By Moray Campbell and Scott Robinson
Moray Campbell is the design engineer who led the team that developed the Melior Point Source speaker for Museatex. Scott Robinson is a Canadian freelance writer and associated with Museatex.
The word "loudspeaker" means very different things to different people. To some, it connotes a small box that can be neatly placed on a shelf or side table. To others, the image is of a towering panel. A hi-fi buff with more esoteric tastes might picture an obtusely chiseled shape that more nearly resembles a sculptured work of art.
Each of these mental images is accurate. Loudspeaker designers have used all of these approaches in making systems that change the signal voltages coming from the amplifier into sound waves that the ear can hear. When all is said and done, however, there are very few basic ways for a speaker system to operate--so when someone does come up with a new way to make a speaker work, it usually captures a good deal of general interest. Museatex Audio believes its Melior loudspeaker has indeed added some new wrinkles to design.
Despite the wide range of styles and methodologies in speaker design, there is still one mechanical principle common to almost all: The use of a simple piston motion by a rigid diaphragm to create air motion. Generally, speaker designers regard any deviation of the diaphragm from pistonic motion as distortion. Yet the Melior Point Source speakers deliberately avoid making the diaphragm behave as a simple piston.
While a piston is indeed an effective method of producing the sound pressure levels in the air that our ears require to perceive sound, it is also a mechanism that presents several fundamental physical limitations in the construction of a full-range speaker. If the diaphragm of a regular dynamic (cone-type) driver is to move as a piston, accurately following the motion of the voice-coil, it must be rigid, lightweight, and non-resonant. In order for this speaker to reproduce lower frequencies, the diaphragm must also be appropriately large. The problem is that there is always a trade-off between the size of a speaker's diaphragm and its weight or rigidity. A diaphragm large enough to reproduce lower frequencies and rigid enough to have low distortion will weigh too much to efficiently reproduce higher frequencies. On the other hand, a diaphragm that is rigid and light enough to operate efficiently at the faster speeds needed to reproduce higher frequencies will not be large enough to reproduce bass. Consequently, full-range dynamic loudspeaker systems must use separate drivers for different frequency ranges. This, in turn, requires a crossover network to distribute the workload among the drivers.
Several planar loudspeakers--some electrostatic, some magnetic-have been able to achieve an effective com promise between size and weight.
These speakers are driven by forces that act evenly over the entire surface of their diaphragms, so rigidity is not much of a factor. This enables them to be large and lightweight enough to accurately reproduce the full frequency range without having to resort to separate drivers, although several of these speakers still make use of crossovers before their transformer stages. The problem with most planar speakers is that, above a certain frequency, the radiated energy begins to concentrate into an increasingly narrow angle directly in front of the speaker. This produces the effect known as treble beaming, where frequencies whose wavelength is shorter than the width of the diaphragm have very narrow dispersion. Such a beaming of certain frequencies requires listeners to sit in a specific "sweet spot" for best sound.
The theoretically ideal speaker has been discussed by designers for decades, and almost always the answer is a pulsating sphere. If a true pulsating sphere were built, it would behave as if the sound came from a tiny point source located at the center of the sphere. There would be perfect omni directional dispersion at all frequencies without any sort of beaming. Early attempts at building such a speaker have included designs using quarter spheres and cylinders.
The problem with a spherical speaker is that its curved diaphragm must pulsate by expanding and contracting over the entire surface. Such technology does not exist at present, with the exception of some gas plasma tweeters. Attempts at curved diaphragms have shown that it is difficult to con struct a diaphragm that is sufficiently rigid to maintain a curved shape while remaining elastic enough to allow the surface to pulsate evenly. This is why existing curved designs have been forced to "borrow" from the elastic material at the edges of the diaphragm in order to allow the more rigid material at the center to pulsate. This can result in the edges of the diaphragm moving out of phase with the center at some frequencies.
Our design team thought that it might be possible to reproduce the hemispherical wavefront of a pulsating sphere from a flat diaphragm. Clearly, a simple piston action could not do this. The key was to imagine what a hemispherical wavefront would do if it passed through a thin, acoustically transparent diaphragm. As the curved wavefront expanded away from its point source toward the diaphragm (Fig. 1A), it would first contact the diaphragm at its center (Fig. 1B). The wavefront would pass through the diaphragm into the air on the other side, while the initial displacement of the diaphragm would expand across its surface to its edges (Fig. 1C), much as ripples on the surface of water.
Following this imaginary example, we designed a speaker that, instead of acting like a piston, operates by driving only a point at the center of a Mylar diaphragm which is fixed at the edges.
The rest of the diaphragm is not directly driven, but follows naturally from the motion of the driven central section to create the desired ripple effect. (Compare this system, illustrated in Fig. 2, with the imaginary example shown in Fig. 1.) Because of the natural time delay involved in the spread of energy across the diaphragm from the center, the sound pressure wavefront travel ling through the air away from the speaker gets a head start at the center of the diaphragm and lags at the edges. This accumulated delay in the transmission of the wavefront across the diaphragm acts to curve the wave-front's edge as it moves into the air, providing a close approximation of the ideal spherically spreading wavefront at all frequencies. Of course, the dipolar nature of our flat diaphragm is able to produce only half of this spherical wavefront in constant phase. So our design complements the forward-moving hemispherical wavefront with a similar wavefront in reverse phase that travels away from the back of the speaker. This is very similar to the effect produced by the Quad ESL-63.
While the Quad uses inductor delay lines for frequency shaping and time delays, the Museatex Point Source speaker produces the same effect by purely mechanical means, without delay lines, crossovers, frequency shaping, or other electronic processing.
Although the Melior diaphragm is made large enough to reproduce bass (much as the diaphragm of a conventional planar loudspeaker), it behaves as if it were the ideal small pulsating sphere. Our measurements show al most constant directivity over a 60° arc at all frequencies from 45 Hz to 8 kHz. (See the polar responses in Fig. 3.) Above and below these limits, the speaker's coverage begins to narrow.
Thus, with careful design to keep the mass of the moving element low enough for high-frequency reproduction, a full-range speaker without treble beaming is possible.
To obtain a linear response over the audible frequency range (Fig. 4), it is necessary not only to keep the mass of the moving system as low as possible but also to keep coil inductance to a minimum in order to yield a relatively flat impedance characteristic. This results in a flat power response and, because there are no other components in the speaker circuit, a very easy amplifier load. Fortunately, a low-inductance coil is consistent with the requirement for low mass. We use a coil with an out-of-gap inductance of less than 0.1 mH and only 60 turns of cop per-clad aluminum wire. The speaker's impedance curve (Fig. 5) therefore resembles a very gentle, classic single-coil plot, reaching a minimum of about 4.5 ohms in the midband and rising smoothly to less than 6 ohms at 20 kHz. The bass resonance is extremely well damped, with a rise to only 7.5 ohms.
The low-inductance format, together with the use of a Kapton former and the copper-clad aluminum wire, yields a total mass of 1.5 grams for the coil and the former. The complete motor assembly--including dust cap, tinsel connecting leads, and all adhesives weighs only 2.2 grams. Our sensitivity rating is 85 dB at 1 meter with 2.83 V rms applied. Very little additional diaphragm or air mass is coupled to the motor, particularly at critical higher frequencies, as the travelling-wave operation of the speaker allows the centrally located motor to work independently of the outlying portions of the diaphragm.
Fig. 1--Passage of a theoretical point source wavefront through a diaphragm.
As the expanding wave nears the diaphragm (A), it first contacts the diaphragm's
center (B), then passes through. During this passage, the displacement of the
diaphragm expands in a ring around the initial contact point (C).
Fig. 2--Action of the Melior speaker. Diaphragm displacements caused by the
voice-coil (A) spread in a ring as the resulting wavefront expands (B and C)
Note similarity to Fig. 1.
Fig. 3--The Melior speaker's polar response at low frequencies (A), middle
frequencies (B), and high frequencies (C).
Fig. 4--Melior speaker's averaged in-room frequency response at 1 meter with
2.83-V input signal.
Fig. 5--Impedance vs. frequency.
The coil relies on the diaphragm itself for positioning relative to the mag net gap. No spider is used in the design, because spiders were found to contribute significant coloration to the sound. To prevent coil rubbing, we developed a "fluid bearing" within the magnet gap using a ferromagnetic flu id, and we believe this to be a unique application in a bass driver. The ferro magnetic fluid also serves to increase the thermal capacity of the coil. Further thermal-protection' measures include the use of a relatively long magnet gap, copper pole-piece sheaths, and damping rings. These items not only function as flux-damping mechanisms but also ensure that under heavy mid-band drive, when current is high but excursion relatively low, the entire coil has effective heat dissipation.
The damping rings and copper sheaths serve the traditional roles of reducing in-gap inductance and con trolling high-frequency roll-off. They also reduce bass intermodulation caused by changes to inductance with position in the gap. Both of these techniques are well known in speaker de sign but are often not used because of their additional expense. Multi-way systems can use separate drivers to get around these problems, but we had no such choice and found these solutions proved to be effective.
Since the diaphragm must be held under tension, the system resembles a drum, and it will generate tympanic resonant modes at certain frequencies if some form of damping is not applied.
We experimented with several different approaches to damping as well as with methods of attenuating or spreading the frequencies at which resonance occurs. Ultimately, we decided against adding stiffeners or mass and against treating the diaphragm with any of the commercially available damping formulations. Instead, a damping system that makes use of resistive air loading was adopted.
The damping system operates in a similar way to that of a conventional box enclosure, providing an air load to the diaphragm that controls the Q of the tympanic modes. However, our system of resistive damping is positioned very close to the diaphragm, employing a nonwoven polyester fabric that is firmly supported by a plastic grid-work behind the diaphragm.
(There is no damping layer in front of the diaphragm.) Despite its relatively complex construction, the damping layer still allows sound to pass, pre serving the speaker's dipolar, double-hemisphere radiation. The system allows us to easily vary the degree of damping across the diaphragm in order to combat specific modes, making it possible to optimally tune the speaker for maximum bass extension while minimizing resonances. (This tuning process began with the basic idea that more damping would be required near the edges of the diaphragm, which are not directly driven.) Further experimentation resulted in critically damped diaphragms for several different-size speakers.
In addition to applying the principle to full-size, floor-standing speakers, our Melior One and Two, we have also used it in our in-wall and satellite models. Naturally, we will seek other applications.
(Source: Audio magazine, Jun. 1992 )
Thunder in the Listening Room--Subwoofer Shootout (Nov. 1992)
= = = =