is made of metal or, more commonly these days, metalised plastic film, and is placed parallel and very close to a similar structure which is maintained in a fixed position, referred to a 'back-plate'. The two surfaces then act as the electrodes of a capacitor (hence the name), with the two electrodes being 'polarised' (charged to opposite polarity) by the application of a DC voltage. The changing capacitance when the diaphragm moves with respect to the back-plate then produces across the electrodes the alternating voltage that we want.
If none of that makes any sense at all to you, you can forget about it; the only bit you really need to know is that condensor mics need power. Not only must the polarising voltage be found from somewhere, but the very high impedance output of the condensor capsule needs to be converted to a more robust signal, hence there is always a small preamp on-board. The presence of active electronics has allowed the output transformer, once a most essential feature of all types of microphone, to be omitted now on many condensor models, with a corresponding gain in subjective transparency.
Some condensor mics do not require a polarising voltage at all, and merely employ on-board batteries solely to supply the low-current needs of the integral pre-amp. These are 'permanently polarized', or electret, models, in which the diaphragm has been fixed with a virtually permanent charge by the application during manufacture of a strong DC electrostatic field. It is 'permanently charged' much in the same way that a permanent magnet is 'permanently magnetized'.
Early electret models often had a rather restricted performance, particularly in the area of pre-amp headroom due to the current delivery limitations of batteries, but many models now compete with 'true condensors' on equal terms when running on phantom power. Often they will automatically detect the presence of phantom power, switching to their internal battery only when necessary.
Electrets are extremely convenient to use in budget systems, for they can be run balanced or unbalanced, and some of them are rather economically priced in view of their comparative level of performance. In terms of basic recording fidelity, a moderately priced electret condensor will comfortably outperform a comparably priced dynamic model. Moving further up-market, some of the very finest models available now use the electret principle. The use of expensive condensor microphones with budget equipment is by no means a waste of time; starting-off with the highest possible quality signal at source at least gives the rest of the system a chance of displaying its full capabilities.
Does this mean that condensors are inherently superior to dynamics? Certainly not. Each has role to play; each has its strengths and weaknesses. Condensors undeniably have superior sensitivity, and more extended frequency response, giving an overall more refined sound. Dynamics in close-miked applications, on the other hand, often offer more 'punch' and character. This may be artificial, but not all recording is about accuracy.
Quite why a 'less accurate' mic should sometimes have more 'impact', particularly on drums and close-miked speaker cabinets, I have never been able to satisfactorily determine, although it has been suggested that it is a function of the moving-coil diaphragm more readily mirroring the transient excursion of a drum head or cone-driver. What I can more readily accept is that you don't normally listen to a drum with your ear two inches above the skin; the inherently 'unnatural' response of some specialised dynamics is perhaps compensating for an inherently 'unnatural' sound at the mic position to produce a musically useful output.
There can be no overall 'best' microphone, for you cannot assess a model without first deciding on the context for which it is being evaluated. Change the application, and it could move from the top to the bottom of the ratings.
Choosing microphones for applications used to be simple, based on the acknowledged limitations of the two main mic types; if the source was very loud and close-miked, you needed a dynamic mic; if the source was very quiet, or if it was a signal with a lot of high frequency detail, it simply had to be a condensor. If it was very loud and needed full-range pick-up... well, you were in trouble. Fortunately, continued improvements in design, manufacture and materials have reduced the need for distinction along these lines; for instance, there are now many models of condensor that you can stick right in front of the bell of a trumpet, such is their tolerance of high sound levels.
Microphone directional behaviour is a vast subject in itself, but we need only touch on the general principles here. Once again, mics can be divided into two main groups; directional mics, with greater sensitivity in one direction than another, and omnidirectional types which, in theory, pick up equally at all angles. It gets a bit more complicated when you start to look at the way in which the nominal directional characteristic actually changes at different frequencies, but in practice the sub-division is still clear.
By far the most common directional type is the 'cardioid'. Many people whose main experience of microphones is restricted to usage of stage mics with a PA system mistakenly believe that this means that the mic will only pick up what is immediately in front of it. It is true that when speaking into a PA mic you appear to get very little out of it until you are right on top of it, but that is actually for an entirely different reason. The name cardioid implies 'heart-shaped', for that is what a two-dimensional representation of the pick-up pattern of a true_cardioid model resembles; a frontal lobe (projection) with two smaller lobes to the rear on either side of the mic housing.
The directional effect is achieved by complex techniques of phase cancellation, in which sound is allowed to reach both sides of the diaphragm. This is sometimes referred to in spec, sheets as 'pressure gradient' operation. The alternative is simple 'pressure' operation, with only the front of the diaphragm exposed, producing an omnidirectional characteristic, offering theoretically equal pick-up from all directions.
The majority of directional mics actually suffer some performance loss as an inevitable consequence of the way in which their pick-up characteristic is generated. The original derivation cardioid designs used the phase cancellation that arises from combining the output of two virtually coincident transducers, an omni and a figure-eight, but the majority of models, particularly dynamics, now actually use acoustic delays, via differing sound path lengths to the diaphragm, to achieve a similar result.
Cardioids gained much of their popularity for their ability to reduce feedback in sound reinforcement (PA) applications, but were always recognised as inferior to omnis in recording applications. The development of multitrack recording however created a much greater emphasis on separation of signals, such that the cardioid soon became the practically universal recording tool. Omnis were reserved almost exclusively for classical music recording, where their extended low frequency response and inherent freedom from colouration was appreciated as more important than directional discrimination.
There are two primary factors to be be considered when evaluating a cardioid for recording purposes; the proximity effect and off-axis colouration. The proximity effect results in heavily emphasised low frequencies when the microphone is placed very close to the source. This can be either a problem or an advantage, depending on the application. Singing voices can be greatly flattered by the proximity effect, giving a fullness and power that is not present in the voice itself, but a speaking voice, where we are perhaps far more tuned-in to the real thing, can sound totally unnatural — listen to the 'voice-over' on some TV commercials and try to think if you have ever heard a voice sound like that other than via a microphone.
We are so used to hearing vocals via proximity-effect mics that a more realistic recording can sometimes seem a bit thin, especially to the vocalist. Rock vocals in particular simply demand the additional power of proximity-effect boosting. Virtually all mics designed for hand-held stage applications exhibit the proximity-effect to a marked degree, indeed in some of them the low frequency response is deliberately curtailed to some extent to prevent it from becoming excessive. This can either be a fixed characteristic, or via a choice of switched low-frequency roll-off settings. The proximity-boosted LF region is often balanced by incorporating a 'presence peak' in the upper midrange, around 3kHz, giving some 'cut' to go with the artificial warmth and power. It is a very odd response really, but it works in the context for which it was developed. It is also the reason why this type of microphone is not favoured for recording other than in close-miked applications.
A full-range cardioid recording model, naturally, will not have a deliberately curtailed LF response, although relative to an omni, it will never be as extended at distances above a few metres (an inevitable side-effect of the pressure gradient principle; this will not be evident from the spec, but that is a function of the measuring technique). Such models give their most neutral response at 'normal' working distances just outside the proximity zone; when used close-up, the LF can be out of control unless it is externally rolled off.
Although nominally 'flat', many highly-rated recording models in fact exhibit some degree of deliberate response tailoring, usually a gentle rise in upper-mid or high frequencies to lend a flattering detail and clarity to their sound. In many models a tendency towards this type of response was specifically designed-in to exactly mirror the loss of transparency and slight HF curtailment inherent in the analogue recording process.
The widespread use of digital recording systems exposed some well-thought-of designs as over-bright for work where accuracy was of prime importance, leading to the development of more neutral second-generation versions, usually with increased dynamic range as well (such as AKG's 'ULS' versions of their C414 and 451). This does not make the 'new improved' version inherently better; many individual signals benefit from a touch of extra brightness when recorded, perhaps to compensate for close-miking, or a dull acoustic, and an appropriately assistive microphone characteristic will always do it more smoothly than EQ.
"Compared to the latest 'black-box-of-electronics' signal processor, for which we seem to think nothing of paying hundreds of pounds, modern mics offer unsurpassable value for money."
The second essential factor governing the performance of a typical cardioid is the effect that the method of generation of the directional characteristic has on the frequency response. As you move further to the side from the head-on position, there will be phase-related cancellation and reinforcement at different frequencies; the smoothest on-axis response can sometimes look decidedly ragged towards a 90° 'angle of incidence' (direction of arrival of the sound).
The specification's 'polar diagram' (the superimposed 'plan-view' plot of the pick-up pattern at several nominal frequencies across the range) will doubtless look very neat and tidy (funny, they always do!) but reality will rarely be able to match it (real polar patterns are three-dimensional to start with, and not many models live up to the claim to be 'rotationally symmetrical about axis'). In close-miked applications this will actually make very little difference to the end result for the ratio of on-axis to off-axis sound will greatly favour the former. At normal recording distances however, where reflections arriving from all directions form a significant proportion of the sound, off-axis colouration is the most significant limiting factor in the performance of some models.
For the absolute minimum off-axis colouration, and no proximity effect either, you could always choose an omni instead of a cardioid. But wouldn't that mean losing all control over what the mic picked up? When you consider that the ratio of direct to reflected sound for an omni is about the same as for a cardioid placed at roughly one and a half times the same distance, you have to conclude that, in some applications, the cardioid is not giving you as much control as you perhaps thought.
With no LF proximity-rise, you can sometimes work closer with the omni, and you will find that it is quality, not level, that really counts when it comes to dealing with 'spill' or inter-mic crosstalk. There is nothing like miking up a drum kit with a set of cheap dynamics to illustrate this.
Every mic that you point at a drum is also picking up its neighbour, but with a modified frequency response. The limited HF models that you decided would be adequate for the toms are also contributing a strangled, highly coloured rendition of the cymbals; the cardioid snare mic has the hi-hat 135° off-axis, in one of its 'dull zones', and is making it go 'clack' instead of 'tsss'! Gating will work fine, until the drummer starts playing the whole kit, when the sound will be rendered with an uncontrollable nasal tonality, inexplicably lacking in punch.
Because the spill is so tonally unnatural, it attracts the ear's attention; it is all too obviously not part of the original sound, and EQ will make it worse, not better. Unless you have enough high quality microphones to do the job properly you are much better off miking a kit with just a bass drum mic and a pair of full-range overheads, reinforced with a close mic on the snare (for a bit of proximity effect artificial depth), if possible.
The introduction of omnidirectional models capable of withstanding very high sound pressure levels has led to something of a re-appraisal of the options, and many more engineers are now using them, where possible, for their more 'open' sound in close-miked applications.
It will be evident already that there can be no single 'best' microphone in any application. A set-up that sounded great in one situation can be totally wrong for another. Why go to all the bother of changing microphones around though? Why not just EQ it until it sounds right?
Microphone frequency response variations are generated by their physical properties. They are unique to each type, and have a level of complexity far beyond anything that can be replicated with an equaliser. Forget about the ruler-flat responses and smooth curves you see on spec sheets; the majority of real mics, particularly cardioids, almost have a 'frequency signature'. You can always take technical measurements in such a way as to disguise it, but you can't fool ears.
Microphone selection sometimes comes down to determining the most desirable (or least damaging) 'character' to be imposed on the source. Obviously, it pays to have the microphone working for you to begin with, before you start EQing.
Curiously, some mics EQ rather better than others; the models that sound best flat are not the ones that sound best EQed, and vice-versa. In general, dynamics for close-miked work will take radical EQ a lot better than any other type, perhaps because the sound is inherently less natural to begin with.
The sensitivity of microphones (how much electrical output they generate when exposed to a particular sound pressure level) actually varies enormously from model to model. The greater the output, the less amplification is required, and less amplification means a better signal-to-noise ratio. A large diaphragm will give more output, but a small diaphragm can more closely equate to the theoretical ideal of the point source pickup (and its smaller housing will interfere less with the soundfield around the capsule).
The situation for the novice is clouded by the fact that microphone specifications do not always quote sensitivity in the same terms. There are two main methods used, the most common being to specify the open-circuit voltage output for a standard sound pressure level (SPL) in millivolts (mv). The level conventionally chosen is a fairly moderate one, 74dB SPL, which roughly equates to talking into a mic from a distance of about eight inches. This normally gives an output of somewhere between 0.1 and 0.2mV for a dynamic model, and usually around 1.0mV for a condensor. Another method of specifying this same figure is in millivolts-per-Pascal (mv/Pa) — with 74dB SPL being 0.1 Pascal, you can expect a single figure specification for a dynamic, getting into double figures for a condensor.
If you see a sensitivity specification in 'dBV' on one model and mV on another it obviously makes direct comparison harder. If you remember simply that -60dBV equates to the 1 mV output of the condensor, and -80dBV to the dynamic at 0.1 mV, you have the whole of the commonly encountered range covered and will be able to roughly calculate any equivalent figure. Even when a spec sheet uses yet another method, such as a 'power level' (milliwatts/microbar) fortunately there will invariably be an equivalent figure for one of the other conventions.
Normal working level output will be much higher than the 74dB SPL figure, however, typically around 2mV for a dynamic and perhaps 100mV for a condensor. The recent trend towards supply voltages of more than 48V in some condensor mic systems can result in an output of several hundred millivolts, virtually a 'line level' signal. As you might imagine, the dynamic range of a model where the pre-amp is powered from a 100V supply compares rather favourably with that of one that receives only 1.5V from a battery!
Microphone 'self-noise' is invariably low enough, in well-engineered modern designs, to make the specification of the mic amp the limiting factor in the noise performance of the system. Designing the microphone input stage of a mixer is quite an art, for it is required to deal with a greater range of signal levels than any other part of the signal chain. It must avoid introducing undue noise on the quietest of distant miked acoustic sources, and yet be able to handle the output from a sensitive mic shoved right up against a guitar speaker or a snare drum, without distorting. Some mic amps stages are fitted with an attenuator or 'pad' (usually about 20dB) to help cope with this degree of variation. High output condensor models are also sometimes fitted with their own on-board output attenuator.
To assess a mic amp, look for a noise figure, expressed as 'Equivalent Input Noise' (or just 'EIN'), of around -120dB or more (the greater the negative figure the better). EIN is measured at maximum gain, with a resistor across the input simulating a typical microphone impedance, usually 150 or 200 Ohms. Be wary of direct comparisons between systems however, without ensuring first that all the test criteria are the same (they should be specified along with the EIN figure).
With the majority of microphone designs having an output impedance of around 150 to 200 Ohms, they will ideally want to 'see' around five to ten times this value as a mic amp input impedance. 1 or 2kOhms is common in modern equipment (although anywhere from 600 Ohms upwards will be fine). Balanced line operation makes impedance 'matching' non-critical, allowing long cable runs without significant losses.
Just what are you paying for when you lay out hundreds of pounds for a microphone? Are you getting value for money to the same extent as in other fields of recording technology? Well, I certainly think you are. What you are paying for are almost unbelievable manufacturing tolerances, without which some designs would not be viable; you are paying for literally years of research into new materials and continuous re-evaluation of the fundamental principles of mic design; you are paying for a guaranteed level of performance from an electro-mechanical system that is pushing at the boundaries of what the basic laws of physics will allow (consider that a condensor mic diaphragm can be as thin as 1 or 2μm, and must be able to respond to everything from a whisper to a gunshot).
In some of the highest quality models, tolerances are beyond automated assembly. Their quality precludes mass production techniques, and limited volume precludes economies of scale. The leading-edge microphones, as in other technologies, are always going to be relatively expensive. As for the rest, in comparative terms, I don't think they are expensive at all. A modern dynamic mic is no longer just a membrane, a coil and a magnet, it is an incredible piece of physical engineering featuring a precision tolerance voice-coil gap, perhaps a rare-earth magnet, sophisticated multi-stage directional tuning network, an advanced shock-isolation system and a humbucking coil. Compared to the latest 'black-box-of-electronics' signal processor, for which we seem to think nothing of paying hundreds of pounds, I think modern mics offer unsurpassable value for money.
Just a few years ago the studio press was full of articles bemoaning the demise of 'mic technique' among young engineers brought up solely on a diet of MIDI and samples. It was said that we would soon have a whole generation who thought that mics existed solely for sample origination! The prophets of doom were, as ever, a little premature, and the inevitable 'back to real music' reaction soon reversed the situation. Actually, there has probably never been a more interesting time for the microphone enthusiast.
New materials, new manufacturing techniques, and a general re-appraisal of requirements in the light of new recording media have all led to a considerable resurgence of interest in microphone principles and technique.
There is no better way of getting to know microphones than by using them, with time to experiment. Always try another mic, even if the first one you put up sounds perfect — how do you know that the next one won't be even better? Only by actually going through this process can you really acquire a useful instinctive feel for what each general type will do. An understanding of principles on its own is of limited worth, but allied to a little practical experience and the inevitable development of some personal preferences, ultimately makes for the perfect combination of informed and instinctive microphone choice.
As 'the first signal processor' in the chain from acoustic source to recording medium, the microphone will never lose its importance, whatever technological developments may overtake its current form. Long after the rest of the program chain has attained the uniformity of 'digital perfection', microphone choice and evaluation will, I believe, be able to remain an intriguing mixture of prejudice, experience, constant learning and accidental discovery. Choosing microphones is seldom a matter of right or wrong; it remains a welcome area of personal expression.
Feature by Dave Lockwood
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!