We begin our journey circa 1957, the dawn of the stereo age, with the introduction of the first full-range electrostatic loudspeaker (later known as the ESL-57) by Quad Electroacoustics. Although the principles of electrostatic transducers were already well established, it was the lack of suitable diaphragm materials that constrained its commercial development. The advent of Polyethylene Terephalate (PET) films made the Quad’s nearly massless moving diaphragm possible. (Nowadays almost everyone is familiar with such films as Mylar, DuPont’s original trade name for this material.)
Peter Walker’s pristine “window” onto the orchestra did justice to Quad’s motto: “The Closest Approach to the Original Sound.” And it quite remarkably remained in production for 28 years. I’m not sure whether it was a question of Quad having set the bar too high in the midrange or the industry’s infatuation with the frequency extremes that focused development effort for the next 50 years on woofers and tweeters. Certainly, from a marketing standpoint, the new notion of high-fidelity sound argued for extended frequency response.
As with the Quad ESL, massless tweeters also became a commercial reality in the late 1950s. An improved version of Dr. Sigfried Klein’s ionic tweeter was commercialized in 1956 by Dukane in the U.S. as the Ionovac. An RF oscillator ionizes a small air mass, creating a plasma, which can then be modulated by the audio signal transferring vibrational energy directly to the surrounding air. No moving parts, no breakup resonances, and an extended bandwidth—truly the ingredients for a superb (but costly) tweeter. Another approach to creating an ionic tweeter is represented by the corona wind transducer. Here a high voltage applied to needle-like electrodes creates a corona discharge, ionizing nearby air molecules. I even recall a couple of “full-range” corona wind speaker prototypes that made it to trade shows, but neither of these was released commercially. Probably the most promising was Nelson Pass’ Ionic Cloud, a prolific ozone producer that nearly did Nelson in. Efficiency of ionic tweeters is generally low and they are usually coupled to a horn for that reason. An exception to the rule, and my favorite plasma tweeter with a reach down to around 1kHz, was Dr. Alan Hill’s Plasmatronics helium-plasma transducer. Helium gas was used to minimize ozone production. I should know, being a former owner (in the mid-80s) and having wrestled lab-size helium bottles across the house to feed the Hill Type 1 loudspeaker. Plasma tweeters are still a commercial reality today and are incorporated in systems offered by two German firms, Lansche Audio and horn-specialist Acapella.
Another tweeter well ahead of its time was the Decca ribbon. Similar in conception to a ribbon microphone, the diaphragm is a thin (think fragile) aluminum ribbon centered between the poles of a magnet. The signal is typically fed to the ribbon element through a transformer in order to maintain decent impedance at the amplifier output terminals. The Decca used horn-loading to improve efficiency.
As you can see, all basic transducer mechanisms were well known by the end of the 1950s, and since the laws of physics have not changed in the interim, advances in the art since then have been mainly technological in nature, aided by a few conceptual breakthroughs. Many of these advances were materials-based. Nowhere is that more evident than in the evolution of the dome tweeter. The 1” soft fabric or plastic dome has adorned the front baffle of countless two and three-way box speakers over the years. It isn’t particularly fast or resolving, but its one saving grace is a smooth response, as it breaks up in a controlled manner without significant peaking. The ideal dome should have low mass for good efficiency and sufficient stiffness to push break-up resonances well above the audible bandwidth. Early light-metal domes such as aluminum and titanium had much faster sound velocities (i.e., excellent transient speed) relative to plastic, but audible resonances were a spoiler. Two fairly recent high-end tweeter developments aimed at scaling stiffness to mass utilize magnesium and exotic beryllium. Significant research and development were also expended on developing ceramic dome and cone materials. The payoff is excellent impulse response with internal damping to prevent ringing. Molding sintered ceramics to sufficiently thin profiles to reduce mass and maintain good efficiency is a challenge. Accuton, an industry leader in this area, produces a very thin concave aluminum oxide diaphragm in a sapphire matrix. Accuton has outdone itself recently with the introduction of the world’s first true diamond diaphragm. About five times harder than Accuton’s own ceramic membrane and an exceptional heat conductor, it demonstrates that diamonds are not just a girl’s best friend.
A similar story can be told when it comes to woofer and midrange cones. In the beginning there was only paper. Over the years various plastic formulations have been used including polypropylene impregnated with additives for improved stiffness, Kevlar, and even carbon-fiber-loaded plastic. Some manufacturers have favored a sandwich cone construction with an inner layer of foam and outer “skin” layers. Recently, Magico has developed the Nano-Tec line of drivers, which feature carbon nanotube outer skin layers and a Rohacell structural foam core. Light metal cones (beryllium, aluminum, and magnesium) continue to have their fans. Nonetheless paper, with a variety of additives and treatments, remains popular on the basis of performance and cost. Moving-coil drivers have also benefitted from improvements in suspension materials, voice-coil wire (e.g., copper-clad aluminum), and magnets (in particular high-strength neodymium-iron-boron magnets, now available at reasonable prices from China). However, due to their substantial loss of magnetic field strength at high temperatures, you are more likely to find neodymium magnets in tweeters than in woofers.
Box speakers in the early 50s tended to be large and efficient, since domestic power amplification was typically in the range of 10 to 15Wpc. Infinite baffles, bass-reflex boxes, and folded rear horns were quite popular. Design theory was quite crude. It was understood that large woofers belonged in large boxes but detailed design models that would correctly combine drivers and enclosures into a system with a predictable low-frequency cut-off were yet to be developed. An enclosure had to be tweaked experimentally to fine-tune it for a particular driver. It was therefore quite a surprise when Edgar Villchur introduced the world to the Acoustic Research AR-1—the first so called acoustic-suspension design. In hindsight, the idea of fitting a high-compliance, low-free-air-resonance woofer into a smallish closed box can be seen as one of many possible closed-box alignments predictable using Thiele-Small (TS) parameter-based models. In the 50s, however, obtaining good bass from a small box seemed like magic. No doubt aided by domestic constraints (i.e., the wife acceptance factor), Acoustic Research almost single-handedly created the genre of the bookshelf speaker. Aided by more powerful solid-state amplification, the bookshelf speaker became firmly established, at least sales-wise, as the dominant domestic speaker type in the 70s.
The task of demystifying what must have seemed like the “black art” of speaker design fell to Neville Thiele and Richard Small. Small, who moved Down Under in 1964, had gotten a part-time teaching position at the University of Sydney. His interest in speaker design was kindled when he landed a small speaker design consulting job. Frustrated by lack of any definitive design information in the American technical literature he was eventually led to Thiele’s paper on loudspeakers in vented boxes published in the Proceedings of the IRE Australia in 1961. The genius of Thiele’s work lay in the application of electrical filter theory to the design problem using a few easy to measure driver parameters. Small expanded on Thiele’s original work (while also acknowledging the contributions of the late Dr. J. Ernest Benson) to form a set of models for the design of low-frequency closed and vented box enclosures. He is fond of saying that loudspeakers don’t have parameters. By that he means that only models have parameters, and that such models are not necessarily definitive in their predictions. But the TS-parameter-based models have proven to be quite successful in predicting low-frequency response, at least for small-signal analysis, within 1dB. With the advent of the PC, these design equations have been imported into software that makes it a snap to investigate the performance of a particular woofer with a given set of TS parameters in a specific enclosure and to optimize its performance. In general, it would be fair to say that computer software has had a significant impact on driver design. For example, computer-aided analysis of magnet structures using finite-element techniques has enabled designers to eliminate significant chunks of iron mass that did not carry magnetic flux from the magnet to the voice-coil gap.
Of equal importance to the TS models was the technical leap in crossover-network synthesis for multi-way speaker systems. This was progress without much fanfare. It seems that prior to the work of Siegfried Linkwitz in the mid-70s crossover filters were being designed and debated without any consideration of driver placement on the front baffle or of phase response. Frequency-dividing networks were being designed strictly along electrical lines under the simplifying assumptions that drivers are nothing more than resistors and that their acoustic centers are spatially coincident. Linkwitz investigated the effect of driver spacing on the speaker’s radiation pattern. He showed that when the spacing approaches a significant fraction of the wavelength, as it would between a tweeter and woofer spaced several inches apart and crossed over at around 3kHz, the end result is interference in the overlap region. As a consequence the main lobe of the radiation pattern could shift in direction and increase in amplitude. As far as the listener is concerned, this is far from an ideal situation, as head movements of only a few inches vertically would generate significant changes in tonal balance. Linkwitz recommended crossover networks based on cascaded Butterworth sections as giving the most stable radiation pattern within the driver overlap region. Today, such networks are known as Linkwitz-Riley types. He also recommended aligning the woofer and tweeter vertically on the front baffle as well as tilting the baffle backwards to align the acoustic centers. The 80s were fertile ground for sophisticated driver-integration techniques. One approach that I have personally had success with is due to Bernd Hillerich and uses asymmetric crossovers of different order to compensate for time offsets between drivers. For example, a fourth-order network for the tweeter coupled with a second-order network for the woofer. Elegant and far more familiar to audiophiles is Joe D’Appolito’s driver configuration, a tweeter flanked vertically by a pair of woofers. The idea is to improve the uniformity of sound dispersion at the cost of an added driver. Software (e.g., CALSOD) became available for driver integration that specifically accounts for various real-world effects such as the drivers’ actual frequency and phase response and gave speaker design a needed boost.
I think that we all owe Magnepan’s Jim Winey a big thank you, not only for inventing the magnetic planar speaker in 1969, but also for popularizing the concept and weaning so many music lovers off box speakers. The planar’s operating principle of “force over area” (as opposed to a driving force applied to the tip of a cone) became an audiophile catchphrase. Since production began in 1971, over 200,000 speakers have been manufactured—a remarkable success story in the high-end arena. Jim says that the concept came to him in a flash while he was staring one day at an air-conditioner grille. While the concept appears simple in principle, translating it into a real-world product was critically dependent on plastic film and magnet technology. As with electrostatics, Mylar came to the rescue as a diaphragm material, to which is bonded either thin wire or conductive strips. The latter arrangement has been dubbed by Magnepan as a “quasi-ribbon.” Much effort was spent on achieving an adequate magnetic flux density in the gap between the permanent magnet array and the diaphragm using a variety of magnet types and configurations. A true line-source ribbon is used in Magneplanar’s upper-end models (MG 3.6 and MG 20.1), which due to its length achieves a reasonable impedance without need for an impedance-matching transformer.
Another influential ribbon/planar speaker manufacturer was Apogee Acoustics, from its inception in 1982 up to its demise in 1998. Designer Leo Spiegel’s full-range three-way ribbon speaker was launched in 1983 and made quite a stir with its astounding dynamics and bass extension. The bass driver was in fact something new and wonderful. It started life as a Kapton-backed sheet of corrugated aluminum foil, which was attached to a trapezoidal frame for resonance control, and then cut into individual ribbons. An array of permanent bar magnets positioned behind the ribbons provided the magnetic driving force. The affordable Apogee Stage was a best seller and featured a conductive coating on both sides of the woofer and tweeter diaphragms.
High-end audio was alive and well in the 1980s and electrostatic fever showed no signs of abating. Some of the important players were Acoustat, Audiostatics, Beveridge, Dayton-Wright, MartinLogan, and Sound Lab. Dispersion and dynamics were two performance areas most often addressed by competing designs. I have fond memories of the Beveridge Acoustic Lens, which converted the output from a flat electrostatic panel to a cylindrical wave front. The innovative Dayton-Wright XG-8 Mk 3 encased the electrostatic cells in sulfur hexafluoride, an insulating gas, to increase the break-down voltage and allow much higher bias voltages than a standard electrostatic. The payoff was much improved efficiency and higher sound pressure levels. They said it that couldn’t be done, but MartinLogan’s curvilinear line-source (CLS), aimed at improving lateral dispersion, has been a signature element of every MartinLogan ESL. Although electrostatic-cone speaker hybrids were not conceptually new in the early 80s, MartinLogan’s knack for optimizing hybrid performance and for cost-effective packaging put such designs on the map. Sound Lab’s full-range electrostatic speakers such as the A-1, which kept me company for many years, also achieved a curved profile for enhanced dispersion by building up a mosaic of small flat cells. In an interesting twist, Final Sound’s recent ESLs invert the function of the transducer elements. The audio signal is applied to a thin conductive diaphragm while the DC polarizing voltage is applied to the stators. One of the benefits of this technology is enhanced safety as no high-voltage AC signal is applied to the stator panels. Another is that the required input transformer size decreases because the stepped-up audio signal can now be applied to just the diaphragm in single-ended fashion.
The paradigm of “pinpoint imaging” was born in the late 70s and early 80s as a result of exposure to a number of British invaders, like the BBC-spec Rogers LS3/5A and Celestion’s SL6 and SL600 mini-monitors, and the Spica TC50 stateside. Dynamics and bass-extension issues aside, these mini-monitors could really image effortlessly throwing a well-delineated soundstage populated by precisely localized and tightly focused image outlines. These are the sort of perceptual attributes that cannot be measured but once heard are hard to forget. In particular, Celestion took driver design very seriously, deploying laser interferometer techniques to study cone behavior under dynamic conditions. The SL6 was the first product to benefit from this technology. And in 1983 the SL600 was launched featuring an aluminum honeycomb cabinet with a stiffness-to-weight ratio far in excess of what average MDF cabinets are capable of. The result was almost no audible cabinet signature. It has been known forever that speaker cabinets flex and vibrate, potentially producing significant sound output, typically in the upper bass, and with a sufficiently long time constant capable of muddying bass lines. The ongoing technical debate has been over how to address this problem. One school of thought, as embodied by the BBC thin-wall cabinet design, is to allow the cabinet to flex but keep the overall mass to a minimum, to minimize energy storage, so that the vibrational energy dissipates quickly. The other approach is to build a thick-walled, well-braced enclosure with the idea of minimizing bending modes and hence vibrational energy. Better yet is the approach of using composite materials, consisting of several layers (possibly including metals) for added stiffness and superior internal damping. The Swiss firm Piega makes all of its cabinets out of extruded aluminum.
No survey of the loudspeaker art would be complete without at least a mention of three unique transducers. Lincoln Walsh’s inverted cone speaker first described in 1965 was the first bending-wave transducer. It offered a wide frequency response and an omnidirectional radiation pattern—at least in the horizontal plane. A recent version is German Physiks’ Dick Dipole Driver, Peter Dick’s variation on the Lincoln Walsh design. Then there is the futuristic looking MBL Radialstrahler which functions much like a pulsating sphere. This driver is designed to be an omnidirectional radiator in the horizontal plane. Finally, the genius of Oskar Heil’s Air Motion Transformer (AMT) lies in a diaphragm which is folded accordion-style, and onto which is bonded a conductive layer. Being immersed in a strong magnetic field, the diaphragm is alternatively squeezed and expanded in tune with the music signal. Air is expelled at a velocity several times (typically four to five times) greater than that of the diaphragm folds. Hence, the air motion or velocity-transformation effect. And because the diaphragm is folded, its effective acoustic area is smaller than the actual radiating area making it appear more like a point source relative to the listener’s position.