Sound pickups

The dynamic microphone, very similar in form and function to the conventional loudspeaker, is the industry standard for most applications and is seen most frequently in the hands of a singer or presenter. However, situations can arise for which other solutions are required, say for example, if a microphone is to be worn, or if a miniature, high sensitivity or discrete approach is required. In such circumstances, hi-fi quality though possible may not be as important as proximity or placement in respect to the required source.

Proprietary systems can cost hundreds or even thousands, but surprising results can be obtained at a fraction of such costs.

When designing a sound pickup system a number of elements must be decided from the outset. The choice of microphone itself can be important. An omnidirectional type will pick up sound from all directions, whereas an unidirectional or cardioid type will exhibit a directional characteristic. A magnetic dynamic type may suit most applications offering a low internal noise figure. Electret or FET capsules, although offering higher noise, are far more sensitive (a chart comparing typical replacement capsules can be found at the bottom of the page). A choice of power source will be necessary, eg; mains, battery or both. At the same time, monitoring of an output may be required. Rugged water/weather-proofing will be vital for external use. Ideally, a complete system can be pulled out of a river, the water shaken off and be used immediately.

Sub-aqua applications have been seen where an (expensive) microphone is sealed within one or two condoms and then thrown into the sea. This approach is not recommended. A brief description of an inexpensive and effective hydrophone is given near the end of this page.

A radio-link can be useful but can suffer from disadvantages arising from weak signals, break-through, etc, especially if the source is mobile or the working environment is unknown or untested. The detection of an obvious RF source may be undesirable and encryption or frequency-hopping may be required to avoid interception. Code-division multiple-access (CDMA) has been used for many years by the military for covert and secure communications, since it is highly immune to interference, carries a high level of inherent security and can be made to appear indistinguishable from ordinary broad-band noise. Professional systems are available, some at very high cost, and can be very effective. However, unanticipated pitfalls can become apparent very quickly that can render even the best system useless. Most easily the commonest problem to address is how to survive being dropped, especially into water.

To increase directivity, say for video use, a 'zoom' facility will be required. JVC patented a method where FET mics were staggered in a vented tube. These could then be mixed into an amplifier in order to enhance the forward lobe by some 15dB. This system has been adopted world-wide with a wide variety of copies.

Sometimes a wanted signal will be submerged in extraneous noise and the actual placement of a microphone with respect to it's target can often be the major determining factor in terms of the quality of the signal obtained, a fact noted by recording engineers when say the close miking of a musical instrument reduces ambient reverberation giving a dead and lifeless quality to the recording. Electronic filtering and compression can improve intelligibility enormously, but simply piling on the amplification will not. Compact battery-driven systems will run down faster if forced to accommodate large and unnecessary signals. Low frequencies, in particular, can play havoc and these are best filtered out at the input. Roll-offs at -3dB of 1 or 2Hz are not productive if the intended target is, say, speech when 100Hz (-12dB) will give better results. In this case, an upper roll-off of 5kHz will meet most needs.

Since the operating noise level of a system can be largely due to the impedance of the source, the outputs of a number of microphones can be paralleled which then feed the input of a bus summing amplifier. Details of these are often found in the datasheets of specialist audio preamps. The layout of such an array can be important (see Bessel arrays).

Simplicity is often the key to a successful system. With the advent of mobile phones and personal digital audio, recording that was previously dependent on tape as a storage medium, has been made much easier.

Tip: if, during a 'live' performance, conference or public meeting, a lead to a mixer, for whatever reason, decides to short or open-circuit much delay and embarrassment can arise. The following circuit can be left in circuit during use to provide an instant diagnosis. Obviously, the line tested must be devoid of DC.

Naturally, in critical situations, there should always be a backup.

Collectors

To increase sensitivity, some form of collector that focusses or directs acoustic energy on to a microphone is as important as an aerial is for a transmission receiver, often the overall performance of the system, regardless how expensive, being limited by this. Extreme examples are sound mirrors.

Experiments were tried with horns and parabolic dishes to focus acoustic energy. These proved adequate and, on occasion, interesting, especially when dealing with reflections, or providing a stereo output. Redundant satellite dishes can be re-used, thus reducing design and construction efforts, although perforated types will require 'smoothing'. However, for those interested here are three web-sites that offer advice on constructing circular parabolic types; one a simple technique, another built from cardboard and another for larger sizes. Other forms exist, for example offset types for microwaves, but most constructors will find the circular form easier to build.

Inaccuracies in a dish's form can be overcome by using a funnel to direct reflected acoustic energy to the microphone, thus offering a larger target area or 'throat'.

If the dish radius equals twice the focal length (f=y=1 and x=2), the microphone supports will be flat and easier to mount. However, the dish aperture will present a signal over a 180° arc to the sensor and, whilst this can be useful with solar heating applications, cancelling can occur. Most microphone apertures will normally be less than this, say 90°, in which case a dish diameter of say 2f can then 'waste' acoustic energy (if mic aperture <90° and f=1, then x=<1). A personally preferred ratio that is easy to remember is that if f=1 and x=root 2 (1.4142135), then y=½=0.5, this supporting a microphone aperture of 140° or less. When physically matching a microphone's 'throat' to a dish, substituting a light source for the microphone will easily identify areas of 'spillage' and 'wastage' in terms of literal illumination.

It must be borne in mind that the larger the diameter of a dish (usually some 0.3-1.2m for audio) the greater it's collecting area and consequent sensitivity will be. This must be weighed against the fact that the greater area of a larger dish can be problematic in high wind speeds in terms of mechanical stability. This can be seen in the table below.

Normally a compromise is required to reduce the dish diameter to a reasonable and manageable size whilst reducing cancellation effects and/or shielding the sensor from unwanted noise emanating from behind the dish. Similar considerations are necessary for storage and transport, since most dishes are not collapsible, together with means to easily demount the sensor and it's supports, whilst being able to remount them again with easily repeatable mechanical accuracy. At the same time, it must be remembered that the total acoustic energy from a large collector can easily destroy a delicate microphone if directed at an even intermittent powerful source (see below). In essence, and rather obviously, sensitive microphones require a smaller dish area than less sensitive types. An assumption that a microphone offering 20dB more sensitivity will require 20dB less dish area can be made although an empirical approach is recommended to determine the best match for any given situation, notwithstanding any spanners that Murphy might decide to throw into the works, susceptibility to extraneous noise being but one.

A microphone fed by a dish, apart from it's aperture (the diaphragm area), should be carefully shielded acoustically to prevent audio pickup from all directions other than that presented by the dish to reduce unwanted signals. Similarly, low frequencies, say from vehicles or strong winds, can make a flimsy dish resonate which can modulate and distort the incoming signal. The back can be ribbed to stiffen it, and can have rubber sheeting bonded to it to dampen vibrations. A ring fitted, or a lip formed, at a dish's edge will help to stiffen it and provide a stronger anchorage for the microphone's/dish's supports. Lightweight, stamped 1m dishes are ideal.

A proprietary solution might include a 20" (50.8cm) segmented and collapsible dish, tripod, amplifier with 3-band equaliser, headphones, quick assembly/disassembly, tape record, patch in of a communications device, adjustable volume for each ear, automatic safety shut off at 95 decibels, tripod mount for extended use and powered by 2 AAA batteries that last about 100 hours. In ideal conditions it might be claimed that it is possible to hear a conversation 300 yards away, with a weight of 1½lbs (0.68Kgs).

To increase the aperture diameter of a small FET mic, curved 'funnels' from redundant smoke alarm output transducers proved very effective, although for external use, some physical means of reducing the input of low frequencies may be required. A dynamic type is less prone to be driven into clipping.

Although potentially extremely effective, if discretion is required a dish is probably the least subtle approach possible unless circumstances allow the use of an obvious satellite dish that has been converted, for example. Under these circumstances, benefit can accrue if the dish is remotely steerable, although the system then may become unnecessarily complex, especially if a visual (camera) output is required as a consequence. Transparent (polyethylene, polycarbonate, etc) types can be obtained reducing visibility. However, the presence of these, if detected, will be a certain give-away. Mottled, drab paint finishes (with matching fabric overlays) can break up a dish's outline, with skill, to the point of invisibility.

In one situation, a wide area was covered by aiming a dish at a slight angle to a large wall adjacent to the target area, reflections from which gave the desired results. Architectural curves can give very distinct acoustic qualities to a building, the Whispering Gallery in St Paul's cathedral being a good example. Some chapels seen have curved ceilings that then merge into the walls presenting an unbroken surface. These tend to reflect any music and singing back down onto the congregation. One former Laura Ashley shop that had been extensively remodelled had a 'dish' in the ceiling above the till area. These principles have been observed in earlier times, for example, take Greek theatres or the very ancient Chalcolithic Goddess Temple on Malta which has chambers cut into the rock which to some ears are harmonic, and obviously resonant, like singing in the shower (see sonics).

Some satellite dishes are offset in that the head will not be aligned on the dish's centre-line but, in normal use, will be angled upwards. For audio, invert the dish so that the microphone is pointed towards the ground, thus helping to prevent the pick-up of aircraft noise, etc, and aiding weather-proofing. Offset types, when rotated 90°, can be more effective for stereo arrangements when two funnels are combined in one 'head' to direct sound on to separate mics. Other non-standard formats are workable too, eg;

Alternatively, for long range use, rifle mikes using a single 'barrel' have been used. These will usually consist of a pipe, open at one end with the mic at the other, to isolate the sound source aimed at. Perforations and damping prevent the tube from resonating.


More complex, but arguably more sensitive, is the 'shotgun' design described by James Hollinger and John Mulligan in the June 1964 edition of Popular Electronics (US). In this approach 37 3/8" aluminium tubes, each an inch shorter than the next (36" to 1") were combined in a hexagonal bundle which was then presented via a funnel to a single crystal microphone. Each tube resonates at a wavelength twice that of it's length giving exceptional sensitivity and directivity. In this case, a frequency response of 186 to 13,392Hz is covered (for speech, 24 tubes of 2-25" can suffice). An unwanted signal at a particular frequency can then be reduced by stopping the relevant tube/s. With the active pickup described below, sensitivity was greatly enhanced. To reduce noise further, say from power-lines or transmitters, a silver-loaded epoxy was used to bond the tubes which were then used as the primary earthing point (spiked out-of-doors).

Provided rigidity problems are addressed, there is no reason why plastic tubes cannot be used, although a metallic structure is stronger and less prone to breakage. As a weight/strength compromise, the longest tube only maybe metal. Advantages, in terms of surveillance say, arise from the much smaller profile presented to the target and portability, compared to a dish.

Some success was obtained subtracting signals - one mic was given a small aperture and a target focussed upon whilst the signal from a second 'wide' mic, in close proximity to the first but aimed in a dissimilar direction, gave a second 'background' signal that could be inverted and subtracted from the first. This arrangement can be improved upon to provide a more discreet form than say a rifle mic or dish. A Bessel array can yield a super-sensitive microphone with a spherical radiation pattern.

With some of these approaches it is possible to record conversations through walls (even in stereo) and across streets through a pane of glass, but every consideration must be accommodated to reduce the input of unwanted noise, including that emanating from the operator. Handling can be reduced with a tripod, it then being possible to pan the microphone with ease. This can be useful since wind can carry a wanted signal away from it's source. Physical isolation of elements, via an air-gap, was found to be best. Wind noise can be curtailed by shielding the pickup with fabric. Similarly, despite water-proofing, the sound of rainfall hitting a mic and/or collector will effectively drown out a signal. In some cases some form of remote handling can be advantageous and, if possible, at the outset of a design process should be considered to increase the flexibility of application.

If experimenting with collectors, it is recommended that provision be made for 'heads' (combined microphone and preamp assemblies) to be changed easily in situ. For example, depending on the situation, a low-gain preamp with an electret capsule may give better results than a high-gain design with a dynamic one. Some designs can include both so that if one head is overmodulated into clipping the other can be switched or mixed in immediately.

The validity of constructing an array for a single pickup can be questioned on the grounds of cost and complexity. However, apart from advantages of reduced noise and greater sensitivity, survival of concentrated and rapid pressure transients from, say, a dish, that can easily exceed a mic's design load is also worth considering. A well-made array can 'lose' half, or more, of it's sensors and still function well, whereas the loss of a single mic in a 'single' system renders it dead and, therefore, useless. A removable physical 'mute' can be considered to reduce sensitivity and/or to protect delicate diaphragms from destructive gradients such as those involved with powerful acoustic sources like thunder, jet engines, fireworks, PA systems, crowds or gunfire. A simple trick, borrowed from aligning satellite dishes, dropped the input by 10dB or more by throwing a towel over the array.

Preamps

A very wide dynamic range of input sound pressure levels, ranging from about 30dB SPL (ambient noise in a quiet room) to over 130dB SPL (pain threshold) can occur. The output of a low impedance (200R) microphone might then typically vary from 20μV to 2Vrms, while it's self-generated output noise would be in the order of some 0.25μV over a 20kHz bandwidth. Ideally since this dynamic range is so large, a preamp's gain should be adjustable so that it can be optimised for the signal levels that can arise in a number of different situations, ie; large signals should be handled without clipping or excessive distortion and small signals should not be degraded by preamp noise.

A good preamp should contribute no more noise to the output signal than does the source impedance. In practice however, it is often considered reasonable to allow a higher level of input noise in the preamp since ambient room noise will usually cause a noise voltage at the microphone output terminals that is on the order of 30dB greater than the microphone’s intrinsic (due to it's source resistance) noise floor. It is eminently feasible to construct designs with a noise floor >20dB lower than a typical microphone’s output from the 30 dB SPL ambient noise level in a quiet room, this being critical if the intended target has a small amplitude.

When long cables are used with a microphone, interference can be expected, especially from mains hum and RF pick-up. To reduce this problem, the outputs of most professional microphones are balanced, driving a pair of twisted wires with signals of opposite polarity. In theory, magnetic fields will induce equal voltages on each of the two wires which will then be cancelled if the signals are applied to a transformer or differential amplifier. Avoiding the use of transformers has several advantages, including lower cost, smaller physical size and reduced distortion. To prevent interaction, signal and power cables should never run alongside each other and if they must cross should do so at right angles.

Ideally, a microphone should be placed in close proximity to a low-noise preamp thus assisting the integrity of the original signal. Both should be integrated into the same assembly, which is then screened, preferably with steel, although larger diameter copper piping with it's associated fittings can be useful.

Below, in approximate historical order, are a number of designs for both single-ended and balanced applications. A considerate approach from the early '70s improved the bias stability and removed the subsonic ringing evident with some designs. The use of MF resistors and separate hum filters in one build reduced the noise floor further, as can the use of ZTX107C low-noise transistors.

Linsley Hood's exploration of his 'Liniac' paid attention to other simple arrangements.


Some microphone (and RIAA) preamp designs have used paralleled arrays (up to 10 have been seen) of input stages to reduce noise.


Older designs might use BC109C/179C combinations, like the input amplifier used in the Revox A77 (below).


Two types that were used frequently in the '80s included the 2SA1085E and the 2SC2547E. Some later designs have paralleled J-Fets, like the 2SK170BL, with a common bias network for a high-impedance input, although a simpler solution might suffice, eg;


and 'bootstrapping' can be employed.


The next design, although simple and definitely not low-noise, proved very effective as a room pickup.


Both crystal mics were mounted 180° with respect to each other (one at each end of the enclosure) helping reduce induced mains noise, and 741 opamps were used with a 50Hz twin-T notch filter, run from a ±12V supply. High overall acoustic sensitivity results with a self-cancelling of unpleasant reverberation effects. Very easy to upgrade.

A JLH microphone preamp from his '82/3 modular preamplifier,

although there are the 'liniac' designs that might interest.


Designs intended for moving coil cartridges can be used offering lower operating voltages, input noise resistances and a more compact approach if using a medium-power input transistor with a low Rbe. This is useful if low-impedance paralleled arrays are used.

One design compared a variety of types, a MJE13007 giving better performance than a BF459, MJE340 or two BF459s in parallel. Noise floors with these types were some 15dB or lower than using a 'standard' low-noise type like a BC549C.

Dynamic microphones are best used with a differential input. Some instrumentation amplifiers can be modified for optimum audio usage.

Used in high quality mixers this low-noise version will also run at the low levels inherent with battery supplies.


For surveillance, the frequency response can be restricted to reduce unwanted input and would normally be used with a limiter or compressor (see NE571 circuit below). For speech, the usual bandwidth is 300-3kHz, although a slightly wider range, or low rate of roll-off (12dB or less), will give a more 'natural' sound (a useful variable low/high pass filter is shown below). On a dish, the gain control would be accessible from the front of the 'head'.


Two applications from the same manufacturer are shown below.


To adjust the CMRR, or Common Mode Rejection Ratio, short both inputs together and apply a signal between these and ground, then set the CMRR control for a minimum signal on the output with a number of progressively higher inputs until no meaningful improvement is obtained.

Low-noise ICs can also help to reduce the component count offering, with care, a succinct solution. Analog Device's SSM-2017 offers typical noise of 950pV (gain 1,000) using a single resistor to set the gain.

The NE5534 opamp has been a personal favourite over many years (still found in nine out of ten CD players as an amplifier in the analogue section) and On Semiconductor's AND8177/D application note gives a good range of designs, although be mindful of the numerous errors that it (Rev. 0) contains.


A similar arrangement, and that for a lower noise spec transformer coupled design, is given in the Analog Devices OP37 (rev. B) datasheet together with useful 'Comments on Noise'. Another arrangement that substitutes the input pair.


This article is useful for circuit details of filters used in noise measurement.

Below are more complex approaches reminiscent of earlier paralleled designs. Refer to the device's datasheets for more detailed discussion.

Brief typical specs of some low-noise opamps are given below for comparison. However, it should be borne in mind that to achieve low voltage noise with a bipolar input, high collector currents may be used at the expense of current noise performance (this being directly proportional to the square root of the collector current) thus the requirement for a low source impedance. With a J-FET input, current noise is considerably less and this should be taken into account in noise calculations.


High frequency types, like the AD797, might require more careful supply bypassing when driving heavy loads. For this reason datasheets should be consulted before using a 'new' IC. There is nothing wrong with testing a variety of opamps in a design to determine which gives the best performance. IC sockets, if used, should be of a high quality (turned pin) but bear in mind that the device dissipation will have to be derated to account for the higher chip-to-ambient thermal resistance.

A discrete JLH opamp with FET inputs is shown below.

High performance low-noise ICs like the PMI SSM-2015/6 are worth considering, although their lower operating voltage limits (±9-12V) may not suit some applications. Low-noise instrumentation types like the INA103 and INA217 can offer excellent performance versus component count.

A compact and sensitive active microphone element consisted of a FET mic (or more paralleled, seven for symmetry) feeding a ZN459, and later SL561, linear IC and have proved very effective, the use of tantalum capacitors offering very compact builds.

A specific bandwidth can be easily tailored to suit the application, and a balanced transformer added. Output impedances are low enough to drive head or earphones directly. Ideally, multiple mics should be matched if an array is used. With a fixed gain of 1000 (ZN459), the input is attenuated physically with baffles, some kind of suspension usually being necessary to prevent pickup of, say, normally inaudible traffic noise which can 'swamp' a system rendering it unusable. Rubber suspensions, as found in some cassette recorders, which cover the entire capsule, bar the opening to the electret membrane, are a good start. Unshielded elements and even moving-coil cartridges can be used to detect sub-sonic and other non-air vibrations.

The following variable filter was used to subjectively determine bandwidths before a fixed type was built.

A fixed 2nd order filter for speech frequencies.


Filters can be cascaded to give steep slopes.


Mains hum and rectification byproducts may have to be removed. To this end 50Hz and 100Hz (UK) filters can be useful. The values then chosen for one circuit can then be easily doubled or halved (paralleled or put in series). When the output of a filter is fed back to a node that is normally grounded the notch width can be varied.


This can be useful when, with temperature drift, f can vary. Although there is nothing wrong with ±5% carbon resistors, for best results use ±1% metal-film resistors and polystyrene, mica or polypropylene capacitors for filtering and equalisation. A low temperature coefficient is considered vital. With the node tied at close to the same potential as the output, the notch becomes vanishingly small and peaking can occur as can oscillation.

A compressor can reduce the distortion from high peaks and these can take a variety of forms

and levels of complexity.


AGC (automatic gain control) ICs are available. For audio a fast attack and a slow decay are usually required. Types like the older 76020 had the attack/decay times set internally. The SL6270 gain controlled microphone preamplifier (VOGAD) compressed and limited the output, reducing overmodulation and clipping.



Some applications may benefit from a variable ratio of overall compression,


A fixed version transformed some cassette decks outperforming the Dolby NE545/646B ICs used.


An interesting IC, the Analog Devices SSM-2167, offers a low voltage microphone preamplifier with variable compression and noise gating, all in a 10-pin small outline package.


For monitoring purposes, head or earphones may be required.


IC options to drive headphones include the TDA2822, LM386, TDA7052, TBA820M , LM380 or the class D MAX4297 (>83% efficient). A discrete low-power design intended for solar use can be found here and a simple but adequate design is shown below.


A suitable tone control for this can be found here.

When prototyping, habit has dictated that two preamps be built, then if a problem arises or modifications are made to one, immediate comparisons with the other can be made. Invariably, it is found that a completed (stereo) 'twin' is a good idea and, once tooled up for one PCB a second made at the same time is quicker and easier than repeating the process at a later date.

If mounting axial-leaded resistors in an upright position to reduce PCB area, the longest lead should be connected to the least impedance 'seen', thus reducing the area of a 'sensitive' high impedance side. The same can be said of polystyrene caps which if marked at one end indicates that wire is connected to the outermost foil and is then ideally connected to the least impedance. Stripboard can be useful to about 5MHz. Finish board by cutting runs to minimum length required and binding unused strips to ground. Tinning connecting runs will reduce impedances. Finished and tested PCBs should be cleaned and tropicalised as a matter of course, since greasy fingerprints can lead, in time, to corrosion.

If enclosing or say potting in epoxy to waterproof, bear in mind capacity of cables, if their cores are exposed, to draw water along them.

Soldered connections on microphones can be varnished and the 'face' or diaphragm end can be covered with a water-proof fabric, such as Goretex, which can be glued in place. Capsules should ideally have a metal case, but if plastic is unavoidable this can be given a couple of coats of nickel-loaded paint which can be bonded to the 'head's' metal-work.

Whatever combinations are built, the primary aim is for the lowest possible signal to noise ratio. For best results avoid carbon resistors and use low thermal coefficient 1% metal oxide/film types. If budgets permit, precision wire-wound types can be used. Be mindful that some ceramic capacitors are microphonic and if subjected to movement or contact can act as pickups in their own right! For the same reason both microphone and preamp should be mounted in a soft suspension within a screened enclosure. Although two-part foams can be used for filling, etc, something as simple as bubble-wrap can be very effective, and even better!