All design is an art and it will remain so irrespective of the amount of computing power available. Whilst computers are very good at predicting what will result from certain choices, they are hopeless at making those choices. No word processor ever wrote a book, instead it streamlined the process for those who could write and made it possible for those who could not write to have it more widely known. Computer aided design is little more than word processing for engineering drawings. It does not design anything, but instead streamlines the process for those who can already design. I would not be without CAD/CAM, but the designs are down to me.

Before a computer can do anything, there has to be a reasonably accurate model of the process that is to be computerised. In accounting, for example, that is relatively straightforward. In design, no such model exists, nor is there any universal agreement about what it might look like. All that can be said for certain is that there isn’t a unique rational series of steps that result in a good design. There isn’t an algorithm for creativity.

Although computer technology has transformed many processes, in many cases for the better, it has had little or no impact at all on the quality of design. Poor design is all around us and to appreciate the problem it is necessary to have some understanding of statistics. There is a distribution of design quality, which might be Gaussian or nearly so, ranging from good to down right squalid. The centre of that distribution is located at less than good.

Bad Speaker Design Is Seldom Lethal

Design cannot be measured by a single number; there are many factors to be considered. One of them is suitability for purpose; another is acceptable aesthetics, almost always subject to economic pressures. Suitability for purpose is an interesting topic and invariably the more critical the application the more educated the purchaser will be and the more stringent will be the criteria. Anything to do with life support, like aviation or civil engineering, has to be highly suitable for purpose and rigorously tested. Failure can cost lives. Where such stringency is absent, where the end user has little or no understanding of the technology, mediocrity creeps in unopposed.

Design is a Cinderella subject, seldom studied, seldom understood and whose tenets are often violated. But wait a minute, acoustics also is a Cinderella subject, seldom studied, seldom understood and whose tenets are often violated. That’s a powerful combination, but we’re not done yet. To add insult to injury, loudspeakers are not really in the life support category. Deaths involving loudspeakers are confined to where their supports fail. Finally most loudspeaker purchasers know nothing about them or how to test them.

In most cases, a product has a specification and then some kind of test is devised to see if the specification is met. In analogue television we used to have the pulse and bar test signal. In television, cinema and photography alike we have resolution test charts and ways of testing and comparing the MTFs of our optics with agreed units in order to establish the quality of our images. Although loudspeakers are intended to reproduce acoustic images, can anyone reading this tell me the unit of acoustic imaging accuracy? What is the standard test procedure?

No, of course you can’t; they don’t exist. Every other imaging technology but audio has them. Without measuring ability, meaningful comparisons cannot be made and better solutions cannot come into use. As a result, loudspeaker design has practically stood still for decades. 

Given those multiple fundamental vectors towards mediocrity, we should not be in the least surprised to find that the overwhelming majority of loudspeakers are so bad that it is questionable whether the term design is justified.

In the same way that the human visual system ultimately specifies all imaging systems, the loudspeaker can only be specified by the Human Auditory System (HAS). Knowledge of the HAS has increased spectacularly in recent years, yet the legacy loudspeaker soldiers on, pretending that these discoveries were never made and violating them. We stopped using asbestos because it was found to be carcinogenic. Many common loudspeaker technologies are now known to result in the death of the input waveform, but they continue in use.

Part of the problem is that the design of loudspeakers has been stultified and commoditised for so long that they all look exactly the same and they all perform exactly the same. The consumer has learned what a loudspeaker looks like and sounds like and has no idea that it is not fundamental. Worse than that, anything different would be automatically suspect because it doesn’t look like a loudspeaker. We note that the new electric car from Jaguar has a conspicuous radiator grille, so it looks like a car. The heat rejected by the generation of energy is dissipated not in the car, but in a power station, but in the absence of that grille, people might think it was something else. 

Whilst the artist can look directly at a painting and assess the quality, the videographer and the audio engineer can only perceive what they have wrought via a transducer. In television production, the need for accurate picture monitors is well understood, but seemingly the need for accurate audio monitors is not. Thanks to their speakers, most audio engineers literally cannot hear what they are doing and as a result many of their decisions are sub-optimal, including the acceptance of poor compression algorithms and poor microphone techniques.

Traditional Loudspeaker Design

It may be illuminating to look at a legacy loudspeaker in order to see all of the features it contains that violate modern knowledge of the human auditory system and serve to make it sound like a loudspeaker. Figure 1 illustrates a typical example of a two-way bookshelf speaker. 

Almost universally, there will be a dome tweeter (3), which looks as though it should radiate sound in all directions but actually doesn’t. The radiating diameter of an ideal loudspeaker diaphragm becomes smaller as frequency rises so that the sound does not become too directional. The dome tweeter drives its diaphragm from the perimeter, which is exactly the wrong approach. As frequency rises, the centre of the dome decouples and the device becomes an annular radiator, which is sub-optimal. In order to palliate the directivity problem, the dome is kept small, which means that it cannot operate down to low frequencies. This forces the crossover frequency to be too high and makes the woofer operate at frequencies where it will become directional. The directivity function of the speaker is all over the place and the off-axis frequency response (2) looks like a dog’s hind leg, with the result that the reverberant sound in the room is coloured.

Not visible in the illustration is the passive crossover (7) that directs the appropriate part of the spectrum to the two drivers. The ideal crossover would produce a pair of signals that add up to the original waveform. Yet no passive crossover can do that. The audio waveform is violated in the region of the crossover. The failing crossover combined with the doubtful directivity results in these speakers sounding as if the audio spectrum had been arbitrarily cut in half and each half had been dealt with by different designers who never met. Sound sources having a broad spectrum that extends across both drivers, such as the cello, sound particularly unrealistic.

The traditional woofer (4) has a small coil in the centre of the diaphragm. This causes a stress concentration and the diaphragm flexes and doesn’t follow the coil under transient conditions.

The typical legacy loudspeaker contains a reflex port (5). This is intended to boost the low frequency response using resonance. Unfortunately it does so at the expense of time accuracy. The time accuracy is so bad that the acoustic source, the point from which the sound appears to be radiated, moves back by several metres at frequencies where the resonance is invoked. Transients are impaired because the resonance takes time to build up. Equally it takes time to decay and the result is hangover. Whilst amplifier designers proudly show the square wave response of their products, and rightly so, the square wave response of legacy speakers is seldom published because it is unrecognisable. Any resonances in the original sound by which we might recognise the source are replaced by the characteristic footprint of the speaker, so that everything sounds the same and listening fatigue is not far away.

The human auditory system relies on timing for source location and to make things like the Haas effect work. Most of the information is sound is in transients. To put it another way, the bandwidth of a sine wave is zero and information theory tells us it carries no information. Yet legacy loudspeakers are optimised to reproduce sine waves to the detriment of transient information.

What do aerosol cans, champagne bottles, submarines and airliners have in common? Simply that they are all designed to resist pressure and have an appropriate cylindrical shape. The legacy loudspeaker enclosure shown in Figure 1 needs to resist pressure, but it has an inappropriate shape. The flat panels (6) are adopted because they allow any carpenter to cobble a speaker together from six pieces of wood. The fact that a flat panel is the weakest structure known, the one most prone to resonance, is neither here nor there. Whereas the enclosure should be inert, the design of Figure 1 displays a characteristic response like that of an empty furniture truck hitting a pot hole.

From a stereophonic imaging standpoint, the square loudspeaker enclosure (1) also lets us down, because the acoustic impedance change at the sharp corners causes sound from the tweeter to be re-radiated. The re-radiation interferes with the direct sound to make the directivity pattern periodic. From an imaging standpoint, the sound source is the width of the speaker instead of a point. This aperture effect destroys the imaging ability as no virtual sound source can be narrower than a speaker.

In a sense loudspeaker design has now become quite easy, because knowledge of the HAS can tell us that a lot of traditional approaches cannot be used and why. The number of remaining acceptable approaches must be smaller and the choice is simpler.

Unfortunately, many of today’s speaker designs result in audio that sounds like a loudspeaker and not like the original sound. 

Credits

Mr. Watkinson is author of more than 20 books on audio and video technology and television transmission systems with a recent book on helicopters. His works are available from major booksellers.

Editor note

John Watkinson's entire series on loudspeaker technology can be located from The Broadcast Bridge home page. Search for "John Watkinson. His other tutorials on audio and video will also be listed.