Who among us hasn’t thought about what matters most in an audio system? The subject often comes up when the total cost of the high-end audio components we’re buying to assemble a system must fit within a certain budget. For example: If you believe that garbage in results in garbage out, then what will most matter to you will be the source component—a turntable, a DAC, a CD player—which means you’ll be willing to spend relatively more on that component. Many audiophiles find that the most significant audible differences among high-end brands tend to be between speaker manufacturers; consequently, those audiophiles are likely to spend the lion’s share of their budgets on getting the best-sounding speakers they can afford. Still others will tell you that the magic resides in the preamplifier—and many of those folks will swear that, all else being equal, what makes the biggest improvement in sound is a tubed preamp. Coming up with an answer to the question of which component matters most, and then acting on that answer, is part of the fun of being an audiophile.
The caveat, of course, is that audiophiles listen not to any individual component but to systems, each of which comprises multiple components. An otherwise great-sounding pair of speakers powered by an amplifier incapable of driving them will sound poor at best. And if your amp can drive your wonderful new speakers but your room is an acoustical mess, same goes. Many more examples abound; when the goal is great sound, the most pressing considerations are often the synergies among components, and between the speakers and the room.
All of us know—or should know—that synergies are critical. For the sake of this discussion, though, if you had to assign the title of Most Important to a single category of audio component, which would it be?
I’d pick the loudspeaker. But for the fun of it, let’s make the question a bit more complex. Instead of limiting the choices to complete, standalone products, let’s widen—or narrow—the discussion, and instead limit our choices to the parts used within a single product: its tubes, transformers, capacitors, drivers, enclosures, etc.—anything that might be a part of the finished component.
Here’s what I chose, and why I chose it: [drum roll, please]
What most determines the sound heard from an audio system is the loudspeaker drive unit.
The loudspeaker drive unit has the hardest job of any component in an audio system: to transduce the electrical signal that flows from the source to the amplifier, through wire and crossover components, into soundwaves that strike the human eardrum. We don’t hear electrical signals—our brains transduce into the mental experience of sound the impact on our eardrums of the soundwaves to which the loudspeaker drive units have converted those signals. I contend that the transduction of electrical energy to acoustical energy has more effect on what you hear than does any other facet of audio reproduction, precisely because that transduction is so fraught with possibilities for error.
I wanted to further explore the loudspeaker drive unit and why its job is so hard, but technically the subject is beyond my pay grade. So I consulted one of the world’s preeminent experts: Laurence Dickie, Technical Director of Vivid Audio. A detailed history of the esteemed Dickie’s career is covered in many places (one is our SoundStage! YouTube channel). Dickie designs the drivers for the Vivid Audio speakers he also designs; his work there, and previously at Bowers & Wilkins (1983-1997), have established his prominent position in the hierarchy of great speaker designers. Here are my questions and his responses:
Jeff Fritz: When a loudspeaker drive unit transduces an electrical signal into soundwaves, what are the sorts of things that can go wrong?
Laurence Dickie: I think we all share the understanding that the ideal speaker driver would convert the electrical voltage signal delivered by the amplifier into an exact air-pressure analog. Sadly, the reality never quite hits the target, and even the first process, the transduction between voltage and force, is subject to a host of nonlinearities that will introduce frequencies not present in the original signal, some inoffensive and even quite consonant, but others dissonant and harsh. Those forces will then cause motion in a diaphragm that will behave differently with changing frequency, and thus emphasize some at the expense of others. And while we would hope that the ideal diaphragm would stop moving when the signal stops, the reality is that some frequencies, once set into motion, will take longer to die away than others. The moving cone creates a soundfield in the air surrounding it, but unfortunately the path from there to free space is still filled with obstacles that may each contribute another blemish to the sought-after flat response.
JF: Specifically, what sort of distortions are caused by the driver’s motor system, as opposed to the diaphragm—the cone or dome?
LD: As the drive voltage first appears across the terminals and begins to push a current around the voice coil, so the magnetic field in which it lies begins to distort. It is this distortion that actually creates the force on the conductor, but it also induces changes in the magnetizing force in the magnet system that surrounds it. The way in which the magnetic field responds to the current is reflected in the inductance measured across the coil. Most magnet systems use steel to guide and concentrate the magnetic flux from the permanent magnet to the gap, but the flux induced in the steel is not linearly related to the incident magnetizing force, and hence the inductance in the coil is not simply related to the current flowing through it. This nonlinearity will manifest itself as a distortion in the current, and hence the forces developed in the voice coil.
A further complication arises because some of the flux induced by the coil finds itself going right round the loop of the permanent magnet circuit, and because the coupling of the coil to this loop depends on the position of the coil, when the coil is all the way into the gap, then the effect is at a maximum, and vice versa. One effect of this is that the force factor experienced by the voice coil will vary with position, and if the exciting signal is a sinusoid, then the effect will be the same as adding a signal of twice the frequency, otherwise known as second-harmonic distortion.
As the inductance measured on the voice-coil terminals depends on the way the current induces magnetic flux in the steelwork, then this inductance will also vary with the coil position. As the large excursions tend to be created by the bass frequencies in the music and the inductance tends to affect the high frequencies, the overall effect of this behavior is that the high frequencies will be modulated by the low.
There are a number of ways of mitigating these shortcomings, the first being in the design of the magnet system. This might involve the use of an extended pole, so the coil is never entirely in fresh air; another is to use an underhung design in which, by definition, the coil is always surrounded by steel. Encasing the pole in a thin sheet of copper can also be effective in preventing the flux induced by the coil penetrating into the steel. This works because the changing flux induces eddy currents in the copper that oppose the changes. Another approach, better suited to reducing the portion of flux that ends up in the overall loop, is to put a thick copper ring around the pole. The change of flux in the pole induces eddy currents in the loop that oppose the change. One final refinement in this mold is to have an extra coil wound directly onto the steel pole. By judicious choice of drive current, it should be possible to neutralize any induced effects.
Of course, one further route might be to leave out the steel altogether, since the permanent magnets themselves behave like free air because all their domains are tied up. This was a relatively easy option for alnico magnet systems, since they have a rather long, thin aspect ratio. This has become a great deal less practical in our world of short, fat ceramic or rare-earth magnets, although some attempts have been made.
JF: How about the diaphragm itself?
LD: The function of a speaker driver is, of course, to create sound, and on its own, the voice coil presents only a tiny vibrating area and must be coupled to a much larger diaphragm.
On the application of the voltage signal, the force experienced by the coil is immediately conducted into the supporting former on which it’s wound. The force acts along the axis of the tube, a direction in which it is stiffest; nevertheless, the speed at which that force is transmitted is finite, and long formers made of pure polymers should be avoided. Best to keep them short, and use a glass reinforcement or even a metal foil (although there lies a new source of problems, as the conductive foil also acts as a short-circuited turn that will draw power and damp motion).
Attached to the far end of the former is the diaphragm, which can take the form of a thin cone, a dome, or a mixture of the two; others have successfully employed flat composite sheets. It can be made from a huge range of materials—paper, plastic, fabric, foam, or metal, or complex composites of any or all of these materials—but they all have the same goal, which is to create a diaphragm that conducts the axial vibration of the former and transforms that energy into soundwaves in the air around the driver.
Perhaps it is the exact form that the diaphragm takes that divides speaker designers more than any other single factor. There are those who believe that the diaphragm should behave as a rigid piston throughout its range, and there are those who feel that, on the contrary, it should be a flat, moderately stiff sheet in which well-behaved transverse waves can propagate like ripples in a pond. Then there’s the world in between, who believe the right way is to use a cone of material excited at its apex and supported at the outside edge by a flexible surround. This last and most popular approach is probably the easiest to manufacture, and can couple a manageable motor size with a large cone, thus producing a reasonable low-frequency output with a moderate excursion. At low frequencies the whole assembly moves as one, but as the frequency rises, the edge of the cone begins to lag behind the apex, until a point is reached where the two are completely out of phase. So the cone is not a rigid piston, but is better thought of as a transverse transmission line swept around the central axis.
It helps to visualize the situation if the cone were made from a soft rubber rather than stiff materials: you should be better able to imagine the waves traveling outwards. On their own, the traveling waves would be fine, but eventually the cone ends and the waves experience an abrupt change of circumstance that causes them to bounce and start the return journey. They end up back at the apex, where a similar shock greets them—so they start the journey again, and if they happen to be in phase with the signal of the voice coil, then the wave will be reinforced. This is known as a resonance, and some frequencies will experience an increase in level while others suffer the opposite effect.
For many combinations of ordinary materials, such as paper, plastic, and doped fabrics, these effects occur in the middle frequencies and, left unchecked, impart a very distinctive character to the sound that results. Since most speaker cones are attached to a flexible surround, the detail of exactly what happens at this transition point holds the key: With the proper choice of materials, the wave may be coaxed into leaving the cone and continuing through the surround, where, again, material choice permitting, the wave may be dissipated as heat. The reality is that no single material exists that can truly satisfy the edge-termination conditions at all frequencies—but still, some very respectable results have been achieved.
Other approaches to minimizing cone-breakup effects might use anisotropic materials so that the wave velocity differs with direction. A popular option is to use a woven material so the properties vary between the 0-degrees/90-degrees and the +45-degrees/-45-degrees directions, thus breaking up the waves. Another might be to use a noncircular cone; clipping the edges to form a slightly square shape might then have the desired effect.
The situation changes somewhat when you move to metals, because they have a very different stiffness-to-density ratio, and now the speed of propagation is such that the first breakup can be comfortably beyond the frequency where you might have stopped using such a driver size for other reasons. Now the driver can be effectively said to be behaving pistonically in its passband.
Whereas cones are the profile of choice for almost all bass drivers, dome diaphragms with peripheral voice coils became popular for use in tweeters because that combination gives you the largest voice coil, hence the greatest power handling, for a given radiating area. It also happens that if you’re going for high breakup, then the metal dome is the design of choice. For diameters of 25 to 75mm, domes offer an excellent route to good behavior.
So is this the end of all our woes? Well, possibly not. For one thing, a pistonic driver is going to have a narrowing beam at the upper end of its band. That’s just one of those rules of acoustics: If the circumference of the piston is one wavelength, then the sound will be noticeably concentrated into a beam. One of the possible advantages of the flexible-cone approach is that while the outer edges are being left behind, the apparent size of the sound source is getting smaller, so it doesn’t beam in quite the same way as the uniform piston.
So we have our diaphragm vibrating in more or less the way we want—can we now expect the translation of this motion into sound to be easy from this point? This depends on the exact shape of the cone, the way the driver edge fits into the enclosure, the magnet shape, and the chassis that holds everything together. We’re all familiar with the resonant qualities of contained volumes of air, but what may not always be appreciated is that there is no sharp transition between a bottle and a depression in a flat sheet—and while it may not be so obvious, the latter situation will exhibit a mild resonant effect. Loudspeaker cones on a baffle are, in effect, a shallow depression in a flat sheet, and therefore are subject to this mild resonant boost. It usually amounts to a 2 or 3dB bump somewhere in the 1-2kHz range and may not be much of an issue, but can be avoided if the diaphragm is convex rather than concave. You may point out that the concavity is still present, albeit to the rear, and can still cause mischief. Well, indeed it can, but there is a complete solution to this that is arguably no longer in the realm of the driver. That would be to attach the rear of the driver to a tube, thereby removing the transition from cavity to free space, and hence the resonance.
Still on the subject of parasitic acoustic resonances, attention should be paid to the way in which the sound from the rear of the cone emerges from the chassis. Again, if the air is somehow enclosed and has to pass through gaps between struts, then it is inevitable that a cavity resonance will result. How this manifests on the front output of the driver will depend on many factors, but, as usual, avoiding it in the first place is the best approach. Thin struts of minimal area will always win over punched holes in a pressed-sheet bowl.
JF: And the perfect drive unit would look like . . .
LD: So, having analyzed the things that can go wrong, what would be my suggestion for an ideal driver? For one thing, we need to specify a frequency range, because spanning the entire audio spectrum with one driver is quite a tall order. But while it would be unlikely that we could produce a single driver capable of a satisfying low-frequency output that would also produce a clean top end out to 20kHz, it is much more possible to imagine a driver that sails past that upper limit while still being able to move a few millimeters, and is therefore admirably suited to being used in large arrays and capable of a reasonable bass output. So my ideal driver would probably use a 50mm beryllium dome with peripherally attached voice coil. The magnet system would use radially polarized magnet segments to permit the smallest overall diameter, while also leaving the largest open area for the rear output to connect to an absorber tube. Actually, I would suggest there be two versions: a long-throw model for use as a full-spectrum driver in large arrays, and a short-throw for use as a midrange driver.
A Vivid Audio D50 driver in its tapered tube
Funnily enough, apart from the beryllium part, this describes the D50 driver, which has been central to our Oval and Giya speaker ranges since the beginning of Vivid Audio. Of course, on its own, a 50mm piston does have a narrowing beamwidth, and for this reason we have always coupled it with a tweeter which is pretty much a half-scale version of the same thing—but just listening to a single D50 in an ostrich-egg-shaped cabinet does have a certain magic. Maybe it’s that single-driver phase-response thing?
The loudspeaker system
As you can see from Laurence Dickie’s detailed account, the loudspeaker drive unit is integral to the functioning of the loudspeaker itself. But that drive unit is only part of the overall device called a loudspeaker. Other aspects of any speaker’s design will have a profound impact on the performance of its drive-units. The shape and thickness of the baffle, what it’s made of, how the drivers are mounted on it—everything matters. What’s behind the baffle also matters—the cabinet’s internal volume, the shape of its interior walls, how the backwave produced by the driver is absorbed, etc. Every aspect of the crossover—its slopes, crossover frequencies, component parts, and more—is also important. The number of drivers, their positioning on the baffle, and their configuration (two-way, three-way, etc.)—all matter. The drive unit can be experienced optimally only within a finished loudspeaker system in which a multitude of factors define its performance.
In short: Drive units are never heard in isolation. But hear them you certainly do.
. . . Jeff Fritz
All photos courtesy Jake Purches.