Loudspeaker

Closeup of a loudspeaker driver

A loudspeaker or speaker, is an electromechanical transducer which converts an electrical signal into sound. The term loudspeaker is used to refer to both the device itself, and a complete system consisting of one or more loudspeaker drivers (as the individual units are often called) in an enclosure. The loudspeaker is the most variable element in an audio system, and is responsible for marked audible differences between systems.

History

Alexander Graham Bell patented the first loudspeaker as part of his telephone in 1876. This was soon followed by an improved version from Ernst Siemens in Germany and England (1878). Nikola Tesla is believed to have created a similar device in 1881 ^[1]. The modern design of moving-coil loudspeaker was established by Oliver Lodge in England (1898). [2]

The moving coil principle was patented in 1924 by two Americans, Chester W. Rice and Edward W. Kellog. There is some controversy in that an application was made earlier by the Briton Paul Voigt but not granted until later. Voigt produced the first effective full range unit in 1928, and he also developed what may have been the first system designed for the home, although using electromagnets rather than permanent magnets.

These first loudspeakers used electromagnets because large, powerful permanent magnets were not freely available at reasonable cost. The coil of the electromagnet is called a field coil and is energized by direct current through a second pair of terminals. This winding usually served a dual role, acting also as a choke coil filtering the power supply of the amplifier which the loudspeaker was connected to.

The quality of loudspeaker systems until the 1950s was, to modern ears, very poor. Developments in cabinet technology (e.g. acoustic suspension) and changes in materials used in the actual loudspeaker, led to audible improvements. For example, paper cones (or doped paper cones, where the paper is treated with a substance to improve its performance) are still in use today, and can provide good performance. Polypropylene and aluminium are also used as diaphragm materials.

The first commercial acoustic suspension loudspeaker was developed by Henry Kloss and Edgar Villchur at Acoustic Research, and further developed by KLH (company).

Additional improvements to loudspeaker technology occurred in the 1970s, with the introduction of higher temperature adhesives, improved permanent magnet materials, and improved thermal management.

Dynamic loudspeakers

Cross-section of a dynamic cone loudspeaker. *Image not to scale.*

The traditional design is a semi-rigid paper fibre cone and a coil of fine wire (usually copper), called the voice coil attached to the apex of the cone. A "gap" is a small circular hole, slot or groove which allows the voice coil and cone to move back and forth. The coil is oriented coaxially inside the gap made with a permanent magnet. The gap is also where the magnetic field is concentrated. One magnetic pole is outside the coil, whilst the other is inside the voice coil. In addition to the magnet, voice coil, and cone, dynamic speakers usually also include a suspension system to provide lateral stability and make the speaker components return to a neutral point after moving. A typical suspension system includes the spider, which is at the apex of the cone, often of "concertina" form; and the surround, which is at the base of the cone. The parts are held together by a chassis or basket. When an electrical signal is applied, a magnetic field is induced by the electric current in the coil which becomes an electromagnet.

The coil and the permanent magnet interact with magnetic force which causes the coil and a semi-rigid cone (diaphragm) to vibrate and reproduce sound at the frequency of the applied electrical signal. When a multi-frequency signal is applied, the complex vibration results in reproduction of the applied signal as an audio signal.

Driver cones may be constructed of a variety of materials, including paper, metal, various polypropylenes, and kevlar. Baskets must be designed in order to preserve rigidity and are typically cast or stamped metal, although injection-molded plastic baskets are becoming much more common in recent years. The size and type of magnets can also differ. Sometimes, larger and more powerful magnets are associated with higher quality speakers. Tweeters are subject to a unique set of variables and parameters; their design and construction is extremely variable.

Despite marketing claims, lighter and more rigid cones do not always sound better. The weight and damping of the cone in a dynamic speaker should be appropriate for the characteristics of the rest of the driver and enclosure in order to produce accurate sound.

The cross-section image shows the construction of a speaker that uses an overhung coil. This term is used when the coil's height is greater than the gap's, and this construction is most commonly used. This construction attempts to keep the number of windings within the gap (and hence the force experienced by the coil) to be constant, as the coil moves back and forth. The other construction is the underhung construction.

Cut-away view of a dynamic loudspeaker. *The lead wires as shown are for illustration purposes. Commonly the voice coil wires are soldered to the lead wires and the soldered joints are glued to the diaphragm, close to the dust-cap periphery.*

Here, the voice coil's height is smaller than the gap's. This construction attempts to keep the magnetic flux constant across the coil as it moves back and forth (this also results in the coil experiencing a constant force).

Both methods try to achieve the same thing — a linear force on the coil, and each has its pros and cons. The overhung design offers softer or gradual non-linearity (compression) when the coil starts exceeding the X_max limits. Speakers using overhung coils have better efficiency and power handling. The underhung design offers greater linearity within the X_max limits but as the coil starts exceeding X_max limits the non-linearity (and hence distortion) rapidly increases. This design also requires a more powerful magnet, and so speakers using underhung coils tend to be less efficient [3]. X_max is the maximum linear peak-to-peak travel of the voice coil. This means that the displacement of the voice coil is a linear function of the input voltage within the limits specified by X_max.

Electrical characteristics of a dynamic loudspeaker

A dynamic loudspeaker presents a complex load to the amplifier as opposed to a pure resistance. It is a combination of resistive, capacitive, inductive as well as mechanical effects. A typical amplifier (amp) is most usually quoted for a given power into a resistive load. However a loudspeaker of a rated impedance of 8Ω/100W can easily overload an amp designed with a purely resistive load of 8Ω/100W as a target. [4]

Loudspeaker types

Multi driver systems

Home loudspeaker systems are generally multi-driver systems. 'Multi driver' refers to any speaker system that contains two or more separate drive units, including woofers, midranges, tweeters, and sometimes horns or supertweeters. In loudspeaker specifications, one often sees a speaker classified as an "N-way" speaker where N is a positive whole number greater than 1, indicating the number of separate frequency bands into which the system divides the sound (not the number of drivers, as one frequency band may be handled by more than one speaker driver). A 2-way system consists of woofer(s) and tweeter(s) sections; a 3-way system is constructed as a combination of woofer(s), tweeter(s) and mid-range speakers, etc. The frequency bands are separated and routed to the correct driver by an N-way crossover defined in the same manner, most usually a passive crossover within the speaker system, but in audiophile systems, sometimes an active crossover placed before the power amplifier stages.

Woofers

A woofer is a loudspeaker capable of reproducing the bass frequencies. The frequency range varies widely according to design. Whilst some woofers can cover the audio band from the bass to 3 kHz, others only work up to 1 kHz or less.

Mid-ranges

A mid-range loudspeaker, also known as a squawker is designed to cover the middle of the audio spectrum, typically from about 200 Hz to about 4-5 kHz. The distinction between woofers and mid-ranges is blurred however since many woofers can operate up to 3 kHz. These are used when the bass driver (or woofer) is incapable of covering the mid audio range. Mid-ranges typically appear where large (>16 cm or 8") woofers are used for the bass end of the audio spectrum.

Tweeters

A tweeter is a loudspeaker capable of reproducing the higher end of the audio spectrum, usually from about 1 kHz to 20 or 35 kHz.

Full-ranges

Main article: Full-range

A full-range speaker is designed to have as wide a frequency response as possible. These often employ an additional cone called a whizzer to extend the high frequency response and broaden the high frequency directivity. A whizzer is a small, light cone attached to the woofer's apex around the dust cap. The use of a whizzer requires that the main cone decouples from the coil at high frequencies such that most or all of the motion at those frequencies is imparted to the whizzer, which then acts like a second smaller coaxial loudspeaker. This gives many of the benefits of a tweeter without the additional expense or circuitry that is required. A whizzer might not be necessary if the diaphragm is small, stiff and light enough. There exist full-range drivers which are capable of reproducing a frequency range from 50 Hz to 20 kHz and higher without a whizzer cone. These drivers are often quite small, typically 2" to 5" (5 to 13 cm) in diameter.

Subwoofers

A subwoofer driver is a woofer optimised for the lowest range of the audio spectrum. Modern speaker systems often include a single speaker dedicated to reproducing the very lowest bass frequencies. This speaker (and its enclosure) is referred to as a subwoofer. A typical subwoofer only reproduces sounds below 120 Hz (although some subwoofers allow a choice of the cross-over frequency). Because the range of frequencies that must be reproduced is quite limited, the design of the subwoofer is usually quite simple, often consisting of a single, large, down-firing woofer enclosed in a cubical "bass-reflex" cabinet. Subwoofers often contain integrated power amplifiers that may incorporate sophisticated feedback mechanisms to ensure the least distortion of the reproduced bass acoustic waveform.

The very long wavelength of the very low frequency bass sounds reproduced by the subwoofer usually makes it impossible for the listener to localize the source of these sounds. Localization starts to happen above the 60Hz point. Because of this phenomenon, it is usually satisfactory to provide just a single subwoofer no matter how many individual channels are being used for the full-spectrum sound. For the same reason, the subwoofer does not need a special placement in the sound field (for example, centered between the Left Front and Right Front speakers). It can instead be hidden out of sight. Placing it in the corner of a room may produce louder bass sounds. A subwoofer's powerful bass can often cause items in the room or even the structure of the room itself to vibrate or buzz. Extended periods of high volume bass can cause items throughout a room to "walk" on a flat surface until they fall off.

Amplified subwoofers frequently accept both speaker-level and line-level audio signals. When teamed with a modern surround sound receiver and full range speakers, they are typically driven with the specific LFE (low frequency enhancement) output channel (the ".1" in 5.1, 6.1, or 7.1 specifications) provided by the receiver. This is because most full-range speakers are incapable of delivering the acoustic power required by the LFE in movies or in some cases, music. When used with speakers that do not reproduce low frequencies well, a subwoofer will often be configured to reproduce both the LFE channel and all other bass in the system, the latter being referred to as "bass management".

Enclosures

A loudspeaker is commonly mounted in an enclosure (or cabinet). The major role of the enclosure is to prevent the out-of-phase sound waves from the rear of the speaker combining with the positive phase sound waves from the front of the speaker, which would result in interference patterns, and cancellation of the low frequencies where the wavelengths are large enough that, in comparison, the front and rear of the speaker driver do not significantly differ in location. The "ideal" technique would be to mount the driver in a divider which extended to infinite distance in all directions to prevent the "back wave" from ever reaching the listener; because it's an attempt to simulate such behavior, the typical simple sealed box speaker system is known as an infinite baffle, although it is not actually infinite in size. More sophisticated designs attempt to improve on the performance of this design.

Connections and phase or polarity

Most loudspeakers have two wires that must be connected to the source of the signal (the audio amplifier or receiver). This is usually accomplished by five-way binding posts or spring wire clips provided on the back of the enclosure. Most professional speakers (except studio monitors) use 1 or more 1/4" phone plugs (in case of being bi-amplified), or a single Speakon connector, when the speaker enclosure is Bi or Tri-amplified. Some speakers have an input and an output to connect directly to another speaker.

Correct signal polarity (phase) must be observed. If both sets of wires for left and right (in a stereo setup) are not connected in phase, the speakers will be out of phase from each other and destructive interference in the sound waves will occur. In this case, any motion one cone makes will be opposite to the other. This type of wiring error creates inverse sound waves that partially cancel out the sound of the other speaker. This does not cause silence, because reflections from surfaces diminish the effect somewhat, but results in a major loss of sound quality. The most prominent effect to the untrained ear is a loss of bass response. The second most noticed is an unsettling feeling.

A similar effect is used in noise-cancelling headphones. The headphones produce the inverse sound waves of the external noise. The inverse sound waves and external noise cancel each other out and produce near silence.

Construction and testing

Speaker design is considered both an art and science. Adjusting a design is done with instruments and with the ear. Speaker designers use an anechoic chamber (essentially a room with soundproofing that inhibits any reverberation or echo) to ensure the speaker will perform the way it is intended to. Some developers (such as Bose) eschew the sole use of anechoic chambers in favor of specific standardized room set-ups designed to replicate likely real-life listening conditions. Some of the issues in speaker design are lobing, phase effects, off axis response and time coherence.

Care of speakers

Exceeding a Loudspeaker's limits by a large factor almost always causes permanent damage. The tweeters are usually the first to go under circumstances of abuse, since they have the lightest voice coil made of thin wire which easily melts if the temperature rises excessively. Tweeters are usually designed (and rated) keeping in mind that a typical music signal doesn't contain a lot of power or energy at the higher end of the audio spectrum. Thus a tweeter rated for 50 W is meant to be used with a 50 W amplifier only if the signals below the tweeter's lower operating frequency are filtered out. Thus, feeding a low frequency (or a DC) signal to a tweeter even though electrically it may be within the tweeter's specification may cause permanent damage to the tweeter. A badly clipping amplifier may also damage the tweeter despite a crossover, since a clipped waveform generates high-frequency harmonics which can contain sufficient power to heat up the tweeter's voice coil.

Most woofers (and mid-ranges) can easily take up to 1.5 times or more power than what their power rating states. However, this is dependent on the particular driver and the duration of the abuse or overload, as well on whether the very loud low-frequency sounds are being amplified. Another factor to take into consideration is whether or not the amplifer driving the speaker was being pushed to the point of harmful clipping, because a clipped waveform can damage speakers. Woofers will usually take a lot of power before burning out or suffering damage to their moving systems.

Physical damage occurs if the signal causes the woofer's cone displacement to exceed the safe X_mech limits for prolonged periods. In rare cases, a very loud signal may cause the coupling between the parts of the woofer to simply give way. A large DC fed to the woofer may cause twisting or deformation of the voice coil such that it rubs against the pole-pieces or magnet. Electrical damage occurs when the voice coil burns out. The latter two typically happen when the amplifier dumps a large DC current into the speaker - a condition typical of a failing (or failed) amplifier. In all cases, replacement or full repair of the driver are the only options.

Efficiency

Efficiency is classically defined as energy out / energy in, and in the case of loudspeakers it is defined as the sound power output divided by the electrical power input, while the voltage sensitivety is SPL / input voltage. Loudspeakers are actually very inefficient transducers. Only about 1% of the electrical energy put into a typical home loudspeaker is converted to the acoustic energy we know as sound - the remainder being converted to heat. The main reason for this low efficiency is the difficulty of achieving proper impedance matching between the acoustic impedance of the drive unit and that of the air. This is especially difficult at lower frequencies. The better the matching, the higher the efficiency. Large horn loudspeakers that used to be used in cinemas, were very efficient by today's hi-fi speaker standards.

Measuring the sound power output is not easy, but the sound pressure is relatively simple to measure. The sound pressure level (SPL) that a loudspeaker produces is measured in decibels (dBSPL). Ratings based on the sound pressure level are known as sensitivity ratings. The sensitivity is often defined as dB/(W·m)—decibels output for an input of one nominal watt measured at one metre from the loudspeaker and on-axis or directly in front of it, given that the loudspeaker is radiating into an infinitely large space and mounted on an infinite baffle. Sensitivity then does not correlate precisely with efficiency as it also depends on the directivity of the source and the acoustic environment in front of the speaker. As an example, a simple cheerleader's horn makes more sound output in the direction it is pointed than the cheerleader could by herself, but the horn did not improve or increase the cheerleader's total sound power output much, it mainly focused it over a smaller area.

Normal loudspeakers have a sensitivity of 85 to 95 dB/(W·m) - an efficiency of about 0.5-4%.
Nightclub speakers have a sensitivity of 95 to 102 dB/(W·m) - an efficiency of about 4-10%.
Rock concert, stadium speakers have a sensitivity of 103 to 110 dB/(W·m) - an efficiency of about 10-20%.

Current state-of-the-art loudspeakers can approach apparent efficiencies of 70% or higher under very special circumstances. This is partly due to a very high magnetic field and partly to a high amplitude displacement (speaker cone movement in and out). The ratio of the sound output to the mass of the cone/coil combination grows significantly at high sound pressure levels i.e. above 140 decibels. In closed or small environments (such as cars or bedrooms) it is far more important to have a speaker with a high X_max (cone eXcursion maximum) as opposed to high (dB/(W·m)) rating. A higher X_max indicates that the driver can move a larger volume of air as power increases. A few top of the line woofers have a very low "sensitivity" rating i.e. 80 to 86 dB/(W·m) (a nominal efficiency of 0.1%). However at full power in an enclosed automobile may achieve 160+ decibels at 20% to 40% apparent efficiency. In general a low frequency speaker designed for high SPL's will have a larger and or heavier magnet, and a higher X_max.

video of 158 dB woofer at 80 millimeter amplitude 450 kB mpeg As shown in this example, sometimes the speaker with the lower sensitivity rating has a far higher amount of acoustic output.

It should be noted that a higher power driver will not necessarily be louder than a lower power one. In the examples which follow, it is implicit that the drivers being compared have the same impedance. For the first example, a speaker that is 3 dB more sensitive than another will produce double the sound pressure (or play 3 dB louder) for the same power as the other. Thus a 100W speaker (call it A) rated at 92 dB/(W.m) sensitivity will be twice as loud as a 200W speaker (call it B) rated at 89 dB/(W.m) when both are driven with 100W of input power. For this particular example, when driven at 100W, speaker A will produce the same SPL or loudness that speaker B would produce with 200W input. Thus a 3 dB increase in sensitivity of the speaker means that it will need half the power to achieve a given SPL, and this translates into a smaller power amplifier and some cost savings.

Specifications

Speaker specifications generally include:

Speaker or driver type (Individual units only) – Full-range, woofer, tweeter or mid-range.
Rated Power – Nominal or continuous or RMS power and peak or maximum short-term power that the loudspeaker can handle (i.e. maximum allowed output power of the amplifier without destroying the loudspeaker. It is not the power that the passive loudspeaker produces).
Impedance – 4 Ω, 8 Ω, etc.
Baffle or enclosure type (Finished systems only) – Sealed, bass reflex, etc.
Number of drivers (Finished systems only) – 2-way, 3-way, etc.

and optionally,

Crossover frequency(ies) (Finished systems only) – The frequency or frequencies where electrical filtering occurs.
Frequency response – The measured or specified variance in sound pressure level over a range of frequencies.
Thiele/Small parameters (Individual units only) – These include the driver's F_s (resonance frequency), Q_ts (the driver's Q or damping factor at resonance), and V_as (the equivalent air compliance volume of the driver).

Interaction with listening environments

A complication is the interaction of the speaker with the listening environment. Most listening rooms present a more or less reflective environment which means that the sound that reaches a listener's ear consists not only of direct sound, but delayed sound that has been reflected off one or more boundaries. These reflected sound waves when added to the direct sound cause cancellation and addition at certain frequencies, changing the timbre and character of sound. This is part of the reason why a system sounds different at different listening positions or in different rooms. A significant factor in the sound of a loudspeaker system is the amount of absorption and diffusion present in the environment.

If one stands in an empty uncarpeted room and claps one's hands, the zippy flutter echoes one hears are due to a lack of both absorption and diffusion. The additions of hard furniture, wall hangings and shelving will substantially reduce the echoes, due primarily to diffusion. Diffusion is provided by somewhat reflective objects with shapes and textures having sizes on the order of sound wavelengths. This texture breaks up the stark reflections otherwise caused by flat walls and spreads the reflected energy of an incident wave over a larger angle.

Adding carpet, curtains or tapestries, and soft furniture will further lessen the echoes by absorbing the sound and preventing reflections. Absorptive materials absorb sound differently at different frequencies. The thinner a material is, the less likely it will have an effect at bass frequencies. The overabundance of absorption at high frequencies that can be caused by large areas of thin absorbing materials can cause a speaker system to sound deficient in treble, and likewise a lack of absorption can cause an otherwise uncolored loudspeaker to sound too bright or sibilant.

For good sound, a room should have a balance of diffusion and absorption. Most systems will sound best when the speakers are set up more or less symmetrically with respect to the listener and also to room boundaries. Early reflections do the most to color the sound (due to the so-called Haas Effect) so placing speakers too near either the rear or side walls is generally something to be avoided, although judicious use of absorbing or diffusing materials can somewhat moderate an otherwise poor placement location. Mounting a speaker in a wall (or in a bookshelf with books flush with the baffle) removes the reflective boundary concerns, but limits placement flexibility.

Another factor in room acoustics is the phenomena called standing waves. A one dimensional example is sound bouncing between two reflective boundaries. Sound will resonate or increase in intensity if the distance between the boundaries corresponds to an integral number of half wavelengths. Since sound travels at ~345 m/s, a pair of boundaries separated by 5 meters will cause a string of resonances to happen at 34.5 Hz, 69 Hz, 103.5 Hz.... Remember, wavelength is simply the speed of sound divided by the frequency.

In a typical rectangular listening room, this resonant phenomena is happening in three dimensions, and there are even more complex interactions that involve four or even all six boundaries. In addition, the placement of the loudspeakers and the listener with respect to the boundaries affect how strongly the resonances are excited or perceived. I am sure the reader is familiar with certain locations in a room or club which have dramatically more or less bass - most notably near room walls or corners. This is because standing wave patterns are most pronounced in these locations and in the bass frequencies, below the Schroeder frequency - which is typically around 200-300Hz, depending on room size.

Variations on the dynamic loudspeaker

Loudspeaker Directivity

In general, a sound source will consist of one of 4 forms, corresponding to [0..3] (zero to three) dimensional geometry. These are: point source, line source, planar source, 3D source. An infinitesimally small point source is often considered the ideal, because it radiates all frequencies equally in all directions in a spherical radiation pattern. Line sources may be finite or infinite, and they radiate sound in a cylindrical pattern. These first two sources are not practical, although real sound sources may approximate them at some frequencies. Several manufacturers have attempted to approximate a point source by approximating a pulsating sphere. In actuality, most sound sources (i.e. loudspeakers) are actually complex 3D shapes such as cones and domes, but most can be approximated (at least at low frequencies) as a planar form that creates a soundwave that becomes more directional as frequency increases, because the wavelength of the soundwave becomes small compared to the size of the diaphragm. That is, the intensity of the sound produced varies depending on the listener's angle relative to the central axis of the speaker. Specifically, the high frequency output decreases when the observer is located further off axis, and the off axis attenuation increases as the frequency is increased. So for high frequencies where the wavelength of sound is smaller than the diameter of the transducer, the size and shape of the radiating surface has a lot to do with the radiating pattern.

A common variation on the dynamic loudspeaker design uses a small dome as the moving part instead of an inverted cone. Perhaps contrary to intuition, making the moving component in the form of a dome rather than an inverted cone does not help to direct sound evenly in all directions. The dome is used because in the case of a tweeter the cone is smaller than the voicecoil and a conical shape is difficult to fit in because it would hit the polepiece. Some tweeters have inverted domes however where the polepiece is hollowed out to accomodate the dome. Some tweeters (TDL and other manufacturer use this shape) use bullet shaped domes instead of domes with constant radius. This is to eliminate the bending of centre of the dome. (if you think of an egg it is harder to push in the pointed side than the softer rounded side). Finally some manufacturers leave out the centre of the dome alltogether and only use the outer ring (ringradiator) because the inner part of the dome causes distortions due to the bending effekt (Scan-Speak, [[[Kea Audio]http://www.keaaudio.com]], Vifa etc). A ringradiator also has a better directivity (less directive). The typical one inch dome tweeter starts to become directive at about 8000 Hz, below this frequency it approximates a point source, above this frequency, it becomes increasingly directive off axis. At distances that are more than ~7 times the diameter of the cone or dome, the response is essentially that of a flat plane, at closer distances, the exact shape of the diaphragm becomes increasingly important.

2D and 3D sources can be monopolar or dipolar in nature. Most planar sources are dipolar, which means that the sound from the rear of the diaphragm is out of phase with the sound from the front. When the rear radiation is absorbed or trapped in a box, the diaphragm becomes a monopole radiator.

Bipolar speakers, made by mounting in-phase monopoles on opposite sides of a box, are essentially a method of approximating a point source or pulsating sphere. This design is essentially omnidirectional.

Various manufacturers employ geometry and the resulting radiation patterns to more closely simulate the way sound is produced by real instruments, or to mimic one of the ideal sound source types, or simply to create an energy distribution that mimics real soundfields.

Point Source

Manger transducer

The Manger bending wave transducer relies on the principle of bending waves, which start from the centre of the speaker and travel to the outside. The rigidity of the transducer material increases from the centre to the outside. High frequencies therefore come to an end in the inner area, while long waves reach the edge of the speaker. To prevent reflections, long waves are absorbed by a damper. The Manger transducer covers the frequency range from 80 Hz to 35,000 Hz, and is close to an ideal point sound source.

Coaxial Speakers

Coaxial speakers have been around since the 1930's. These approximate a point source by moving the radiating axes of the various drivers closer to the same point, with benefits in polar response. Coaxial mounting eliminates crossover lobing (interference between drivers caused by non coincident placement) by bringing the drivers radiating centers to the same point. The tweeter response tends to suffer somewhat and the woofer must act as a horn in most cases.

The technique of using concentric radiating elements for a multiway system has been used by several previous manufacturers, notably Technics. Cabasse recently published a paper showcasing the development of 3-way and even 4-way coaxial speakers using concentric ring-shaped radiators.

Several manufacturers (Tannoy, Eminence...) still build 2-way coaxials where the tweeter fires through a horn that passes through the woofer pole piece.

Several manufacturers (KEF, SEAS, Kea-Audio, Tannoy...) build coaxial units where the tweeter is mounted on the woofer pole piece. The small form factor was made possible by recent developments in rare earth magnets.

Omnidirectional

One school of thought is to approximate a pulsating sphere, below are some of the techniques used.

Direct/reflecting speakers

Amar Bose of Bose (also a professor at MIT) spent many years trying to reproduce this spherical wavefront by constructing a one-eighth sphere covered in small drivers that would be situated in the corner of a room, thus mimicking one-eighth of a spherical wavefront emanating from that corner; in practice this idea never became workable, but Bose's experience with combining multiple small drivers in one loudspeaker cabinet gave rise to the popular speakers which use multiple small drivers pointed in various directions to help create the balance of direct and reflected sound that Amar Bose determined is usual in a concert hall. The technique is actually somewhat controversial.

Conical bending wave transducers

The Ohm model "F" speakers invented by Lincoln Walsh feature a single driver mounted vertically as though it were firing downwards into the top of the cabinet, but instead of the normal almost flat cone, having a very-much extended cone entirely exposed at the top of the speaker.

The usual problem with designing a driver is how to keep the cone as stiff as possible (without adding mass), so that it moves as a unit and does not become subject to traveling waves on its surface. The Ohm drivers were designed so that the entire purpose of the electromagnetic driver was to generate traveling waves that traversed the cone from the electromagnet at the top downwards to the bottom. As the waves moved down the truncated cone, the effect was to reproduce the omnidirectional soundwave, as with a cylinder that changed diameter. This created a very effective omnidirectional radiator (although it suffered the same "planarity" effect as ribbon tweeters for higher-frequency sounds) and eliminated all problems of multiple drivers, such as crossover design, phase anomalies between drivers, etc. However, in practice it was found necessary to use a very complex cone made up of various materials at different points along its length, in order to maintain the waveform traveling evenly.

Line Sources

Ribbon speaker

The ribbon speaker consists of a thin metal-film ribbon suspended in a magnetic field. The electrical signal is applied to the ribbon which vibrates creating the sound. The advantage of the ribbon loudspeaker is that the ribbon has very little mass; as such, it can accelerate very quickly, yielding good high-frequency response. Ribbon loudspeakers can be very fragile, thin ones can be torn by a strong puff of air. Ribbon tweeters emit sound that exits the speaker in a roughly cylinder-shaped pattern. Above and below the ends of the ribbon there is often less treble sound, but the precise amount of directivity depends on the length of the ribbon. Ribbon designs generally require powerful magnets which make them costly to manufacture. Ribbons have a very low resistance that most amplifiers cannot drive directly. A step down transformer is therefore used to increase the current through the ribbon. The amplifier "sees" a load that is the ribbon resistance times the transformer turns ratio squared. The transformer must be carefully designed so that its frequency response and parasitic losses do not degrade the sound, further increasing cost relative to conventional designs.

Planar Magnetic speakers (having printed conductors on a diaphragm - see below) are sometimes described as "ribbons", and they are related, but are not truly ribbon speakers.

Planar designs

While most loudspeakers essentially have planar diaphragms, the term planar is generally reserved for speakers which have roughly rectangular shaped planar radiating surfaces. Below are some of the methods of making speakers having flat plane-shaped diaphragms.

Flat panel speakers

There have also been many attempts to reduce the size of loudspeakers, or alternatively to make them less obvious. One such attempt is the development of voice coils mounted to flat panels to act as sound sources. These can then be either made in a neutral colour and hung on walls where they will be less noticeable, or can be deliberately painted with patterns in which case they can function decoratively. There are two, related problems with flat panel technology; firstly, that the flat panel is more flexible than the cone shape and therefore fails to move as a solid unit, and secondly that resonances in the panels are difficult to control, leading to considerable distortion in the reproduced sound. Some progress has been made using such rigid yet damped material as styrofoam, and there have been several flat panel systems demonstrated in recent years.

Planar Magnetic

These consist of a flexible membrane with a voice coil printed on them. The current flowing through the coil interacts with the magnetic field of strategically-placed magnets, causing the membrane to vibrate. The driving force covers a larger percentage of the membrane and reduces the resonant problems inherent in coil-driven planar diaphragms. Many designs touted as "ribbons" are in fact planar magnetic designs. The long planar tweeters that are used in some designs have a small cavity in front of the diaphragm that is used to accommodate the magnets. This is not ideal and it sometimes creates a "cavity resonance" response peak that requires corrective filtration. Failure to correct this cavity resonance is sometimes the cause of the steely or shrill sound often attributed to these designs.

NXT

A newer implementation of the Flat Panel involves an intentionally flexible panel and an "exciter", mounted off-center in a location such that it excites the panel to vibrate. Speakers using NXT design methods can reproduce sound with a wide directivity pattern (paradoxically somewhat like a point source) with adequate quality.

Other technologies / without moving coils

Other technologies can be used to convert the electrical signal into an audio signal. These include piezoelectric, electrostatic, and plasma arc loudspeakers.

Piezoelectric speakers

Piezoelectric transducers are frequently used as beepers in watches etc., and are sometimes used as tweeters in less-expensive speaker systems, such as computer speakers and portable radios. Piezos have several advantages over conventional loudspeakers when applied to such purposes:

Piezoelectric transducers have no voice-coil, therefore there is no electrical inductance to overcome; it is easy to couple high-frequency electrical energy into the piezoelectric transducer, especially under the low-power, non-critical applications in which they are usually employed.
Piezoelectric transducers are resistant to overloads that would normally burn out the voice coil of a conventional loudspeaker.
Because piezos comprise a capacitive load, they usually do not require an external cross-over network; they can simply be placed in parallel with the inductive woofer/midrange loudspeaker(s).

There are also disadvantages:

Piezoelectric tweeters (in most cases) have a frequency response that is not as good as most other technologies. This is why they are generally used in single frequency (beeper) or noncritical applications.
Some amplifiers do not operate well into capacitive loads, causing high frequency oscillation, which can cause distortion or amplifier failure. Installing a series resistor can help prevent this.
unable to produce high enough excursions to play loudly at frequencies as low as conventional tweeters. Piezoelectric elements are sometimes seen coupled to larger cone shaped radiating diaphragms in some applications to increase their output.

Plasma arc loudspeakers

The most exotic speaker design is undoubtedly the plasma arc loudspeaker, using electrical plasma as a driver [5], once commercially sold as the Ionovac [6]. Since plasma has minimal mass, but is charged and therefore can be manipulated by an electric field, the result is a very linear output at frequencies far higher than the audible range. As might be guessed, problems of maintenance and reliability for this design tend to make it very unsuitable for the mass market; the plasma is generated from a tank of helium which must be periodically refilled, for instance. A lower-priced variation on this theme is the use of a flame for the driver [7], flames being commonly electrically charged. Unfortunately, the recent marketing of plasma displays as high-end television sets and computer monitors has caused the "me-too" labeling of many speakers as "plasma" which have nothing whatsoever to do with plasma [8], much as the advent of digital audio caused the marketing of a large number of "digital" headphones and speakers whose drive-units were analog in nature.

Digital speakers

Actual digital speaker driver technology not only exists, but is quite mature, having been experimented with extensively by Bell Labs as far back as the 1920s. The design of these is disarmingly simple; the least significant bit drives a tiny speaker driver, of whatever physical design seems appropriate; a value of "1" causes this driver to be driven full amplitude, a value of "0" causes it to be completely shut off. (This allows for high efficiency in the amplifier, which at any time is either passing zero current, or required to drop the voltage by zero volts, therefore theoretically dissipating zero watts at all times). The next least significant bit drives a speaker of twice the area (most often, but not necessarily, a ring around the previous driver), again to either full amplitude, or off. The next least significant bit drives a speaker of twice this area, and so on.

There are two problems with this design which led to its being abandoned as hopelessly impractical, however; firstly, a quick calculation shows that for a reasonable number of bits required for reasonable sound reproduction quality, the size of the system becomes very large. For example, a 16 bit system to be compatible with the 16 bit audio CD standard, starting with a reasonable 2 square inch (13 cm²) driver for the least significant bit, would require a total area for the drivers of over 900 square feet (85 m²). Secondly, since this system is converting digital signal to analog, the effect of aliasing is unavoidable, so that the audio output is "reflected" at equal amplitude in the frequency domain, on the other side of the sampling frequency. Even accounting for the vastly lower efficiency of speaker drivers at such high frequencies, the result was to generate an unacceptably high level of ultrasonics accompanying the desired output. In electronic digital to analog conversion, this is addressed by the use of low-pass filters to eliminate the spurious upper frequencies produced; however, this approach cannot be used to solve the problem with this digital loudspeaker, since it is the last link in the audio chain.

Electrostatic loudspeakers (ESL)

Some speakers are electrostatically driven rather than via the usual electromechanical voice coil, thereby giving a more linear response. Today, modern materials and insulative coatings have allowed engineers to design electrostatic speakers that are safe and reliable, but this wasn't always the case. For many years electrotstatic loudspeakers had a reputation as a dangerous and unreliable product; the disadvantage was that the signal must be converted to a very high voltage and low current, which was problematic for reliability and maintenance as they attracted dust, and developed a tendency to arc, particularly where the dust provided a partial path; the point where the arc occurs often became more prone to arcing, as carbon built up from the burned dust.

Electrostatic loudspeakers are large by nature. The sole objective of loudspeakers is to move the air analogue to the electric signal applied to them. The amount of air that can be moved is determened by the size of the membrane and the allowed excursion. With electrostatic loudspeakers the excursion is limited to millimeters as where dynamic loudspeakers can move centimeters. This means that the membrane of an electrostatic loudspeaker has to be larger than an equivalent dynamic loudspeaker.

Converting ultrasound to audible sound

A transducer can be made to project a narrow beam of ultrasound that is powerful enough, (100 to 110 dBSPL) to change the speed of sound in the air that it passes through. The ultrasound is modulated-- it consists of an audible signal mixed with an ultrasonic frequency. The air within the beam behaves in a nonlinear way and demodulates the ultrasound, resulting in sound that is audible only along the path of the beam, or that appears to radiate from any surface that the beam strikes. The practical effect of this technology is that a beam of sound can be projected over a long distance to be heard only in a small, well-defined area. A listener outside the beam hears nothing. This effect cannot be achieved with conventional loudspeakers, because sound at audible frequencies cannot be focused into such a narrow beam.

There are some criticisms of this approach. Anyone or anything that disrupts the path of the beam will disturb the dispersion of the signal, and there are limitations, both to the frequency response and to the dispersion pattern of such devices.

This technology was originally developed by the US (and Russian) Navy for underwater sonar in the mid-1960s, and was briefly investigated by Japanese researchers in the early 1980s, but these efforts were abandoned due to extremely poor sound quality (high distortion) and substantial system cost. These problems went unsolved until a paper published by Dr. F. Joseph Pompei of the Massachusetts Institute of Technology in 1998 (105th AES Conv, Preprint 4853, 1998) fully described a working device that reduced audible distortion essentially to that of a traditional loudspeaker.

The technology, termed the Audio Spotlight, was first made commercially available in 2000 by Holosonics, a company founded by Dr. Pompei.

There are currently two devices available on the market that use ultrasound to create an audible "beam" of sound: the Audio Spotlight and Hypersonic Sound.

External links

Cable Nonsense -This newsgroup message from a speaker manufacturer is very informative about issues of speaker cables.
The Audio Circuit - An almost complete list of manufacturers of loudspeakers.
Article on Tesla's contribution