And the next speaker is...!
The first two parts of this series emphasised the early history of loudspeaker technology. Both articles mapped the route by which current technology originally developed and then evolved, rather than the white heat of modern loudspeaker research and technology. This is appropriate - the story has not changed that much in recent times and loudspeaker technology has of late been, not entirely stagnant, but seen few thoroughly innovative directions. In the past 100 years most principles and methods that might produce a sound have been fully explored. Nowadays, one might be forgiven for seeing little beyond the constant effort to evolve and perfect the moving coil loudspeaker, and its enclosure, which we know and love so well. Nevertheless, invention and development does continue.
Companies like Tannoy, B&W, KEF, Bose, Mission and Goodmans have made great strides in doing a lot better with what we have already. Indeed, an effort has gone into improving the basic cone loudspeaker and its enclosures. Bass reflex and acoustic labyrinth designs are commonplace, and moving coil loudspeakers themselves are now reaching the limit of their potential. Even the vacuum enclosed miniature speakers (supposedly bass efficient) are not entirely new. However, I suspect that most players in this field, subject to a little hard interrogation, would concede that nothing done of late really amounts to a technological revolution, unless you count the NXT tile loudspeaker. Here, prima facie, is something of a revolutionary move in loudspeaker technology.
Related further reading...
The story, it appears, begins with attempts by DERA (Defence Evaluation Research Agency) engineers to identify and prevent sympathetic vibration or resonance, in laminates used in the inside of helicopters and aircraft. In an inevitably noisy environment at the best of times, the added rattle from internal cladding was a distinct distraction for air crew. In the course of investigations it was unexpectedly discovered that the laminates in question were acting as wide band, multiple pole radiators. That is, given a range of extraneous vibratory frequencies, they did not simply act as a simple, spring-mass resonator at one frequency or harmonic, but were amplifying a range of frequencies cross their surfaces. This, we are told, was not good and, more to the point, classical physics decreed that the laminates should not have been doing what they were doing!
Further investigation showed that with critically thin sheets of stiff materials, a series of discreet flexure (bending) wave effects could be generated, producing a series of distributed modes of vibration across the whole surface. In short, the sheet becomes a variable sound radiator across the whole of its surface area. This appeared to be unlike a conventional speaker, where the essential piston action at low frequencies gives way to combined piston, flexural action above a critical frequency.
Trying to find out the best way to create this distributed mode behaviour and exploit it was, to say the least, far from straightforward. Early attempts at exciting this distributed mode action gave a less than uniform response across the audio range. So, DERA, having patented the basic principle, looked for an appropriate partner with the right background to pursue a technology that showed a lot of promise, but needed a wider expertise in acoustics and loudspeaker technology. They opened negotiations with the Verity Group Pic. (Wharfedale, Mission etc.) and ultimately, the technology was transferred from DERA under a joint collaborative and licensing agreement, forming a new research and IP holding company, NXT New Transducers Ltd.
Even Verity had their doubts, but viewed it as an analytical problem whose solution promised a new era in loudspeaker technology. The story goes that subsequent work by a range of analysts, Neil Harris and Henry Azima in particular, slowly unravelled the way the sound radiation properties were optimised. Over time, Harris and Azima, along with Verity's IPR. advisor, Martin Colloms, have applied for some 23 patents covering the NXT rigid panel and driver technology.
The research followed a series of experimental and analytical procedures all intended to elucidate the conditions which determined panel acoustic behaviour, and would point to design factors enabling a near-ideal acoustic radiator to be produced. As might be imagined, the factors were complex and interrelated. Radiation effects (nodes and modal distribution) are dependent on surface density, stiffness, geometry, surface area, position of drive point, parameters of drive point and method, shear modulus of core, internal damping and load stress caused in suspending the tile. In compensation, it was found that a number of methods could be contrived to create the right sound radiation - none of which create complex loads for drive amplifiers.
The result is a thin, dimensionally controlled rigid panel which creates a multi-modal sound source able to radiate sound over a very wide frequency band. It is heavily advantaged by the fact that, unlike conventional loudspeakers, the front to back path length is not critical, especially in terms of 'out of phase' destructive interference creating sound cancellation. Instead, the front and back panel surfaces produce in-phase signals. Furthermore, unlike the narrow radiation field of a cone speaker the polar diagram of the tile is virtually omni-directional. A typical frequency range for Hi-Fi applications is lOOHz to 18KHz.
Nothing Is Perfect
Nevertheless, nothing is perfect. Its detractors would argue that NXT may have its good points, but at bottom it may not be all that much of a theoretical breakthrough. It is in operation a complex composite of stiffness controlled beams operating as separate oscillators - each nodal plane is part of a distributed vibrational state across the surface. These states are produced as a series of defined vibrational elements created by the complex amplitude of the applied audio signal. Indeed, though not as simple as, say, a violin sounding board, any physicist will tell you that the phenomena is, fundamentally, part of classically understood acoustic phenomena investigated by Chladni, Rayleigh and others.
Furthermore, in this context, it is quite possible to argue that the NXT is no more than a planar version of an ordinary loudspeaker diaphragm, operating in flexure mode above its 'piston' frequency. In short we are back to the large surface, low displacement model.
However, there is a difference, and what differs is the multi-modal surface - that is, the whole surface is capable of becoming a large array of separate acoustic radiators at any one time. By choosing the right material properties, dimensions and driver characteristics for the panel it is possible to manufacture a widely dispersive sound radiator with very low distortion in the upper scale.
On the plus side, this makes the mid-range and high frequency response very uniform with little distortion or colouration. But at low frequencies, the low distortion level is disguised somewhat by neglecting a severe roll-off in low frequency response. Published data gives, as typical, the 50Hz output being 12db down compared to the 200Hz level. In light of this, there is much to question about the meaning of other published data giving a significant increase in distortion below lOOHz (up to 10% 2nd and 3rd harmonic). Comparisons with the low frequency distortion of a (typical?) moving coil speaker on the same scale look good, but not against what is also a rapidly decreasing response (-12db). If the roll off response and distortion were normalised the comparison would certainly not flatter the NXT.
With no base line reference datum for response at these low frequencies the data is thus somewhat facile. A good deal of argument will also ensue about the exact effect of the drive method. The favoured method is a driver coil which is essentially the same annular arrangement found in conventional speakers, with the coil located by a rubber septum. The coil is bonded to the panel and, like an ordinary loudspeaker the varying magnetic forces produce lateral forces on the coils. However, unlike your standard cone unit these forces result not in push-pull motion but, because of the magnet polarisation, lateral bending moments in the panel, (called an inertial magnet driver). The panel has an electrodynamic equivalent impedance having no significant reactive term. Instead the load is predominately resistive. Using the Inertial Magnet driver, the electrical load looks like a low pass filter with the high frequency extension determined by the coil mass and inductance, the panel resistance, and the propagation attenuation. Yet another method uses a magnet mechanically coupled to the panel and the coils are independently mounted by the magnet (clamped (bender) driver).
These methods are constant velocity, mass controlled drivers and, given the nature of the sound radiator, low mass drivers are best able to optimise performance. The favoured method, therefore, is to use the inertial magnet driver critically positioned on the 'tile'.
In some cases, a piezoelectric driver is added since, being a constant displacement device, it may be used successfully to compensate for high frequency roll-off, thereby producing a flat response.
However, let us be cautious. This approach would, after all, be a drive
system typical of a conventional mid to high range piezo driver, and the published NXT performance data is suspiciously like that of your average upper range piezo loudspeaker.
But this is where the similarity ends and, if at this juncture, the reader detects a tendency to be somewhat scathing, pedantic even, the criticism ends too!
By virtue of its planar-like dimensions and ability to cover a considerable area easily, it becomes a hugely dispersive sound radiator moving away from the virtual 'point source' or porthole effect of a conventional loudspeaker. With a superbly diffuse, effortless distribution of the mid and higher audio spectrum, and very low coloration, the NXT technology gave the author an impression of vastness and expanse in the reproduction of music. It almost puts the quality of electrostatic speakers to shame. Voice reproduction is disturbingly realistic and, curiously, the low end roll off and peaking distortion is hardly noticeable, since all the higher bass harmonics are still well reproduced. As to the poor bass fundamental response, NXT is the first to recognise that the panel will take a lot of development to meet the power and bottom end response of your average bass unit. But coupled to a woofer/sub-woofer, the panel or panels can produce the most wonderful listening experience
A typical system presents a near constant impedance over its whole frequency range. It is, therefore, more than suitable for standard amplifiers presenting nothing in the way of a complex reactive load. This is a definite plus. Of distinct advantage is the fact that the sound intensity versus distance plot is unlike the inverse square relationship of a conventional driver - the NXT's large area and diffuse radiation pattern means that the measured and subjective intensity hardly changes with distance.
Where the NXT technology will go next (pun intended) is yet to be announced. It has enormous interest, from applications in cinema sound reproduction systems (using the screen) to zero profile loudspeakers for mobile phones, car sound systems and laptops.
It could be very close to the solution we have all been waiting for, namely a sound reproduction system which makes sound without the need for bulky enclosures or expensive drive units and can be shaped or moulded almost at will.
The sheer flexibility of the technology has created a flood of licensing and technology transfer agreements. It offers the potential for a widely dispersive source that one would expect at a live performance. The hi-fi purist, just 3% of the market incidentally, may be sceptical about its future for high-quality sound reproduction, but it is likely that any minor performance limitations will be resolved as the technical development proceeds.
The panels are expected to give good value for money. In terms of cost against quality, the promise of being able to double or treble the radiation surface, for less than the cost of a typical high quality enclosure, is an enormous bonus. Whatever the analogy or affinity with standard moving coil units, the NXT technology exploits it in a controlled and well considered way. Not everyone will agree with NXT (New Transducers Ltd) about how their NXT panel actually works, or if in reality they are simply doing something a bit different with what we have already. But frankly it doesn't matter - it's bloody good!
Sincere thanks to John Visor and Russ Brown at NXT for much kindness.
Typical multi-modal surface of operational panel