Lynn Olson's terrific article about the basics of speaker design and construction. PART 1 This is the first section of an article published in the Vol. 3, Issue 6 of Positive Feedback magazine. It is my *personal* view of the art of designing high-end loudspeakers, and is not intended as any kind of "bible" for folks new to high-end audio. Since I've gotten a lot of questions about the kind of thing that appears in Positive Feedback, it's probably simplest to demonstrate by reposting a complete article in this forum. ------------------snip----------------* Looking Over My Shoulder, Part I * by Lynn Olson * Introduction Where did I come from? Where am I going? Who am I? These ancient questions of philosophy are the first questions you must ask yourself if you are serious about designing audio equipment. These questions repeat themselves in only slightly altered form: What is the history of the art of sound? What is is the true potential of the future? And what, really, do you want to do? * The Future If you relax and make a mental journey to the far future, it is easy to imagine the perfect loudspeaker. It would made of an immense number of tiny point sources that would create a true acoustic wavefront (or soundfield). Resonances due to massive drivers, cabinets, or frames would be a thing of the distant past. A myriad of waveform distortions (harmonic, intermodulation, crossmodulation, frequency, phase, and group delay) would be utterly absent ... the sound would be literally as clear as air itself.
This perfect loudspeaker would be made of trillions of microscopic coherent light and sound transducers, integrated with signal processing circuits all operating in parallel. (This is similar to present-day military phased-array radars, with tens of thousands of tiny antennas with integrated electronics subsystems.) It would be constructed by a combination of nanotechnology and genetic engineering and operate at the molecular level, appearing simply as a very thin film when not in operation. Just for a moment, imagine a thin-film mirror for 3-dimensional sound and images with a near-infinite random-access memory that is also connected to all other mirrors; it would be "transparent" in a lot more ways than one. * The Present Whew! Now that we've glimpsed perfection it shows just how far we have to go in 1994. Here's a partial list of the major problems we face now: * 2-speaker stereo falls woefully short of a true acoustic wavefront, producing a phasey, unrealistic image of small size that causes listening fatigue for many people (particularly non-audiophiles). The virtual image is unstable with respect to listener position, spectral energy distribution, and room characteristics. Even a simple central mono image has been shown to suffer from deep comb-filter cancellation nulls between between 1kHz and 4 kHz, which is why a solo vocalist sounds different coming from a single mono speaker and a conventional stereo pair. It now appears that 2-channel stereo requires a minimum of 3 speakers to faithfully represent the tonal quality of centrally located sound sources, such as vocalists. * Large amounts of harmonic, intermodulation, and crossmodulation distortions combine with mechanical driver resonances to concentrate spectral energy at certain frequencies. Driver damping techniques usually improve spectral characteristics (the frequency response curve looks better as a result) but do not provide much improvement for the underlying breakup modes, so the distortions may actually be spread over a much broader frequency range. The narrowband nature of resonant distortions in loudspeakers is why a single-frequency THD or IM measurement is useless; it takes an expensive tracking-generator type of measuring system in order to
create a usable frequency vs. harmonic distortion graph. In order to eliminate these resonant distortions, the driver diaphragm needs to have a density equal to air and absolutely uniform acceleration over the entire surface at all frequencies. As you can imagine, we are nowhere close to meeting this criterion. As a result, all speakers have tonal colorations ranging from subtle to gross, with some types of colorations present at all times, and other types of colorations appearing only at high or low levels. A reviewer's preferences in music can easily mask the presence of these problems, so make sure to get a second (and third) opinion. * Standing-wave resonant energy is stored in drivers (all types, except for "massless" exotics), cabinets, and in the listening room itself. The unwanted mechanical energy must be quickly removed in two ways: rigid, low-loss mechanical links to the earth itself (a rigid path from the magnet to stand to floor to ground), and also dissipated as heat energy in high-loss, amorphous materials such as lead, sand, sorbothane, etc. The energy that is not removed is slowly released from every single mechanical part of the speaker, each of which has its own resonant signature. In any real speaker system, regardless of operating principle, there are hundreds of standing-wave resonances at any one time, which are gradually released over times ranging from milliseconds to several seconds. These resonances continually overlay the actual structure of the music and alter the tonal color, distort and hide the reverberent qualities of the original recording, and deform and blur the stereo image. In speakers that measure "textbook-perfect", this type of "hidden" resonance is the dominant source of coloration. This is also the reason that 1/3 octave pink-noise measurement techniques have fallen out of favor, being replaced by much more sensitive techniques such as TDS, FFT, MLSSA, and others. * Radiation patterns shift dramatically with frequency, and change sharply at crossover points; in addition, the radiation pattern is further deformed by diffractive re-radiation at every sharp cabinet edge (regardless of cabinet size or type - this includes planar types of loudspeakers). Diffraction, which occurs at every sharp cabinet boundary, creates delayed,
reverse-phase phantom sources that combine with the direct sound from the actual driver. These secondary phantom images create significant ripples in the midrange response (up to 6 dB) and create delayed sounds which interfere with the timing cues necessary to perceive stereo images. These dispersion problems are audible as room-dependent colorations, harsh or dim midrange and treble, fatiguing stereo, and a "detenting" effect that pulls images in towards the loudspeaker cabinets. * This list only covers some of the problems of contemporary loudspeakers. There are other problems, not as severe, but still quite audible to a skilled listener. These problems occur in all loudspeaker types - dynamic direct radiator, horns, ribbons, electromagnetic planar, electrostatic planar, you name it. They all have lots of THD and IM distortion concentrated at certain frequencies, they all store and release significant amounts of resonant energy, and they all have frequency-dependent dispersion further degraded by diffractive re-radiation. This is why I treat claims of "perfection", or of a "major breakthrough", with a big grain of salt. Does it address even one of the serious flaws cited above? Not too often. By contrast, what we're really seeing is a steady, progressive improvement in materials technology and big steps in measurement technology and computer modelling. (I, for one, am grateful, since it's been a major improvement compared to what I had to work with at Audionics in the mid-Seventies!) * The Past Modern speakers are far better than the speakers of the Fifties (which I still remember). Very few people had full-blown Altec "Voice of the Theatre" A-5 systems, 3-way Bozak B-305's, or Klipschorns. Most "hi-fi nuts" had to put up with University, Jensen, or Electro-Voice 12" coaxial drivers in big plywood boxes with a single layer of fiberglass on the rear wall. A large cutout served as the vent, resulting in boomy, resonant boxes tuned much too high, with 6 to 12 dB peaks in the 80 to 150 Hz region. (Have you ever heard a restored jukebox?) The coax, or worse, triax drivers went into paper cone breakup at 200 Hz and above, cavity resonances (due to the horn element mounted in the cone driver) at 800 Hz and above, horn breakup throughout the working range of the short horn, and phenolic diaphragm breakup at 8 kHz and above. A "good"
driver of this type usually had a plus/minus tolerance of 4 to 8 dB, and it took a lot of judicious pen damping to get it to measure that well. It wasn't for nothing that early hi-fi acquired a "boom-and-tweet" reputation. The sound quality was closer to modern-day autosound, or an un-renovated neighborhood theatre, than a modern system. The tube electronics helped smooth out much of the shrillness, but they couldn't rescue the truly mediocre phono cartridges and loudspeakers of the day. (Even so, the first-generation Quad, the RCA LC-1A, the Tannoy, and the Lowther compare very well with modern systems ... but these were quite rare at the time.) I find it interesting that second-generation transistor amplifiers like the Dyna 120, Crown DC-300, and Phase Linear 400 were favorably compared to the classic vacuum-tube Dyna Stereo 70 and Marantz 9 in the early Seventies. (Even by J. Gordon Holt's "Stereophile" magazine!) That tells us a lot about the resolving power of the best speakers of the early Seventies. Progressive improvements in speaker design now reveal the actual sonic quality of these second-generation transistor amplifiers as quite repellent, while the "freshened-up" tube amplifiers sound as good or better than many expensive transistor amplifiers made today. It does make you wonder about the resale value of the current crop of multi-thousand dollar transistor amps and D/A converters - what will they sound like on speakers ten years hence? By then we'll be listening to speakers with evaporated diamond diaphragms and other materials with very low distortion and stored energy, a long-overdue high-fidelity digital storage medium with 24-bit depth and 96 kHz sampling, and the possibility of replacing conventional 2-speaker stereo with something more natural. You can be confident the sound will be far more transparent and realistic than what we're accustomed to now. * Where Do You Start Your Design? What kind of sound do you like? People really do hear in quite different ways, and different people assign importance to different qualities of sound. Some audiophiles value timbral (or "tonal") qualities above all else, treasuring the sound of their favorite instruments or voices; some like a sense of immediacy, directness, and emotional impact; some like the sensation of an immense 3D space; and others like a see-through
transparency, a palpable "you are there" quality. It helps a lot if you know what's important to YOU, what you tend to dismiss, and what you don't hear at all. We all tend to unconsciously define the bounds of "reality" by our own personal perceptions and thoughts, but this just isn't so. Reality is far, far larger than any personal set of boundaries. That's why getting a second and third opinion is so important. Since all speakers have serious flaws in the absolute sense, it's up to you to select the qualities that are most important, and most believable, to you personally. "Perfect Sound Forever" is an arrogant marketing slogan, not a realistic goal for an artist. For one thing, the materials to build anything of the sort just don't exist. (Unless you've found a way to generate a controllable room-temperature plasma. If you have, you'd better talk to the Department of Energy first.) * Major Schools of Speaker Design Since all designers are forced to choose on a subjective basis, there is no single "right" or "wrong" way to design a speaker. If anyone tells you that, it might be amusing to investigate their personal beliefs a little further and see if they worship at a church of religious fundamentalism or the invisible church of "scientific" fundamentalism. Where do I stand personally? Well, I'm certainly not a fundamentalist! Oh, you meant speaker design! I care about to spectral flatness, very low energy storage, minimizing IM and FM distortion, and low diffraction, in that order. Subjectively, I seek out an elusive quality I call "the bloom of life"... that rare "You-Are-There" sensation of being in the presence of musicians creating beauty. In the section that follows, I'll describe the various paths that designers must choose as they make their way to sonic perfection. PART 2 This is a continuation of an article published in the Vol. 3, Issue 6 of
Positive Feedback magazine. It is my *personal* view of the art of designing high-end loudspeakers, and is not intended as any kind of "bible" for folks new to high-end audio. Since I've gotten a lot of questions about the kind of thing that appears in Positive Feedback, it's probably simplest to demonstrate by reposting a complete article in this forum. ------------------snip----------------* Pulse Coherent (3D-Imaging) Dynamics Duntech, Thiele, Spica, and Vandersteen systems fall in this group. The designer takes expensive steps to control diffraction, offset the drivers for a coherent arrival pattern, and usually employs a first-order (6 dB/Oct) crossover. Some, such as Spica, may use 3rd (18 dB/Oct) or 4th order (24 dB/Oct) Gaussian or Bessel crossovers. This is the only type of dynamic to offer accurate pulse reproduction, sometimes even bettering exotic electrostats or ribbon loudspeakers. However, the audibility of phase, and pulse distortion, is quite controversial in the engineering community, with the more conservative engineers feeling it is a waste of time and money to ensure accurate pulse reproduction. In a typical pulse-coherent design, the drivers are asked to be well controlled 2 or more octaves out of their normal operating ranges, so power-handling and IM distortion are compromised. Expensive drivers are required to partly overcome this problem, along with accurate resonance correction in the crossover. Controlling the radiation pattern with first-order crossovers and offset drivers is difficult; as a result, speakers of this type may sound quite different sitting and standing up. The yardstick for this type of design is the spatial quality of the stereo image; if it isn't signicantly better than any other type described below, the design is not a success, since that is the primary potential benefit of this design approach. * Flat Response (Objective-Design) Dynamics Most British and Canadian speakers fall in this group. They have very flat spectral responses, with the British paying more importance to the 1 or 2 meter on-axis response curve, and the NRC-influenced Canadians paying more
importance to the frequency response averaged over a forward-facing hemisphere. These design priorities have been arrived at by BBC broadcast professionals and NRC listening panels respectively, using statistically verifiable double-blind listening tests. This school of design is most closely identified with an "objective" engineering-oriented philosophy. Not by accident, the engineers with the most advanced academic credentials tend to design speakers using this philosophy. These folks are not going to be sympathetic to exotic wires, resistors, capacitors, the directly-heated triode mystique, or anything not audible to a double-blind listening panel. To its great credit, the BBC is known as the first group to accurately measure and identify sources of driver and cabinet resonances in the early Sixties, and many British speakers still excel in this area. Since audible resonances may be as far as 20 dB below a conventional sine-wave response curve, the BBC was the first organization to identify and measure colorations that were completely missed by the conventional sine-wave or 3rd-octive pink-noise measurements. These types of measurements are now considered a standard part of FFT, TDS, or MLSSA systems. Objective-design crossovers are usually 3rd-order Butterworth or 4th-order Linkwitz-Riley, which offer the flattest, most accurate response curve and the best control of out-of-band IM distortion (at the expense of pulse distortion and overshoot). Laurie Fincham of KEF deserves credit for designing the first computer-optimized crossover using acoustically-accurate rolloff curves combined with driver resonance correction (in the early Seventies). This crossover optimizing technique is now available at far less cost by using a 386, 486, or 586 PC with programs such as XOPT or LEAP. Competent designers are now expected to be knowledgeable in the design of acoustically-accurate crossovers, regardless of their design philosophy. Recent British designs also focus on a high-quality mechanical ground path to the floor (exotic stands). Objective-school designers usually ignore pulse response, diffraction control, and subjective areas such as capacitor, inductor, and wire quality. In contrast, research is focussed on steadily improving driver quality, cabinet resonance control, and accurate pair-matching in production.
* Minimalists (Subjective-Design) Dynamics Some Italian, Scandinavian, English, and American speakers fall in this group. The crossover is extremely simple, sometimes consisting of one capacitor to protect the tweeter (no, I don't know how Sonus Faber protects their tweeter without a capacitor). Drivers and crossover components are of the very highest quality, along with exotic wire and cabinet materials. Measurements usually play a minor role in the development of this type of speaker. Since this design philosophy leaves driver resonances uncorrected and accepts the resulting frequency and pulse response aberrations produced by the minimal crossover, compatibility may depend strongly on the sonic flavor of the rest of the audio chain. Personally, I find these types of speakers somewhat colored, but with an exciting, dynamic, and involving sound. * Horns and High Efficiency Dynamics Most Japanese high-end speakers fall in this group, as well as a handful of French, Italian, English and American systems. They use design philosophies derived from Western Electric theater speakers, Paul Voigt's Tractrix horn, or a combination of both. Nearly all horn systems are at their best when used with low-powered, sometimes single-ended, Class A directly-heated triode amplifiers, with transistor amps, even Class A types, frequently giving very poor results. Horns typically have extremely low THD, IM, and FM distortion, reasonably flat response, and sharp cutoff characteristics at both ends of the frequency range, which may be fairly narrow. Historically, horns were usually beset by intractable problems with impulse response, diffraction, and smooth dispersion, which is why most high-end designers (in the West) avoided horn systems, leaving them to the studio monitor and PA markets. In the past decade, though, Dr. Bruce Edgar in the US, and others in Japan, have made very significant improvements in horn design for high-end and ultra-fi audio, which are just beginning to be recognized by magazines like Sound Practices. According to the movers and shakers in the American ultra-fi renaissance of the 90's(Joe Roberts, Herb Reichert, and Mike LeFevre), these new horns, and the Edgarhorn in particular, are in a class
of their own, superior to ribbons, electrostats, planars, exotic dynamics, etc. My own opinion? Well, I feel that the trio I mentioned above have pretty good taste, so it's entirely possible I'll also like the new horns. Right now, I can't say anything until my own set of Edgar midrange horns arrive and I design them into a full-range speaker system. Stay tuned. * Electrostatic Planars A few English, American, and Japanese companies make this class of speaker, which I have to admit are old-time favorites of mine. A well-designed electrostatic offers the most linear and completely uniform diaphragm motion of any class of loudspeaker (and low IM distortion), as well as the potential for the best pulse response. The first Quad electrostatic is the most famous example of a speaker decades ahead of its time. There is a downside, of course, and that is very low efficiency, an extremely reactive amplifier load, restricted dynamic range, fragility, limited bass, and a tricky room-sensitive dipolar radiation pattern that becomes quite directive at high frequencies. These problems are not easy to solve, particularly the large-panel dispersion, which is not an asset, but a serious problem for stereo imaging. The old Quad pioneered the most widely used solution, a side-by-side 3-way system, using progressively narrower panels for the higher frequencies. The new Quad uses a complex phased array system which approximates a spherical radiator. The Martin-Logan uses an unusual cylindrical panel relying on curved damping pads stretched across the perforated high-voltage stators. Problems still remain, though. All of the electrostats I have measured show moderate resonances below 200 Hz (primary room-diaphragm resonance) and multiple sharp resonances above 8 kHz (non-homogenous diaphragm motion and standing waves in the HV stators or metal grill-frame assembly). In short, wonderful midrange and depth perspective, and good-but-not-great at the frequency extremes, reasonable-to-fair stereo imaging, and somewhat limited dynamic range. * Electromagnetic Ribbons and Planars
A few are made in the USA, being represented by Apogee, Magnepan, Eminent Technology, and others. These fall in two classes: the true ribbon, which is a very thin corrugated aluminum "voice coil" hanging freely like a streamer in a side-by-side magnetic field, and the magnetic-planars, which are sheets of stretched Mylar or Kapton film with the "voice coil" either printed or glued on the film. Magnetic-planars use arrays of magnets on the back side of the film (not good for distortion) or on both sides (operating in push-pull, but also creating a cavity between front and rear magnet pairs). The arrays of magnets provide a somewhat uneven drive field, so the uniformity of diaphragm motion is not in the same class as an electrostat. Then again, HV arcing is not problem, so the magnetic-planars can play a lot louder than their electrostatic cousins. The magnetic-planars have weak magnetic coupling, which is much lower than a dynamic driver due to the large magnet spacing and shorter length of wire in the gap. In a dynamic driver, a high "BL-product" means a strong magnetic field in the gap (the "B") and a long helical voice coil immersed in the field (the "L"). Dynamics with a high BL-product provide the tightest amplifier-speaker coupling, which is why they are sensitive to amplifier damping factor and wire resistance. By contrast, in a planar-magnetic, damping is mostly provided by the stretched film and the air load, and little by the amplifier. Both impedance and efficiency are lower than dynamics, and attempts to raise both by increasing the amount of aluminum wire on the film usually degrades the transient response. The true ribbon is free of the stretched film resonances and obstructing magnets of the planar-magnetic, so it offers outstanding pulse response, uniform drive, and a good approximation of a line source, but the efficiency and impedance are both phenomenally low and it is not usable as a woofer due to the small area. Most practical ribbons either use a stepdown transformer or ask the amplifier to drive a 1/2 ohm load (not a joke, unfortunately). Magnetic-planar sound quality is usually midway between a good dynamic and an electrostatic, with a significant freedom from cabinet colorations at mid and high frequencies. Radiation pattern is similar to an electrostatic, which means problems with imaging and bass, with the minimal amplifier
damping making some rooms unusable. This type of speaker is probably the most room-sensitive of any type, requiring the listener to choose between smooth bass and accurate imaging in some rooms. These kinds of speakers aren't my cup of tea, but I know many people who really enjoy the neutral, relaxed type of sound they can offer. In addition, a true ribbon offers some of the best treble around, superior to dynamics or electrostats, exceeded only by the "massless" exotics. * Hybrids Each type has a quite distinctive sound, and some, naturally, are hybrids. This gets pretty tricky when the dispersion patterns are different, along with different IM distortion spectra. It helps if the original designer is the one making the decisions, because a consistent philosophy is then being used for the entire system (we hope). Even so, if the designer is unfamiliar with the strengths and weaknesses of each type, the result can be the typical disjointed "hybrid sound", with the crossover region quite obvious. The most common blunder seems to be high-performance electrostats unhappily mated to low-performance woofers in thin-walled closed-box cabinets. At least there's a precedent for this, I guess, going right back to the AR1-W and the JansZen 130 tweeter. * "Massless" Exotics One day, I'd like to design one of these myself. The "massless" speakers fall into this category ... Ionovac, Magnat, and Plasmatronics (what a name!) They DO sound exotic, and measure the same. No resonances at all, and accurate pulse and frequency response up to 100kHz or more. Low distortion too ... like an amplifier. Actually, the "diaphragms" do have mass. But it's not much. It's the same as air, so the acoustic coupling is 1:1. Efficiency is a little difficult to state, though, since the output tube plates of the power amplifier are providing a high voltage that directly modulates a conductive gas. I remember hearing the Plasmatronics at the 1979 Winter CES, and I must say I've never heard a tweeter that even came close to that one. They darkened the room, and you could see this weird purple glow through the grill cloth
that looked for all the world like a gassy triode ... but it was the tweeter! It glowed and pulsed with the music! The rest of the speaker, though, was a pretty mundane paper-cone setup in a huge cabinet ... oh well. Even so, the Plasmatronics was a wild thing, a glimpse of the future, a SR-71 Blackbird next to a bunch of Cessna 172's. Not too surprisingly, the designer was a plasma physicist at Los Alamos Labs. Talk about being ahead of your time! This was 12 years before anyone was talking about turning swords to plowshares, or in this case, Dr. Teller's atomic toy into the next breakthrough in audio! The real-world problems? Well, if you ionize air (by using RF heating) you strip apart oxygen-2 and get oxygen-3 (ozone gas). Not very healthy, and illegal in the USA. The Plasmatronics ionized helium gas, which got around that problem, but it required a fresh tank of helium every month (I'm serious!). I remember seeing a full-size helium tank, gauges and all, in a special compartment inside the subwoofer enclosure. * Summary In Part 1, I covered the basics of the art of speaker design from a personal perspective, briefly illuminating the past, the future, and the present, and went on to describe the many camps that now exist in the ranks of speaker designers and manufacturers. In Part 2, I'll discuss the different qualities of dynamic drivers, the way they affect the sound of the total loudspeaker, and the effect they have on the sound quality of the listener's high-fidelity system. PART 3 The second in a series called "The Soul of Sound" where I wear my professor hat. These articles are my *personal* view of designing high-end speakers. I expect this article to be thoroughly out of date in two years or less ... so enjoy it while it's fresh! -------------------snip------------------------Looking Over My Shoulder, Part II
by Lynn Olson * Drivers The speaker driver determines the ultimate potential of the entire loudspeaker, and plays a dominant role in the sound of the entire high-fidelity system. As mentioned in the first part of the series, there is no perfect driver at the present state of the art, and that goal is many decades away, since it requires a driver with a density equal to air, completely uniform motion at all frequencies, and no distortion of any kind. We have a long road ahead of us ... but take heart! Major advances in materials sciences are happening right now, with many improvements due to occur in this decade. I confidently expect we'll be seeing breakthroughs every 2 or 3 years at the present rate of progress. This is thanks to major advances in computer modelling of mechanical behaviour and the research conducted in the aerospace, automotive, and sports/recreation industries for lightweight, high-performance materials to replace costly and heavy traditional materials. We now have Kevlar, carbon fiber composites, and forged aluminum drivers; we can expect synthetic diamond, ultralow density aerogel-silica glasses, new types of metallic and carbon monocrystals, and new classes of composite drivers before the turn of the Millenium. * Why Drivers Sound Like That The major challenge facing the driver designer is combining uniformity of motion (rigidity) with freedom from resonance at mid and high frequencies (self-damping). This is the major tradeoff in speaker systems of all types (except the "massless" group discussed in the first part of this series). There are additional problems introduced by cavity resonances and magnetic non-linearities, which are discussed later. * Uniform Motion Rigidity means accelerations from the voice coil are accurately translated into cone or dome acceleration over the entire driver surface; this translates to ruler-flat frequency response, fast pulse risetime, low IM
distortion and a transparent, "see-through" quality to the sound. Audiophiles usually describe this type of sound as "fast", much to the dismay of measurement-oriented objectivist engineers. "How can can a mid or woofer possibly be fast, since the crossover limits the pulse risetime to a fifth, or tenth, of what any tweeter can do?" This quickly leads to what diplomats call a "full and frank exchange of views", in other words, a mutual exchange of misunderstandings. As usual, both sides are right, and both sides are wrong. They're just speaking about different things. What the audiophile is actually hearing is uniform cone motion; this phenomenon can be measured by the absence of IM distortion, a flat frequency response in the working range, and good pulse response with a clean and quick decay signature. Well, that's great, you might think, just make the cone, or dome, or whatever as rigid as possible. How about a metal, like bronze, perhaps? That's nice and strong, and it can be formed into nearly any shape. You can see the direction this is taking. Bells are made out of bronze. Another problem raises its head .... resonance! After all, why does a bell, or any other rigid metal, ring so long, for many thousands of cycles? The answer has two parts, one obvious, one not so obvious. First, the metal is rigid, and formed in a shape that increases the rigidity even further. Second, the only path for the bell to release mechanical energy is to the air itself, which takes a long, long time, since the density of air and bronze are quite different, resulting in very weak coupling, and very little damping by the air load. This leads us to another desirable property for the speaker driver, which is ... * Self-Damping We also want the voice coil to stop the cone or dome, not have the cone or dome play a tune all by themselves. Unfortunately, the most rigid materials (traditionally metals) have very little self-damping, resulting in vibrations of very long duration (high Q). One way to control the problem is to extend a heavy rubber surround partway down the cone, and pay a lot of attention to the damping behaviour of the spider and surround materials.
At the present, though, even the best Kevlar, carbon-fiber, or aluminum designs show at least one high-Q peak at the top of the working range, requiring a sharp crossover, a notch filter, or preferably both to control the peak. Unfortunately, this peak usually falls in a region between 3 and 5 kHz, right where the ear is most sensitive to resonant coloration. Self-damping results in an absence of coloration, as well as contributing to a relaxed, natural, and unfatiguing quality. Interestingly, many audiophiles (and reviewers, too!) are unaware of the particular sound of driver-material resonance, calling it "amplifier sensitivity", "room sensitivity", or similar term that seems to point away from the loudspeaker itself. There are highly-reviewed (by the large-circulation "underground" magazines) 2-way speakers that use 7" Kevlar drivers crossed over to metal-dome tweeters. Technically, these loudspeakers operate with uniform motion over the range of both drivers; in practice, though, the crossovers are hard pressed to remove all of the energy from the Kevlar breakup region between 3 and 5 kHz. The reviews of these particular 2-way speakers go on at considerable, and amusing, length about the trials in finding an amplifier that "revealed" the full quality of the loudspeaker. In reality, the reviewer was forced to use an amplifier that was particularly free of coloration in the region where the Kevlar driver was breaking up. Since most audiophiles and reviewers are unfamilar with the direct sound (and measurements) of commonly-used raw drivers, they can't evaluate how much "Kevlar sound", or "aluminum sound", remains as a residue in the finished design. This is a problem, by the way, that plagues all current 2-way Kevlar, metal, or carbon-fiber loudspeakers ... at the current state of the art, the 6.5" or 7" drivers are forced to operate right up to the edge of their working ranges in order to meet the tweeter in a moderate-distortion frequency range. If you lower the crossover frequency, tweeter IM distortion skyrockets, resulting in raspy, distorted high frequencies at mid-to-high listening levels; if you raise the crossover frequency, the Kevlar breakup creeps in, resulting in a forward, aggressive sound at moderate listening levels, and complete breakup at high levels (unlike paper cones, Kevlar, metal, and
carbon fibers do not go into gradual breakup). This presents the designer with a tough choice: rough sound in the entire treble region, or the characteristic Kevlar forwardness, which can at times actually give a snarly sound to the speaker system. At the present, the best choice is a fourth-order (24dB/Oct.) crossover with a sharp notch tuned to the Kevlar resonance. I should add, by the way, that I like Kevlar and carbon-fiber drivers very much ... but they are difficult drivers to work with, with strong resonant signatures that must be controlled acoustically and electrically. As mentioned above, rigid cones have advantages, but are difficult to damp completely. A different approach is to use a cone material that is made from a highly lossy material (traditionally, this was plastic-doped paper, but this has been supplanted by polypropylene in most modern loudspeakers). The cone then damps itself, progressively losing energy as the impulse from the voice coil spreads outwards across the cone surface. The choice of spider and surround are then much less critical. This type of material typically measures quite flat and also allows a simple 6dB/Octave crossover; personally, though, I don't care for the sound of most polypropylene drivers, finding them rather vague and blurry-sounding at low-to-medium listening levels. Without access to a B&K swept IM distortion analyzer, I have to resort to guesswork, but I strongly suspect that this type of cone has fairly high IM distortion since it is quite soft. In addition, it is quite difficult to make a material that has perfectly linear mechanical attenuation; in practice, distortion creeps in when you actually want a progressive attenuation of energy over the surface of the cone. I think what is really happening is similar to many soft-dome tweeters; the cone is actually breaking up throughout the entire frequency range, but the heavy damping hides this from the instrumentation (but not the ear). To overcome this subjective effect, the best drivers of this type (Dynaudio, Scan-Speak, and Vifa) are actually composites, adding silica, talc, or metal dust to the plastic, which significantly improve rigidity without losing the characteristic polypropylene smoothness. * Cavity Resonances
Even though the dust cap in a mid/woofer (or the dome in a tweeter) looks pretty harmless, the space between dustcap and the polepiece of the magnet creates a small resonant cavity. One example of this was the (in)famous KEF B110 Bextrene midbass driver dating from the early Seventies (as used in the BBC LS 3/5a). Although this driver was probably the one of the first high-quality midranges available, it also had a number of problems, such as low efficiency, limited power-handling, a broad one-octave peak centered at 1.5 kHz (corrected by the crossover), and group of 3 very high-Q peaks centered around 4.5 kHz (only slightly attentuated by the BBC third-order crossover). These upper peaks, which reviewers mistakenly ascribed to the tweeter, were also very directional, which is typical of dustcap resonances. The popular tweeters of the 1970's, including the Audax and Peerless 1" soft-domes, also had similar resonances between 9 and 16 kHz, which were partially damped by a felt pad nearly filling the space between the dome and the polepiece. Since the soft-domes were much more lossy than the stiff B110 dustcap, the resonances were much broader and only 1 to 3 dB in magnitude ... but they were still there, and they were responsible for some of the fatiguing quality noticed by attentive listeners. Not surprisingly, the problems were much worse in the phenolic, fiberglass, and hard paper domes used in the more mundane speakers of the day. (Ah yes ... who can remember such paragons of excellence as the BIC Venturis? The Cerwin-Vegas? The Rectilinears? The JBL L100's? In a prior life, I actually had to sell these awful things! "Wait'll you hear 'Dark Side of the Moon' on these babies!") Returning to the present, the best midbass and tweeter drivers now sidestep this problem in two ways: a vented polepiece assembly, used by the Scandinavian manufacturers Dynaudio, Scan-Speak, Vifa, and Seas; and a bullet-like extension of the polepiece, which replaces the midbass dustcap entirely, used by the French manufacturers Audax and Focal. The Dynaudio Esotec D-260, Esotar T-330D, and Scan-Speak D2905/9000 tweeters are the most notable examples of using a vented polepiece loading into a tiny transmission-line to damp the backwave from the tweeter dome;
their use in the Sonus Faber Extrema and ProAc Response Threes is a comment on how successful this technique can be. By contrast, the Focal T120 and T120K, which use a rigid fiberglass or Kevlar inverted dome directly above an undamped polepiece cavity, show a series of high-Q resonant peaks at the top of their operating range, which are caused by the resonant cavity coupling to the first breakup region of the rigid dome. I must admit I was rather puzzled by the public acclaim for these drivers when they first came out; I didn't like the way they sounded, and I wasn't too impressed by their measurements. >From all accounts, though, the new Focal titanium-dome T120Ti and titanium-dioxide T122Ti-O2 are excellent, and I liked what I heard when I auditioned a speaker that used the Focal T120Ti at a recent Triode Society meeting. Magnetic Non-linearities Most audiophiles are aware that loudspeaker drivers are inductive; after all, the voice coil is wound around a ferrous polepiece, and that's how you make an iron-core inductor (or "choke"). Not as many audiophiles know about the myriad of problems this creates. If the inductance were constant, like an air-core inductor, there would be no problem; just adjust the crossover design to allow for it (using a simple R-C network) and off you go. Unfortunately, this is an iron-core inductor, and much worse, the inductance varies with the position of the voice coil. The varying inductance has profound consequences, since the inductance is actually a important factor in determing the upper rolloff frequency of the driver, and its resulting acoustic delay (relative to the tweeter). Vary this inductance, and the rolloff frequency and acoustic delay move along with it. When does this happen? Whenever the driver moves a significant proportion of the linear region of voice-coil travel, which is less than you'd think. In the excellent 8" Vifa P21W0-12-08, this linear region is only 8 mm (plus/minus 4 mm either way). A more typical figure for linear travel would be 6 mm for most 8" drivers, and 1 to 3 mm for most midranges.
Play some deep bass, and the effects of inductance modulation begin to show, creating IM and FM distortion over the entire frequency spectrum. This is a genuine problem for 2-way systems and 3-way systems using a low midrange crossover; it means that any time you can actually see the drivers move, there are quite significant amounts of IM and FM distortion. What does this sound like? You can expect a loss of resolution that is depends on the bass content of the program material, which may be masked by problems in the power amplifier (such as output transformer saturation or inadequate power supplies). Are there solutions? Yes. The best drivers from Scan-Speak (SD System) and Dynaudio (DTL-System) plate the polepiece with copper to short out eddy currents induced within the magnet structure by the voice coil. The specification that gives this away is the voice coil inductance. The 8" Scan-Speak 21W/8554, probably one of the best 8" drivers in the world, has an inductance of 0.1mH, which is far lower than the 8" Vifa P21W0-20-08, which has in inductance of 0.9mH. Both are excellent drivers; the Scan-Speak, though, is almost certainly going to have more transparent sound when asked to carry bass and midrange at the same time. The inductance figure also has a another hidden meaning; remember, the upper rolloff frequency of the driver is the combined function of the mechanical rolloff and self-inductance of the voice coil. If you calculate the electrical rolloff frequency by using the VC inductance and the DC resistance, a few drivers have an electrical rolloff well above the measured acoustical rolloff. This is desirable; it means that the interaction between the two rolloff mechanisms is going to be small. Other drivers (and this is true of most drivers) are going to have an electrical rolloff well below the measured acoustical rolloff. How is this possible? The mechanical system actually has a broad peak which is masked by the self-inductance of the voice coil. This is not good; any change in either the mechanical system or the electrical system is going to strongly modulate the frequency and transient response. This, by the way, is the same kind of problem found in the old moving-magnet phono cartridges. Most moving-magnets (typically Shure and Stanton) were mechanically peaked, then rolled-off electrically by the
combination of cable capacitance and cartridge inductance. Not surprisingly, this type of cartridge usually sounded much less transparent than its high-end moving-coil brethren, which had less than one-tenth the inductance and a much flatter, more accurate mechanical system. In the section that follows, I'll show you how you can make your own decisions about which drivers you like (and second-guess the manufacturer, reviewer, and your audio friends). * Selecting A Driver I use a method that's so crude it might sound kind of dumb; I put the driver on large, IEC-sized baffle (135cm by 85 cm) and listen to it. No crossover, no enclosure, and if it's a tweeter, not very loud at all. I listen to pink noise (to assess the severity of the peaks that may appear in the sine-wave and FFT waterfall measurements) and music (to get a sense of how much potential resolution the driver posseses). This does take an educated ear, though, since you have to listen around the peaks that the crossover might notch out, and not hold the restricted bandwidth against it. However, this listening process tells you a lot about how complex the crossover has to be, particularly if you remember that the crossover can never totally remove a resonance ... it can just make it a lot more tolerable. In addition, I very carefully assess the results of the MLSSA PC-based measurement system (using the same IEC baffle), looking at the: 1) Impulse Response. (How fast does it settle to zero? Is there chaotic hash in the decay region or is it a single, smooth resonance? Are there two or more resonances?) 2) The Group Delay vs. Frequency Response. (How ragged is the frequency range above the first breakup? Can it be fixed in the crossover?) 3) The Waterfall Cumulative Decay Spectrum. (Can I accept the resonances that can't be fixed in the crossover? If crossover correction is required, how complex is it going to be?) 4) The flatness of Frequency Response in the working band. (Can I accept
the broad, low-level colorations that may appear here?) Listening and measurements are equally essential. Both give only a partial picture of the actual driver. Even the finest modern audiophile system will have very serious sonic deficits 5 years from now; measurements provide a reality-check on colorations that present-day equipment may not reveal. In turn, the MLSSA system can point out troublesome colorations to listen for; some are much more audible than others. The thoughtful designer is obliged to be as careful as the craftsman (or craftswoman) who lavishes a full measure of care and attention on even the hidden parts of their creation. ------------------snap------------------This is a revision of an article published in Vol. 4-2 of Positive Feedback magazine, a publication of the nonprofit Oregon Triode Society located at: 4106 NE Glisan Portland, OR, 97232 USA For reprints, back issues, T-shirts, or subscription information, call John Pearsall at: (503) 234-4155 ->From Lynn T. Olson PART 4 The second in a series called "The Soul of Sound" where I wear my professor hat. These articles are my *personal* view of designing high-end speakers. I expect this article to be thoroughly out of date in two years or less ... so enjoy it while it's fresh! -------------------snip------------------------* Types of Drivers It helps when you start listening and comparing to have a good grasp on the basic characteristics of the driver, so you can determine if it is a good example of its type. By listening carefully and examining all of the relevant specifications, you can find out just how well the driver
designers solved the problems of making a good driver. * Paper Cone Drivers This dates back to the original Rice & Kellogg patent in the late Twenties. Paper ranges in quality from the worst TopTone clock/radio speakers to the superb Scan-Speak 5" cone/dome midrange used in the Theil line of speakers and the SEAS 6.5" midbass used in the Wilson Audio WATT. This oldest of materials is actually a composite structure, and changes properties significantly when it doped with an appropriate plastic (the choice of doping is invariably a trade secret of the driver vendor). The doping is quite important, since paper undergoes significant alterations with changes in humidity and time if it is left undoped; the doping stabilizes the material and typically improves the self-damping. Strengths are: Good-to-excellent self-damping, potentially excellent resolution and detail, very flat response potential, and a gradual onset of cone breakup. It can be used with low slope linear-phase crossovers without much trouble. Paper is a material that sounds better than it usually measures ... this is an asset, not a disadvantage. Weaknesses are: Not as rigid as the Kevlars, carbon fibers, and metals, so it lacks the last measure of electrostatic-like inner detail. Doesn't go as loud as the materials above, but the onset of breakup is much more gradual. Paper-cone drivers usually require modest shelving equalization in the crossover for best results. Paper is not as consistent as synthetics, so pair-matching isn't quite as exact, which may affect imaging, depending on the precision and quality of manufacture. Properties may change with time, even with doping. Best Examples are: Scan-Speak 8640 5" cone/dome midrange, with linear response out to 13kHz, very low distortion, excellent pulse response, and excellent inner detail. SEAS 6.5" midbass (as used in the Wilson Audio WATT, but possibly modified). Audax PR170M0 6.5" high-efficiency midrange. (100 dB at 1 meter!)
I have heard from several sources that Kurt Mueller makes the best paper cones and surrounds, which are used by Scan-Speak, Seas, Vifa, and others, as well as many high-quality proprietary drivers made in Britain and the United States. * Bextene Cone Drivers This is an acetate plastic derived from wood pulp, not petrochemicals, and is always damped by a layer of doping material to control the strong first resonance it displays around 1.5 kHz. It was originally developed by the BBC in 1967 to replace paper with a more consistent and predictable material for monitoring purposes. It came into widespread use in the early Seventies, with the typical audiophile speaker using a 8" KEF or Audax Bextrene midbass driver with an Audax 1" soft-dome tweeter. The BBC-derived designs always employed equalization to flatten the Bextrene driver in the midband; the most famous (or infamous, depending on whether you were the listener or the designer) driver was the KEF B110 used in the BBC LS 3/5a minimonitor. Bextrene has been replaced by BBC-developed polypropylene, which gives a much flatter response, does not require a layer of doping material, and provides a 3-4 dB increase in efficiency due to the decrease in cone mass. Bextrene is now considered an obsolete material by nearly all speaker designers. Strengths are: Consistent batch-to-batch, excellent potential imaging (by mid-Seventies standards). Inner resolution higher than most paper cones. Weaknesses are: Very low efficiency (85 dB at 1 meter), requires a strong notch filter in the midband, "quacky" coloration by modern standards, sudden, unpleasant onset of breakup at not-so-high levels, and numerous resonances at the top of the working band. Best Examples are: None. Modern designers are not willing to tolerate the complex notching and shelving equalization required to make these drivers acceptable. * Soft-Dome Tweeters
These came into common use in the early Seventies with the introduction of the Peerless 1" soft-dome (remember the tweeters of the original Polk speakers?), followed by the superior Audax 1" tweeter, which found its way into many British and American designs during the Seventies and early Eighties. These designs fell into disfavor with the introduction of titanium and aluminum domes and the Focal inverted-fiberglass domes in the mid-Eighties, which swept the Audax-class soft-dome drivers off the audiophile market. In the last two years, the soft-domes have made a surprising comeback with the introduction of the Dynaudio Esotec D-260, Esotar T-330D, and Scan-Speak D2905/9000 1" tweeters, which compete on even terms with any metal-dome around. These new designs combine sophisticated transmission-line back-loading with new dome profiles and new coating materials. As a result, they have the sonic resolution and detail of the best metal domes without the characteristic 22 to 27 kHz resonance. Strengths are: Intrinsic self-damping and potential for extremely flat response and first-class impulse response. Potential for natural, open sound without intrusive and fatiguing resonances, a valuable quality when listening to many digital recordings. Weaknesses are: The older class of soft-domes had a dull sound with a hard-to-pin-down fatiguing quality. Many had quite limited power-handling and required a high-slope 18 dB/Octave crossover to minimize IM distortion. The high-profile dome, required for rigidity, has more restricted HF dispersion than the competing metal-domes, which have flatter profiles. The new units mentioned above do not have these faults, however, with the exception of dispersion. Best Examples are: The Dynaudio Esotec D-260, Esotar T-330D, and Scan-Speak D2905/9000 1" tweeters. * Soft-Dome Midranges These things are dogs! I've listened to the AR-3, AR LST, ADS systems, Audax 2", and the Dynaudio D-52 soft dome midranges, and they barked, they snarled, they chewed the rug, and made a mess on any loudspeaker they
approached! They measure flat, all right, but they sound opaque, fatiguing, strongly colored, and 2-dimensional. The first problem is that soft-dome midranges have a limited bandwidth resulting from restricted linear excursion (1-2 mm typical) and do not gracefully tolerate even a 500 Hz crossover, operating best over a limited 800 Hz to 3200 Hz range. A second problem is that they are quite prone to side-to-side rocking modes, since there is no spider combined with the surround to force the movement into a linear back-and-forth motion. A third problem is that the doped silk dome is just, well, too soft for the job it has to do in the power band of the midrange. The newer class of cone-domes, such as the 5" Scan-Speak 13M/8636 and 13M/8640, and the 5" Dynaudio 15W-75, are another story. These 3 drivers are actually constructed as high-precision cone drivers, not midrange domes. The only thing they have in common with the soft domes is a large dustcap, which does act as a dome at high frequencies. This class of driver has much more excursion, much lower distortion, and much wider frequency response than the older soft-dome midranges. The cone-dome drivers are capable of realistic and transparent sound. They are described in more detail in the other sections, since they use Kevlar, paper, and polypropylene respectively. Another "special case" is the professional-grade ATC 3" dome with an integral short horn. This driver uses a dual spider to eliminate the rocking problem that plagues most soft-domes, reducing the IM distortion very significantly. Ron Nelson (of Nelson-Reed) recommended this driver as one of the very best midranges around, and I take his recommendation seriously. This is a very expensive driver (around US$300/each) and needs to be hand-selected so the resonant frequencies of the left and right channels match. Strengths are: None. Metal-dome midranges have some potential, but they require sharp crossovers on both ends with an additional sharp notch filter at high frequencies to remove the first (and worst) HF breakup mode. Note: This does not apply to the ATC driver or the cone-domes.
Weaknesses are: High distortion, fatiguing sound, high crossover frequency, limited bandwidth, limited power-handling, and misleading frequency response measurements. It takes a detailed swept IM distortion measurement and laser holography to get the goods on these drivers. Note: This does not apply to the ATC driver or the cone-domes. Best Examples are: ATC 3" professional-series - a totally different animal than the usual soft-domes. About 4 times as expensive, though. The Dynaudio D-54 in a Edgarhorn is reputed by the "Sound Practices" group to be the finest midrange in the world ... yours truly has a pair on order, so I'll let you know. * Polypropylene Drivers This material was developed by the BBC in 1976 (my dates may be off) as a replacement for Bextrene. Since it is intrinsically highly self-damping, a correctly designed polypropylene driver is capable of flat response over its working range without equalization. In addition, it typically attains efficiencies of 88 to 91 dB at 1 meter, which is a significant improvement over Bextene. This material has become nearly universal, since it requires a minimum of hand treatment to assemble a loudspeaker - the only difficult problem was finding adhesives that would stick to polypropylene, and that problem was solved in the beginning of the Eighties. This material is used in speakers ranging from mundane rack-stereo trash to the first-rank ProAc Response Threes and Hales System Two Signatures. The cone profile and additional materials added to the polypropylene mix strongly determine the ultimate quality of this type of driver. Strengths are: Very flat response if correctly designed, very low coloration, good impulse response, crossover can be as simple as one capacitor for the tweeter, good efficiency, and gradual onset of cone breakup. The best examples can be as transparent as the best paper-cone drivers, which is an excellent standard. Weaknesses are: Not quite up to the standard of transparency set by the rigid-cone class of drivers and the planar electrostatics. Many poly
midbass units do not mate well with the popular metal-dome tweeters, with differences in resolution that can be audible to the skilled listener. Not the best choice for woofers 10 inches or larger unless the polypropylene is quite thick and reinforced with another, more rigid, material. Large woofers are better served with stiff paper or carbon fiber. Best Examples are: The Scan-Speak 18W/8543 7" midbass, as used in the ProAc Response Threes, is probably the finest polypropylene driver in the world. Another closely-ranked contender is the Dynaudio 17W-75 Ext. 7" midbass, as used in the Hales System Two Signature. The Vifa P13WH-00-08 5.5" midbass/midrange unit is another excellent performer, well suited for midrange or minimonitor use. It is unique in having a textbook-flat midrange combined with a completely smooth 2nd-order rolloff. * Metal-Dome Tweeters Advances in German metallurgy (at Elac and MB) resulted in thin profile titanium and aluminum domes in the mid-Eighties, with drivers from several vendors in Germany, Norway, and France now available. This type of driver can offer very transparent sound, rivaling the best electrostatics if correctly designed. The downside is the lack of self-damping, with aluminum coming ahead of titanium in being better behaved in the ultrasonic region. At the present, all metal-dome drivers have significant ultrasonic peaks, ranging in magnitude from 3 dB (excellent) to 12 dB (not so good). There's controversy about the significance of this peak, since the all-wise, all-knowing, all-seeing potentates of Philips and Sony have gone to great lengths to make sure that none of our wonderful new "perfect sound forever" recordings will ever have any musical information above 20kHz. Maybe one day we'll finally have a digital system designed with high-fidelity in mind ... something with an upper limit of at least 32 kHz and a true resolution of 20 to 24 bits. Strengths are: Uniform piston action right up to the HF resonance, providing sound of very high resolution, transparency, and immediacy if
correctly designed. Dispersion is typically excellent, since the metal domes have flatter profiles than soft-domes. Weaknesses are: Potential for (I hate to say it) "metallic" coloration caused by the HF peak intermodulating with the inband sound. Some early designs have restricted power-handling. If significantly overloaded, breakup distortion is obvious over the whole frequency range. Best Examples are: Vifa D25AG-35-06 1" aluminum dome, which is even better with the plastic phase disk removed. This dome has a vented pole piece, so power handling is quite good, and the ultrasonic peak is only about 3 dB even with the phase disk removed. Also, the brand- new Focal T122Ti-O2 is reputed to be outstanding. * Rigid Drivers Aluminum The first rigid drivers to find limited use in high-fidelity applications were the small Jordan Watts 4" aluminum-cone units. The hand manufacture, high price, and low efficiency limited the market for these drivers, and very few appeared in the United States (I have heard of them by reputation but have never auditioned them personally). Some new British speakers are also using 5" and 7" aluminum-cone midbass drivers, which tend to have very low efficiency and require a notch filter in the crossover. Expanded-Foam The next generation were the expanded-foam bass units, with the KEF B139 being the most famous examples. This class of driver offered piston-band operation through the midbass, but suffered from very low efficiency, limited power-handling, and severe high-Q resonances in the midband. (It was not generally known that the B139 had a 12dB peak at 1100Hz with a very high Q. Many reviewers blamed the midrange for problems that were actually caused by the B139 not having a notch filter in the crossover.) They were quite popular in 3 and 4-way transmission-line systems in Britain (IMF) and the United States (Audionics) in the Seventies. The Audio Professor Digresses
I remember working with the KEF B139, B110, Richard Allen 7", and front and rear KEF T27's in my first commercial design, the Audionics TLM-200 4-way system. My baptism into the mysterious arts of speaker design went as follows: Charlie, my boss: "Hey Lynn! You remember what Laurie Fincham was talking about when we visited KEF last year? All that stuff about impedance correction and frequency response target functions?" Yours Truly: "Well, I didn't write it down, but I remember some of it." Charlie: "Great! You can do the crossover for this!" ... pointing at a massive six-foot-tall loudspeaker with the 4 aforementioned drivers. The previous designer had left town without warning, leaving Audionics with a monster transmission-line complete with drivers and no crossover.True story, folks. Carbon-Fiber (The Audio Professor Returns to the Topic) The next generation were the Japanese carbon-fiber units, which made their first appearance in the pro studio monitor (prosound) 12" TAD units with very high efficiencies and very high prices (around $300 each in 1980). Carbon fiber prices have now dropped, and Vifa and Audax make good examples of this type of driver. The Japanese make lots more of them, having pioneered the technology, but they have been difficult to obtain if you are a non-Japanese small-run specialist manufacturer. These drivers have true piston action, outstanding bass and midbass response (the best I have ever heard), but also have nasty, chaotic breakup modes at the top of their range. Removing these breakup modes requires a sharp slope and one or two very sharp notch filters (this type of driver and filter is used in the top-of-the-line Linaeum speaker). Although I very much dislike drivers that require filters as complex as this (after doing the TLM-200, I vowed I would *never again* design a 57-component crossover), I must admit the Vifa 8" and 10" carbon fiber woofers are the only direct radiators where I've actually felt, not heard, tactile bass.
Kevlar Kevlar drivers made their appearance in the mid-Eighties with the French Focal and German Eton lines, with the Eton having superior damping due to the higher-loss Nomex honeycomb structure separating the front and rear Kevlar layers. The Eton and much newer Scan-Speak Kevlar drivers now share the limelight as the worlds pre-eminent high-tech drivers. A unique and quite desirable property of the latest Scan-Speak Kevlar drivers is a smooth rolloff region above the usual Kevlar peak. All of the other Kevlar drivers (that I have measured and listened to) have chaotic breakup regions; the Scan-Speaks are the only ones that appear well-controlled in this region, which is certain to provide a significant improvement in smoothness and transparency compared to other types. Composite Audax has made a surprise reappearance in the high-end market with an unusual composite technology, called HD-A. This is an acrylic gel containing a controlled mix of grain-aligned carbon-fiber and Kevlar fibers. The initial factory measurements show a striking combination of piston-band behaviour combined with minimal high-frequency peaking and a smooth rolloff above that point. The Future Evaporated diamond coatings are now available at low cost with a recent Russian breakthrough (cheap enough to coat floppy disks!), and I very much hope that Scan-Speak and others pick up this technology quickly. As I said earlier, things are changing fast. * Overall Strengths & Weaknesses of Rigid-Class Drivers Strengths are: Best available transparency, imaging, and depth presentation of any type, equalling or exceeding electrostats if carefully designed. High efficiency, high peak levels, and very low IM distortion in the best examples. This class of drivers is considered at the state of the art by many designers, and this field is expected to advance quite rapidly as material technology advances.
Weaknesses are: Older designs have severe peaking at the top of the working band, and nearly all have a uncontrolled chaotic breakup region above the high-frequency peak. This would cause fatiguing sound over the long run and a compression of depth perspective and "air". Any loudspeaker that does not use a correctly designed notch filter with a Kevlar or carbon-fiber driver can be considered faulty; since the HF peak does not lend itself to correction with a conventional low-pass filter, and will be quite obvious to any listener familiar with the sound of an unequalized Kevlar or carbon-fiber driver. Unpeaked rigid drivers are not currently available ... stay tuned, this will probably change in less than a year. Although these types play quite loudly, the onset of breakup can be quite sudden and unpleasant, akin to clipping in a amplifier. Some Kevlar and carbon-fiber drivers require an extremely long break-in period (>100 hours) to soften the fibers in the cone and the spider; this is a fault, since it indicates the materials may not be mechanically stable with extended use. Best Examples are: The Scan-Speak 13M/8636 5" midrange, 18W/8544 7" midbass, and 21W/8554 8" bass drivers. These are the only Kevlar drivers that have reasonably well-behaved rolloff regions above the high-frequency peak. The Scan-Speak drivers also have vented pole-pieces that are copper-coated, reducing inductive types of IM distortion by tenfold or more. These drivers are probably at the pinnacle of rigid-driver technology (in Spring of 1993 at least). The Audax HM130Z0 5.25" midrange, HM170Z0 6.5" midbass, and HM210Z0 8" bass drivers in the HD-A series also look very interesting. The German Etons also bear close watching, since the manufacturer is continuing to work on techniques to maintain the cone rigidity while improving the self-damping characteristics. * More Coming Attractions The next two articles that follow aren't on the Net, unfortunately, since
they make extensive reference to construction drawings of the cabinet, suggested placement in the room, and lots of MLSSA measurements. The two articles (The Soul of Sound, Part III and IV) describe an efficient (92dB @ 1 meter) 2-way transmission line suitable for use with triode amplifiers of modest power (8 to 22 watts work just fine). Look for these articles in Vol. 4-4 and 5-1 of Positive Feedback magazine - back issues are still available, as far as I know. If I can convert the Macintosh PICT artwork to something that will survive the uuencoding process and figure out an appropriate FTP site, I may post these in the future. In the meantime, if you good folks on the Net are interested, I could post my review of the Reichert 300B and the Ongaku in this forum. (The review articles are entitled "The Outer Limits" and are *not* the Authorized Version from StereoPriest, Absolution Sound, or the Audio Cleric.) ------------------snap------------------This is a revision of an article published in Vol. 4-2 of Positive Feedback magazine, a publication of the nonprofit Oregon Triode Society located at: 4106 NE Glisan Portland, OR, 97232 USA For reprints, back issues, T-shirts, or subscription information, call John Pearsall at: (503) 234-4155 ->From Lynn T. Olson A series of meditations and thoughts on the differences between tube and transistor amps Yah sure, you betcha, it's time for another Positive Feedback article ... a preview from the next issue, Vol. 5, Issue 2. This is a continuation of the "Soul of Sound" series on the art of high-end audio design. Readers who are collecting the entire series might notice the absence of Part III and Part IV ... well, those two parts are the construction articles for the Ariel high-efficiency transmission-line loudspeaker, and I'm still working on the politically correct way to transmit the rather large graphics files over the Net (I may set up an FTP site here at
Teleport). If you just can't wait to find out about the Ariels, you're probably better off subscribing to the magazine, which is chock-full of tweaks, reviews, construction articles, and off-the-wall pieces of all kinds. ------------------------The Soul of Sound, Part V by Lynn Olson This month's column is a collection of meditations on the art of audio design. They arise from a series of questions from fellow OTS members and e-mail that appeared in my Internet mailbox - which is:
[email protected] For those who haven't followed these electronic discussions, here they are in print. I also have a bonus feature at the end of the column, a newly-arrived little brother for the Ariel. *Frequently Asked Questions* Q: Why do tubes sound different from transistors? Is this due to "euphonic" second harmonic distortion? I don't think "euphonic distortion" exists. Granted, some sonic flaws are a lot more tolerable than others, but I've never found any class of distortion or signal bending that makes the sound more transparent, more real, and more lifelike. People have been making gizmos like Aphex Aural Exciters, delay circuits, and "T-function shaping" for decades now, and all they really do is make music sound more like commercial FM radio. To my ears, at least, when I remove a known problem (such as a resonant mode in a cabinet, a standing wave on a driver cone, a problem in a power supply, or improve the linearity of an amplifying element) it pretty much always sounds better ... more real, more truthful, more expressive, etc. In short, fixing problems makes it sound truer to life. So why do tubes typically, but not always, sound better than transistor
circuits, despite worse overall distortion? The answer doesn't lie in circuit simplicity, since if that were true, first-generation transistor units like the Dyna PAT-4 and the Marantz 7T preamps would be audio classics. Well, they're not, and a quick listen on even a half-decent system tells the tale. I don't like measurements that only apply to the entire "black box", be it a speaker, an amp, a CD player, or whatever. They tend to mislead both the reviewer and the buyer. The measurements that tend to be the most meaningful are ones you can make on elements of the black box ... an active device, a single driver, the way the crossover behaves far off-axis, a single reflection at a cabinet edge, the frequency vs. source impedance of a power supply, etc. In other words, measurements are primarily useful as a tool for the designer to isolate problems, not as a go-nogo signal for the reviewer or buyer. The reason is pretty simple: when you work at the component level, you can actually isolate cause-and-effect, and deal directly with the cause of the problem. When you get the vector sum of all of the stuff in the black box, the measurements get a lot less useful, unless the designer really made a big mistake. Case in point: I've lived with the Audio Note Ongaku SE211, as well as the Kassai PSE 300B and the Reichert SE 300B. I've also had access to my trusty Audionics CC-2 (not a bad transistor amp), a modern multi-kbuck Class A transistor unit, and a souped-up Dyna Stereo-70. They all sounded different, particularly to non-audiophile friends. The Ongaku, by far, had the worst THD and power measurements ... 22W at 3% distortion. It also made the Ariel sound better than any electrostat I've ever heard ... in fact, the best sound I'd heard in many years. It certainly sounded better than anything I heard at the 1994 Winter CES. So what's going on here? Maybe THD is simply measuring the wrong thing. I have a hypothesis I want to explore. Any of you who have access to serious measuring equipment are invited to join in. The hypothesis is based on a very simple premise: If it sounds much clearer, more natural, more true-to-life, that implies a problem (or type of distortion) has been removed. (Even if the problem
cannot be measured with present-day equipment.) I am hypothesizing (bear with me here) that the problem with analog transistor devices is actually non-linear Miller capacitance. You see, bipolar transistors, MOSFETS, and even diodes exhibit very significant changes in capacitance with applied voltage (possibly current and temperature as well). As capacitors go, it is my understanding that transistors are very poor quality, worse even than electrolytic capacitors. To test this, of course, you'd want to bias the device and measure the base-collector capacitance with a high-quality bridge that measures DF and DA. Even though the capacitance is only in the pF region, it is Miller-amplified by the gain of the transistor. This means any capacitance non-linearities are multiplied by the forward gain, which is also non-linear. So we have two non-linearities multiplied by each other, and worse, one is in the time domain, and constitutes a type of FM modulation of the audio signal. Now, this effect is probably very small, and hard to measure as well. Nevertheless, Walt Jung successfully identified very small problems related to the DF and DA of capacitors more than a decade ago, and it is generally agreed now in the upper reaches of prosound and audiophile engineering that the choice of dielectric is very important in speakers and electronics. I'm just taking the same thing and applying to it to active devices, and throwing in the implications of Miller-amplification of DA and DF problems. Note that tubes are very different in this respect. As capacitors go, the dielectrics are basically a little mica, glass, and a high vacuum ... pretty close to theoretically perfect. In addition, this capacitance is unmodified by voltage, current, temperature, or even if the tube is stone cold and completely inactive. In a working circuit, of course, the capacitance itself is still Miller-amplified, and yes, the tube is only somewhat more linear than its silicon brother, but the non-linear part of the capacitance basically drops out of the picture. Well, I agree, this is all pretty esoteric, but non-linear Miller
capacitance is one area where the tube and the transistor are very different indeed. Of course, none of this shows up in a THD measurement, which is why I feel that the search for the ever-lower THD spec has led the audio industry in the wrong direction over many years, even decades. What we're looking for in the 21st Century may have very little to do with THD ... I suspect that more meaningful measurements are awaiting invention, and don't yet exist. Q: What about crossovers and measurement techniques? How can I design my own crossovers? A: This isn't going to be the answer you want to hear. Using an unmeasured "textbook" crossover is certain to produce poor results ... nowhere close to the potential of the driver. Why? Drivers are nowhere close to a purely resistive load and do not have a flat response ... both of which are requirements for a "textbook" crossover. Using a stock crossover will produce errors from 3 to 20dB, as well as wildly variable inter-driver phase characteristic, resulting in nulls that sweep through the listening position. Results? Grossly non-hifi sound. Don't do it. Unfortunately, a FFT (or similar) measurement system is pretty much the minimum these days, although some designers still swear by (or at) 1/3 octave pink-noise systems. The major benefit of FFT-class systems is that they allow examination of frequency, phase, and group delay responses, as well providing the frequency resolution required for automatic crossover optimizers like LEAP, XOPT, CALSOD, etc. Of course, all designers have their favorite measurement systems, but I don't know of any well-regarded speakers that are designed by ear without benefit of instruments ... the days of Bob Fulton are long gone in the speaker business. The minimum system is probably the IMP FFT board published in Speaker Builder magazine. The IMP and the Mitey Mike are about $400-500 from Old Colony. Be aware that pure FFT systems have severe problems with signal-to-noise ratio, requiring a quiet room and an amp that can throw big pulses at the speaker. Old Colony Sound Lab is at: (603) 924-6371. The next step up is the LMS from LinearX, made right here in Portland USA (I have no affiliation with LinearX ... I don't even get a discount). This is a chirp-based system with excellent frequency resolution and noise rejection, a very good integral mike, but no directly-measured phase or
group delay info. The LMS is used by many big-name manufacturers for production-line testing, as well as design work. If you don't need fancy waterfalls, or aren't into linear-phase designs, the LMS is a great choice, particularly since it is well-tuned for making measurements for LEAP, XOPT, CALSOD, etc, crossover optimization programs. Glenn Philips, a fellow designer in Portland, knocks out custom designs all day long using mostly LMS and LEAP. (They sound pretty good, too.) The LMS + mike cost $1000 last time I checked. LinearX is at: (503) 620-3044. I use MLSSA (so does Glenn Philips and plenty of big-name manufacturers). The MLSSA has largely displaced the Crown TEF, since it delivers pretty much the same performance at about 1/3 the price and doesn't force you to buy a reworked Kaypro as a host computer. (I've heard the new TEF unit uses a PC host and has a price comparable to MLSSA ... maybe even cheaper. I haven't kept up on this.) The MLSSA measures freq. response, phase, group delay, does waterfalls, calculates polar response, sum-and-difference curves, washes the dishes, rents the movies, etc. (Well, maybe the last two items are in the next upgrade.) MLSSA uses a pseudo-random noise sequence called a Maximum Length Sequence. The math coprocessor of the PC does a cross-correlation on the known sequence, deriving a pulse, then does a FFT on the pulse, providing the freq. response with excellent noise rejection. All of these mathematical monkeyshines happens in a fraction of a second (on a 486DX). I also use the ACO Pacific 1/2" condenser microphone, since I couldn't handle the heavy price of the B&K original. The MLSSA is a big step upward in price: $3500, and possibly $4000 by now. The ACO Pacific mike, FET amp, and charge box are $1050. The MLSSA is made by DRA Labs, but unfortunately the manual shows no phone number. I think they are in Pennsylvania and they do advertise fairly often in the AES Journal. Yes, there are instruments that cost more, with the Audio Precision coming to mind. Since I remember an all-up price of >$12,000, I haven't checked it out. I'm not sure if the AP offers true time-domain measurements, but it does have a lot of slick automated test features that lend it to fast production checking. The AP is just the thing for amps and CD's, though,
so if my little startup ever shows a profit, I'll buy one (after I pay back my wife/sweetie/partner). Now that I've thoroughly snowed many readers, I should stop and make a point. Discussions of measurement systems quickly degenerate into religious wars for many speaker designers, with distinct overtones of Windoze/Macintosh, UFO/Phil Klass, Fundie/Materialist/NewAge wrangles. Phooey on all of 'em. I use these machines to find out why the speaker sounds strange, and I don't expect the instrument to think for me. There are plenty of things that are plain as day and don't show up with any measurement technique. Likewise, there are things that are just slightly audible, yet look quite nasty on the screen. This is where experience comes in. So, to design a crossover, you must measure. Rock-bottom minimum is 1/3 octave pink-noise or the IMP board. After that, you can design using the ole cut-n-try (don't go beyond 18dB/octave) with Butterworth as a starting point, adding inductance compensation as necessary, notch filters as necessary, and bumping the Q up and down until it sounds and measures halfway right. Or you can save yourself a lot of trouble and optimize by computer ... but you may very well hand-tune the end result in 5% steps anyway. A point about crossover tuning: a lot of people don't realize that inter-driver phase is far more audible than actual absolute phase (that is, phase referred to the original source material). Why? The inter-driver phase characteristic controls the polar pattern above, below, and at the crossover frequency, and I can assure you that a null sweeping across the listener at a certain frequency is very audible. That's why subtle adjustments of Q (rolloff shape) are a lot more audible than you'd expect. I use a simple subjective technique to check for coherency and final tuning: I turn the speaker on its side, play good-quality pink-noise, get about 1 meter away, and listen very carefully for a smooth spread of sound between the mid and tweeter. A good crossover should actually sound like one driver ... and yes, it can be done, and must be done if you want cohesive sound with decent image quality. Q: When you design a system, where do you start?
A: In speaking with designers whose work I respect (and in looking at the ways I've done my best work), I've found three main themes, or guidelines. These "guideline" aren't rigid rules, although I may have made them sound that way; they are just straightforward techniques for designing audio equipment that meets your subjective goals. The essence of these rules is minimizing the number of variables and not making "automatic" assumptions about what's audible and what's not. Audio has a way of humbling the "know-it-alls"; this is a field that requires an open mind and an inquiring spirit. 1) Use the appropriate test equipment intelligently. This requires equipment at a professional standard; fortunately, PC's have reduced the "entry price" very significantly. For loudspeakers, the realistic minimum is the LMS IMP (as seen in Speaker Builder magazine), the ATI LMS system, the DRA Labs MLSSA, or the Crown Techron TDS system (in order of increasing price). All of these systems require a high-precision calibrated microphone (which is included in the price of the LMS system). Not good enough: General-purpose FFT systems, 1/3 octave analyzers, and warble tones used with sound level meters. These systems do not have adequate resolution to disclose narrowband resonances, are nowhere accurate enough for crossover optimizing software, and do not provide information on time-domain behavior (the FFT does, but it suffers from serious problems with signal-to-noise ratio). For amplifiers, a distortion analyzer is required that can resolve out to the 5th harmonic or higher. The Audio Precision is an obvious but quite expensive choice; secondhand Sound Technology, Hewlett-Packard, or Radford units work well, provided that the distortion residual is then fed to a spectrum analyzer or oscilloscope. (The systems mentioned above for loudspeaker analysis also work as FFT analyzers, but they must be used together with a distortion analyzer, not as stand-alones.) Not good enough: Total Harmonic Distortion analyzers with no way to discern the proportion of the harmonics. THD, by itself, is an utterly meaningless and potentially misleading number. What is much more interesting is the proportions of the harmonics, which requires an additional FFT unit connected to the analyzer output. Although an
oscilloscope is an essential piece of equipment for examine high-frequency response and overload behavior, it provides no direct information on distortion behavior unless the distortion is very high (distortion isn't visible on a scope until it reaches 3% ... which is an overload condition on any amplifier). Used intelligently is far more subtle than it appears at first glance. This means you are well aware of the limitations of the equipment mentioned above. For example, MLSSA is a superb tool for analyzing the performance of drivers and complete loudspeakers. However, it won't tell you beans about the swept IM distortion of a loudspeaker, which is an important part of the character of any loudspeaker. Not only that, but swept IM measurements require a top-quality anechoic chamber and some very expensive B&K equipment to get meaningful results. In addition, MLSSA only provides a very indirect picture of the radiation pattern of the speaker versus frequency: This requires many tedious measurements and a motorized stand to get repeatable results. Also, there are plenty of things that are quite audible that just flat-out resist meaningful measurements. The sound of wires and resistors fall in this category, and there are lots of subtle decisions you can make with loudspeakers or amplifiers that just don't show up on any equipment at present. 2) All components going into the new device must be of known-good subjective quality. This includes speaker drivers, tubes, transistors, enclosure materials, interconnects, capacitors, resistors, wire, solder, transformers, rectifiers, even dumb things like power cables. It's all audible. In differing degrees, to be sure, but everything makes a difference. Please don't assume just because you can't hear some subtle difference that no-one else can. Not true! Everyone hears in different ways. Well, how do you select these "known-good" components? There are two ways: A) Pick out things you really, really like. (For example, I really, really like the Audio Note Ongaku and the Reichert amps. You might like Krell or
Levinson.) Find out what's inside. Don't be a pest, just quietly track down the capacitors, resistors, wire, etc. on your own. Do research. I subscribe to Sound Practices magazine, and I take what Kondo-San and Herb Reichert say quite seriously, no matter how outlandish it may appear at first glance. Why? Because I respect their products, and I know from direct experience that quality never happens by accident. B) Do your own sonic investigations. Go ahead, tweak away, and see which parts you like! Just keep good notes, ask friends over and record what they say, and tweak something that's already good to start with! Don't design something new and start playing with parts! You'll just go around in circles, since you'll be dealing with far too many variables all at once. Of course, once you start researching parts quality, you'll soon discover that tweaking takes just much time as "real" designing! That's why I ask around and find out what other designers are already using, and start from there. Speaking for myself, I just find it easier to start constructing with "known-good" parts that have worked well in other designs that I respect. Simple example: I'll probably use a 5AR4/GZ34 or 5R4-GY rectifier tube in my first amplifier. Why? Amps that I like personally (based on the way it sounds in my system) use these rectifiers. When I ask around, the people that design these favorites are quite definite in their opinions on the importance of a given type of tube rectifier, even talking about the differences between brands and year of manufacture. Frankly, I have no idea why these archaic rectifiers with all their problems sound better than solid-state rectifiers. I don't care. I want to design a good tube amplifier, not become an expert on rectifiers. Now when I'm happy with the overall sound of the amp, and not before, I might explore other types of rectification ... and maybe not. 3) To audition the new device, you need a "standard of reference" for the rest of your hi-fi system. Don't assume, by the way, that anything in the "Recommended Components
List" or the latest hot-dog $50,000 system at your local salon represents a "reference system". If you live in a medium or large metropolis, chances are excellent that someone has a superb system ... and audio clubs are a great way to find them. In my experience, and in the experience of designers I respect, hi-fi systems at this level of performance are not off-the-shelf systems, do not represent mainstream AES engineering philosophy, and may actually be comparatively inexpensive in dollar terms. Audio systems at this level have hundreds of listening hours invested in subtle alterations at every point, and the development process takes nothing for granted. For example, I used to think that tweaking power cables was basically nuts. I don't anymore, because I've heard the difference for myself, and that ended that question. If you don't have access to a high-quality system, there is a near-certainty that you'll start compensating for the defects of your other components. Mass-market CD players can either be shrill or very flat and one-dimensional sounding. Most commercial speakers have unnatural treble, a forward midrange, overblown bass, and poor imaging. Most transistor amps sound coarse and one-dimensional. The list goes on and on. That's just the way most mass-market stuff sounds, and not all "high-end" equipment lives up to the reputation. You just have make your own assessments. The real secret is to track down first-rate systems, hear for yourself, and find out what others have discovered. Tune into pleasure ... the right equipment will have the extraordinary ability to touch the emotional truth of nearly any recording, regardless of "sound quality." Seek it out! --------------------This article is continued in "Positive Feedback V (2/2)" -Lynn T. Olson ........
[email protected] Senior Editor, Positive Feedback magazine, a publication of the nonprofit Oregon Triode Society. For subscriptions, call (503) 234-4155 or fax (503) 254-3866.
----------------------Q: Why did you draw such big distinctions between schools of design in Part 2 of Soul of Sound? A: Well, I was stretching things just a little. There is a philosophical tension in high-end audio, though, and I wanted to point it out. There are plenty of well-respected designers who are thoroughgoing materialists; as an example: "If I can't measure it on my AP/MLSSA/pinknoise system that was blessed by the National Bureau of Standards, you must be fooling yourself again, since you are a totally unscientific nitwit that couldn't reason your way out of a paper bag." In response to this kind of name-calling, I can't resist quoting from the "The Science of Beauty" article by Herb Reichert which appeared in Vol. 5, Issue 1 of Positive Feedback: "In America, we assume that if it were to sound "just like live", all of the emotional content of the performance would be conveyed automatically. I submit this is a poor premise to base an entire industry on. There is no law that demonstrates or suggests that a perfect reproduction of the original soundfield would convey any of the original artistic quality of the performance. Additionally, there is no parallel in other media. A photograph is no substitute for a painting. Motion picture film or video will not capture a theatrical production or sports event. Bronze castings of marble sculpture lose most of the original beauty. What makes us think that vibrating transducers will communicate artistic quality just because they are almost linear? What is important to remember here is to remember that the record playing system is a media of its own like film, paint, or clay. With any artistic media, communication is achieved through the creative use of dynamic contrasts (drama) and profound architecture (form and structure). In this country, the creators have chosen to build audio that emphasizes the mental picture of the performance. Image, depth, transparency, and grain are of high importance to American audiophiles. Alternatively, the Euro/Asians have chosen qualities that emphasize the emotional content.
These music lovers design for maximum dynamic contrasts, presence, vividness, and effortlessness." As you can see, Herb's article is an open challenge to the philosophy of "perfect accuracy" that has dominated the US high-end industry and the allied slick magazines. As I have hinted in my previous articles, there is real turmoil in the US industry at this time. Going to the Winter CES was an interesting experience ... the great-numbers boys were doing THX demos with their megawatt amps and high-definition projectors, and the weirdos were playing LPs (in about 1/4 of the suites) with their directly-heated 300B and 211 amplifiers. The majority of the exhibitors at the Sahara were using tube amps ... for the first time since the mid-Sixties! As my patient readers have noticed, I digressed pretty far from the last question. Well, it covered a lot of territory. It's hard to convey the flavor of this industry until you spend a little time inside of it, which is why I slide over into a kind of gonzo journalism when I describe the social side. The dry, "this-is-what-I-saw-at-the-CES" stuff just doesn't convey the zany aspect of the high-end business. My heart is with the zany side ... I saw the Japanese eat Scott, Fisher, and Marantz whole, watched as Gordon Holt, Bob Fulton and Audio Research rose from the ashes of the American consumer electronics business to create an independent high-end market, then saw the two "underground" magazines mature and become the all-powerful gatekeepers for the high-end industry, and now ... full circle again ... a new set of magazines and a new philosophy is sweeping through! Maybe we'll all be listening to Sovtek 10 and 45-powered Tractrix horns in the next Millenium (the rebels at Sound Practices win). Maybe we'll watch digitally-compressed HDTV, with sound provided by diamond-cone drivers, powered by German 500-watt PWM digital amps, and Windows software-based crossovers (the numbers-guys win). All I can say is that future of audio is wide open.
*Yes, A Bonus Construction Feature!!!* Folks have been asking me about an easy-to-build version of the Ariels, then David mentioned that this issue might not have any construction articles in it. This prompted me to turn on the LEAP program on my PC and design a modestly-sized near-Bessel vented enclosure for the Ariel drivers. (I selected the near-Bessel alignment because it has the least overshoot and time-ripple.) Since the new arrivals are minimonitors, efficient, and 2-way, and are the little brother of the Ariels, Karna suggested christening them the ME2. Why not? So ME2 it is. The ME2 is 18" high, 8" wide, and 8.5" deep, with an internal volume of 521 cubic inches, and a pair of 1" diameter vents that are 4.125" long. It retains the large diameter radius (1.25" on top, bottom, and one side) of the Ariels. Everything else is the same as the transmission-line Ariels ... same drivers, crossover, front panel layout, composite MDF/plywood construction, etc. The LEAP modelling system shows a -3 dB point of 70 Hz, a -6 dB point of 53 Hz, and a box frequency of 48 Hz. (Since they are intended for use with triode amps, the alignment is centered on a Damping Factor of 8 when they are connected to the 4 ohm tap of the output transformer. They are usable with Damping Factors between 4 and 100). Your initial reaction might be: 70Hz! That's not very much bass! Well, actually, it's not too different than plenty of other minimonitors ... a lot of them are actually spec'ed at the -6dB point, which gives a better match to the perceived low-frequency limit. Still, I agree, you're not going to shake the room with these speakers. The ME2 aimed at three groups: folks who like the unadorned minimonitor sound, multichannel systems, and/or systems that use one or two biamplified subwoofers. A Multichannel Digression Part of the reason I designed the ME2 was listening to the Cogent Research SPI, which is a modern sum-and-difference 4-channel decoder using psychoacoustically optimum decoding coefficients (whew!) This unit offers
high-resolution "frontal quad" without the listening fatigue that most people associate with surround and quadraphonic systems. Long ago, I invented the Shadow Vector quadraphonic decoder for Audionics, which became one of the first dynamic-matrix decoders ever built. (This was far more advanced than the commercial CBS/Sony "Full-Logic" system, which used gain-riding and a static matrix.) As a result of months of fine-tuning the Shadow Vector, I became quite aware of the action of a dynamic-matrix decoder on various types of recordings. Many years later, after quad disappered from the map of hi-fi, the Dolby Pro-Logic Surround decoders became popular for theatres and home video systems. The modern Pro-Logic family of decoders all use a dynamic-matrix technique closely related to the Shadow Vector, the CBS Paramatrix, and the Sansui Variomatrix of the quad era. Since they use the same basic operating principles as the previous devices, movie soundtracks are mixed down in the presence of an engineer from Dolby Labs, to ensure that the mix doesn't baffle the logic circuits of the decoder used in the theatre. (Yes, it is quite possible to encode a mix that can knock the logic circuits for a loop. Happens all the time in multimiked classical recordings.) After many years (decades!) of designing and listening to multichannel systems, I've arrived at two main conclusions: 1) Multichannel systems require *better* electronics and speakers than stereo ... in the exact same way that stereo requires better equipment than mono. Also, as with stereo, all of the speakers and amps must be *precisely* matched. Unlike stereo, though, quality mismatching doesn't result in a left/right asymmetry, but instead, a very rapid increase in "phasiness" and fatigue. This is why so many people who use mismatched low-cost rear speakers eventually give up on quad. Using a "logic" system to force the localization into the mismatched speaker simply results in more listening fatigue, not less, as frequency response, phase, and polar patterns go through extremely rapid changes as the sound source is steered around the room. 2) Any system that requires the user to flip through mode switches for
each type of program (mono sources, movie surround, classical, jazz, rock) is fundamentally flawed. You don't need to do this in stereo, after all. A correctly designed multichannel system should provide stable, accurate, and fatigue-free localization of all possible encoding possibilities ... mono, panpot stereo, multimike stereo, Blumlein stereo, Dynaquad, EV-4, Ambisonic, Dolby Surround, etc. The fundamental principle of all logic-steered systems is that they sharpen the localization of the loudest sound and defocus the quieter sounds, with "attack" and "release" times of the logic-steering system carefully adjusted to be as unobtrusive as possible. Unobtrusive, though, isn't the same as inaudible, and this kind of fool-the-listener monkyshines can be detected when you play an ungimmicked stereo recording. Of course, logic-steered systems always provide logic-off or "hall" switches for this condition. This is **bogus**! It requires the user to choose between sharp localizations and natural sound. The Cogent, and the Ambisonic/Gerzon family of decoders, take the different tack of very carefully optimizing the fixed decoding coefficients, and trusting the user to build a fully matched multichannel system with correctly located speakers. This an approach I agree with. There's really no point in a multichannel system unless it is better in every way than a 2-channel system ... after all, stereo can be pretty fatiguing and unnatural if not done right, and multichannel is *much more* fatiguing and unnatural if not done right. My prelimary listen to the Cogent showed a lot of promise ... very wide and spacious soundstage, stable imaging, with reasonable but not great sound. BUT we were listening to a quad mid-fi Rotel amp and four Definitive Audio speakers at $300 each. Comparing it to the two Ariels and the Kassai 300B was ludicrously unfair ... the Kassai is a superb $40,000 amplifier, and the Ariels would have to sell for $4000/pair in a retail environment. Which brings us to the ME2 speaker. It's more compact, less expensive to build, and offers the same sound quality as the Ariels, with 25 Hz less bass. It's a prime candidate for an audiophile-quality multichannel system using the Cogent decoder, particularly with a high-quality pair of stereo subwoofers, which brings us to ...
The ME2 and Powered Subwoofers I recommend trying a 80 or 100Hz crossover as a starting point, and using no more than a simple high-pass capacitor for the high-frequency amplifier. I also recommend a simple, high-quality triode amp for the high frequencies, and a good transistor or MOSFET amp for the lows. This lets each amp function in its best frequency range ... in particular, the triode amp will sound far better if it doesn't need to amplify low frequencies, which is what the input capacitor accomplishes. Stereo subwoofers are a good idea, but not for the reasons you might think. It is indeed true that localization is very difficult to perceive when the wavelengths are longer than 10 feet. However, there is an effect known as "Spatial Impression", which conveys the impression of size, or space, and it is quite important at very low frequencies. A mono subwoofer, in other words, won't distort the image, but it won't sound as *spacious* as stereo subwoofers! In addition, it has been shown that a 2-speaker stereo image grows narrower as the frequency is lowered ... which means that it is desirable to place the stereo subwoofers much further apart than the normal 60 degree stereo-pair. Good results have been reported with 120 or even 180 degree spacing of the stereo subwoofers. Level-setting shouldn't be a big chore ... select some piano music that uses the bass register (I use the full-keyboard scale on the ProSonus test CD) and adjust the subwoofer level and phasing for the most natural and "fast" sound. Using piano music as a reference should curb the natural tendency to let the subwoofers creep up in level. The ME2 Crossover The crossover is the same as the Ariel, since the factors that affect the mid and upper regions are unchanged (cabinet width, driver layout, edge radius, etc.) I would stick with North Creek Audio as a supplier, particularly for the resistors ... do not use the typical sand-cast power resistors found in most crossovers. I recommend and use Spague 730P polypropylene capacitors, North Creek 12 or 14 gauge air-core inductors, Ohmite 10W 1% non-inductive power resistors, isolated grounds for the low-pass and high-pass sections, and a
bi-wired connection to the amplifier. Cabinet Construction The drawings pretty much tell the tale. Use Baltic Fir or Apple-Ply plywood for the internal members, and premium-grade MDF for the outer shell. The 1.5" thick front panel plays a significant role in quieting down the cabinet and is worth the hassle of bonding two 0.75" panels together. Be sure to make the speaker in mirror-imaged pairs and try to achieve the 1.25" radius shown on the drawings ... I found out that there is a pretty big sonic difference between 0.75" and 1.25" radius when I made the Version 1 and Version 2 Ariels. Lightly fill the V-shaped rear half of the cabinet with long-fiber wool (best for mids) or crimped Fortrel (like Acousta-Stuf), but don't cover the inlets of the vent tubes. If you're using the ME2 with subwoofers, though, feel free to stuff the vents themselves (this slightly improves the midrange at the expense of bass response). You'll notice that there really isn't any room inside the cabinet for a crossover ... that's intentional, since vibration of crossover components can degrade the clarity of sound. A small external box for the crossover is the best solution ... but keep it at least 18" away from steel stands and transformers. Setup I recommend placing the ME2's at the tips of an equilateral triangle with your listening position at the bottom of the triangle. Aim them at a point about 1 foot in front of you, with the tweeters on the inside of the stereo pair. You should be able to see just a little of the side of the cabinet that has the large radius (consider the speaker in the drawing the Right speaker). Next, adjust the height of the stand so the centerline of the tweeter is between 38" and 40" high. If the tweeter is much below ear level, the stability and overall size of the stereo image will be degraded, so adjust the height so the tweeter is at or above ear level. When the ME2's are correctly set up and used with the right electronics,
they will have a very big, spacious sound, a stable stereo panorama well off-axis, and will sound quite natural and lifelike even from another room. They are "tuned" for the best possible voice reproduction, so use recordings of male and female singers to do the final in-room tuning with crossover levels, cables, damping material, location from the back wall, etc. When you are all through with the tune-up process, the ME2's will have all of the sonic quality of the Ariels you've read about in previous issues of PF ... they're just as well braced, use the same drivers and crossover, and just as efficient at 92 dB/metre. So turn on your favorite 2A3 or 300B triode amp, get out your favorite sides, and have some fun! Suppliers: Cogent Research 25145 Manzanita Drive Dana Point, CA, 92629 (714) 493-0721 There are three models of the Cogent SPI, with identical sum-and-difference decoding coefficients, but progressively more elegant internal electronics. For people who enjoy multichannel sound but don't care for the subtle fatigue that logic decoders can induce, the Cogent units are well worth a listen. North Creek Music Systems Route 8, PO Box 500 Speculator, NY, 12164 (518) 548-3623 Top-flight drivers, capacitors, inductors, resistors, binding posts, glue, etc. as well as good information on subtle points of cabinet construction. I haven't heard the kit systems, but they look very good from what I can see of the descriptions. Madisound Speaker Components 8608 University Green Box 44283 Madison, WI, 53744-4283
(608) 831-3433 Many, many drivers, all with spec sheets. Also, all of the little bits-n-pieces needed for a complete loudspeaker. A&S Speakers 3170 23rd Street San Francisco, CA, 94110 (415) 641-4573 Even more drivers, but no spec sheets unless you ask. Good info on which vendor is using what drivers in their commercial speakers. VMPS kits as well as their own. ORCA 1531 Lookout Drive Agoura, CA, 91301 (818) 707-1629 They sell the very effective "Black Hole Five" damping material as well as the interesting Axon speaker cables. All kinds of interesting odds-n-ends from these folks. The Speaker Clinic 1 SE 47th Portland, OR (503) 230-0366 Glenn Phillips offers a wide range of consulting and cabinet construction services using his MLSSA, LMS, and LEAP computer systems. Speaker Builder Magazine and Glass Audio PO Box 576 Peterborough, NH 03458-0576 (603) 924-9464 A unique and outstanding resource for information. I've subscribed to both for many years now, and the quality has steadily improved the whole time.
About the only source for info on transmission-lines, build-it-yourself electrostats, and long-forgotten vacuum-tube innovations. Standard Disclaimer: All trademarks and registered names mentioned in this article are the property of the respective trademark and registered name owners. -Lynn T. Olson ........
[email protected] Senior Editor, Positive Feedback magazine, a publication of the nonprofit Oregon Triode Society. For subscriptions, call (503) 234-4155 or fax (503) 254-3866.