Part Two: Implementing a Philosophy of Sound
Two months ago, in "Developing a Philosophy of Sound," the first half of this feature interview, Andrew Jones focused on the studies, experiments, and experiences that underlie his thinking about sound, especially as it applies to his loudspeaker designs. Jones describes it as "A balanced design approach which maximizes, along multiple parameters, real-world performance and original artistic intent across the musical spectrum."
In Part Two we discuss Jones’s role as Director and Chief Engineer of TAD Laboratories, how he has applied his philosophy to the speakers he’s designed for TAD, and the systems he builds around those speakers at international audio events.
Peter Roth: Now that we’ve explored your personal history, perhaps you could give us a brief history lesson regarding TAD.
Andrew Jones: The history of TAD is also the history of Pioneer, because TAD is, and always has been, a division of Pioneer. Now, however, Technical Audio Devices Laboratories Inc. is a separately incorporated division rather than just a "skunkworks." Being a division of Pioneer makes some people nervous -- "Pioneer? Why would I be buying Japanese speakers for $78,000?" Yet TAD Labs is a separate division staffed by people who are primarily just TAD. There are a few oddballs like myself who do some work for other Pioneer divisions -- while I’m focused primarily on TAD, I do design some speakers for the particular needs of Pioneer USA -- but most work exclusively for TAD.
I don’t think many people actually realize that Pioneer started as a speaker company over 70 years ago and has been developing speakers ever since. Nozomu Matsumoto, the company’s founder, came to the States in 1935, ’36. This is where he got passionate about hi-fi, and, realizing there was nothing similar happening in Japan, he returned to develop a driver. For 1937, it was a really cool driver. It was a flared paper-cone driver with an aluminum cone-dome dustcap to extend the high frequencies. That dome dustcap was mechanically decoupled from the cone at the high frequencies -- almost a concentric driver. That single, full-range driver was the centerpiece of Matsumoto’s first product, the A-8 dynamic speaker. Really well engineered, especially given it was the first thing Pioneer ever did.
In the late ’70s through the mid-’80s, Pioneer had a run where its speakers were rather well known. The HPM series is what everybody remembers. [Specifically, the HPM-100 was a high-fidelity, four-way, four-driver bass-reflex loudspeaker system designed by Bart Locanthi and his team of ex-JBL engineers, and built by Pioneer in Japan from 1976 to 1979.] That is really the only time when people had a general impression of Pioneer making speakers; otherwise, it was all electronics, and then plasmas, before we exited that business. So Pioneer has continually been making speakers, developing some remarkable technologies and ideas throughout that history.
TAD was formed in the ’70s when Pioneer developed a beryllium diaphragm. If you look at all the research papers from the Audio Engineering Society in those days, it was always the Japanese with new driver materials -- diaphragm materials. They were just researching everything. Were they implementing these new materials correctly? That’s another question. But were they developing astonishing technology? Yes, they were. If you look at a patent search, there is tons of stuff there, predating most of what we (US, UK, and European speaker manufacturers/designers) had been doing. So Pioneer developed a beryllium compression driver for professional use, and created TAD for the "pro" side, whereas Pioneer remained consumer oriented.
Going into the professional field as "Pioneer" wasn’t going to work, so TAD was established as a separate group. It was named Technical Audio Devices because it was producing technical audio devices. TAD made inroads into the studio market, which became hugely successful -- the best compression drivers in pro use. Nothing else could really touch it in terms of bandwidth. The larger you make the diameter of the compression driver to achieve lower frequency response, you lose highs. This is due to breakup modes and tolerances between the phase plug and the diaphragm. For the first time, with beryllium, we could get to 20kHz in a 4" diaphragm, and it was just a marvelous device. TAD developed a range around that driver and became established in the pro market (but not in the home). For 25 years, TAD stayed "pro." There is such a split between the pro market and the home that very few products cross between the two, and neither side really pays attention to what’s developing in the other silo.
So I joined TAD and, in around 2000, I headed up the development team for a new project. The questions were: "What should be the next direction for Pioneer/TAD in drive-units? For what future technology do we start developing new drivers?" We decided on concentric drivers. That’s where my history was, and, surprisingly, that’s where Pioneer’s history was too. Pioneer introduced the first concentric driver, in 1954, and had been developing concentric drivers ever since. When I started with Pioneer in 1998, I developed a new concentric driver (with flexing diaphragm), which was awarded a patent. In performing related patent searches -- I designed new magnet construction for a concentric driver that was really cool -- everything looked fine until I decided to check Japan patents. Our Japanese engineers did the research, pulled up a 1977 Japanese patent, and I said, "Shit, that’s exactly the magnet structure I’ve just been developing!" Who did this? Pioneer! Which group in Pioneer? Takausara’s. He led my group before he moved on through Pioneer. To top it off, that magnet assembly was shown in the patent diagram in a concentric-driver implementation. It was obvious that Pioneer itself had predated us in what we were trying to do.
Concurrent with this concentric-driver project, Pioneer bemoaned being locked into lower-end speakers and yearned to break out. We needed to do a higher-end speaker. But a high-end "Pioneer" speaker would be saddled with credibility issues -- you can start at the high end and come down in price, but you can almost never go the other way. We had Elite, but that is limited to consumer electronics. Should we originate a new brand? Instead, we cast our attention to our TAD brand. But to be a true TAD, it couldn’t simply be a higher-end speaker, but rather the best we could possibly do. Bells started going off, ding ding ding: "Let’s load it all together, this new direction in drivers is what we’ll develop for TAD, and because TAD equals beryllium, let’s do a beryllium concentric driver." I found myself exclaiming "Yes! Yes!" because every problem I’d ever faced in developing concentric drivers around existing materials should start to disappear by working with beryllium. That is what started it, and it has been the heart of everything we’ve been doing at TAD ever since. We spent several years before we were ready with the original TAD Model One, in 2003.
PR: We’ve touched on the balance between theory and real-world results. Yet many designers seem to dogmatically adhere to a single approach that they deem theoretically superior, or even perfect: e.g., only single-ended-triode amps sound good; only fourth-order crossover networks work; anything utilizing acrylic sounds like crap; etc. With your body of scientifically derived experience and knowledge, how do you approach the development of TAD speakers?
AJ: It is interesting, because a lot of people focus on such a limited aspect of the performance, whether relating to how the equipment performs or the particular aspect of music that they try to optimize. When people talk about single-ended tubes, they talk about a "magic" midrange, or vocal quality (setting aside whether or not SET amps do, in fact, have that quality). But even those fervent SET adherents will admit they don’t do bass or high frequencies. "But it has a magical midrange!" If that is all they care about, they optimize that singular aspect without regard to any other aspect. That is not balanced design. Instead, I’m designing to a market where I need a thoroughly balanced design approach. To incorporate midrange quality and great bass and good treble -- that is what I’m trying to do.
It brings me back to seeking a natural, balanced performance across the spectrum of all types of music. I’ve had this conversation with Paul Stubblebine (of The Tape Project) and Keith Johnson (of Spectral and Reference Recordings). In discussing the fringe, tweak-oriented parts of the market, and having listened to the sound that part of the market produces, we have asked ourselves, "What is it those folks are listening for?" Once, after a reviewer praised my demonstration, I ended up listening to his personal setup. I listened and wanted to say something nice about it without being two-faced, but I really didn’t get it. How could he have so enjoyed the sound that I presented with my demo while his home setup was totally alien? There must have been something in particular that he was looking for in his home sound that was contained within mine, some subset of what I was producing. I don’t know what it is, because his system’s sound was such a restricted subset. What aspect was he seeking? I wish I knew, because I’m fascinated to know how people are satisfied, how reproduced sound represents what they hear in real life. The recording people, the Stubblebines and Johnsons, have these same questions; they don’t get it. What is it this audience is looking for? At TAD, I’m trying to present a more broad-based product that has the best of everything as far as I can do it, and doesn’t focus on just one narrow part of the performance.
PR: Apparently, research on the ear/brain interface indicates that different people hear in dissimilar ways -- their brains react and respond differently to identical stimuli, which is a possible explanation for the diversity of sound sought by various niches of the marketplace.
AJ: This reminds me of when I started work on this TAD project with some of the Japanese engineers. The traditional bias is that the Japanese can’t design speakers. Why not? They are arguably the best engineers in the world. So why this bias? Maybe in the past they were designing for a different philosophy of what the sound should be. Now, however, in working with my Japanese team during the design and implementation process, we never had a philosophical impasse, which was a relief. Whenever we improved things, we felt that the results were an improvement, to all of our satisfactions. We reached a level of performance that encompassed everything they need to hear from a system and everything I need to hear, even if we were hearing differently. Anything that made it sound better to me did not make it sound worse to them, because it was always additive rather than subtractive. It gets back to my goal of ultimate balance; as long as it does everything well, you’ll get consensus regardless of what unique people may search for to hear. But if you limit the engineering to a particular aspect of sound, then you won’t get the consensus; it will be a very niche product. It helps to understand what they may be listening for, but don’t leave it to the market niche to develop the system by themselves. Rather, develop it by someone looking at the whole, which also satisfies what the niche person needs from the product. Then you’ll get broad-based acceptance.PR: In connection with this balancing act, you noted that successful speaker designs maximize the potential benefits of a given approach and minimize its downsides -- e.g., the many strengths of compression drivers, despite their bandwidth limitations. What are your thoughts on the weaknesses of different approaches to loudspeaker design, such as planar speakers (ribbons and electrostats), hybrids, dipoles, line arrays, and horns?
AJ: I’ve always been a fan of electrostatics. I own lots of different types. They do have a certain quality of detail, resolution, and transparency that is difficult to achieve with regular speakers. But they have huge practical limitations in terms of output levels, large-scale dynamics, and bandwidth. A full-range electrostatic or ribbon just can’t do it. I don’t care how many people claim a full-range electrostatic can. I listen. They will boast that it doesn’t have a subwoofer. I can tell straight away.
Because I have this past history with electrostats, I’ve always wanted to integrate their strengths while obtaining everything else. Electrostatics do provide a subset of what you need to do. They work great for that subset, but they can’t encompass everything. As soon as you try to make it wider bandwidth, you change the technologies, and integrating those together becomes very difficult. MartinLogan -- great speakers -- but the challenge in a hybrid is to get them to work adequately together. Another limitation with electrostatics is their uncontrolled directivity. The only electrostatic speaker that has ever come close to controlled directivity is the Quad ESL-63 and its variants, but even that is a bit too directional at high frequencies. I believe in controlled directivity, and the Quad took a very elegant approach to achieving it, but because of the practical limitations of output capability, Peter Walker was forced to use a 4"-diameter central element. This restricted the high-frequency directivity too much.
Then you’ve got the rear wave of a dipole. Classically, people will say that a dipole interacts less with the room. But once you start angling the panel, it is not clear-cut because the axial wave will be joined by tangential waves "in room." There is all this reflected sound energy from the back of the room. When you do the critical analysis, it is not that clear-cut that a dipole will be better. Theoretically, in a large space, the direct-to-reverberant ratio is better by 4.8dB. But it’s no good if that direct-to-reverberant ratio is not uniform at all frequencies, and it never is. Because of their large size and their lack of controlled directivity, the response changes drastically off axis, very quickly (especially with a flat-panel electrostatic/ribbon). You move a few degrees off axis and everything goes away. Trying to measure them becomes a challenge, because you’ve got to be exactly on axis, and it’s hugely dependent on frequency.
Which leads to line sources. Many electrostatics claim to be line sources, and most line-source speaker systems claim to be, in fact, a true "line source." Nonsense. A true line source is infinite. The only way to approach that in a room is with a true floor-to-ceiling line source, with a hardwood floor and a hard ceiling. Anything that isn’t that is not a line source. Theoretically, line-source radiation dissipates at one-over-R instead of one-over-R-squared. The in-room, direct-to-reverberant ratio ought to be superior. The stereo imaging as you move off-center should be more stable. But when you do the math, even a 6’ line source acts as a true line source only above about 1 to 2kHz. Below that, it acts as a regular monopole. So what you find is that the frequency response varies dramatically with distance. It will be flat, and then it will start rolling off at 3dB/octave, and that transition point is about a kilohertz.
So where do you design it for? A particular listening distance? I did research using 6’ and 8’ line-source arrays, and I did the measurements. I expected the transition point would be a few hundred hertz, but in reality it was up around 1kHz. So then, do you balance it for a particular distance? The direct-to-reverberant soundfield is so frequency dependent, and so room dependent, that it is going to be very sensitive to room placement, room acoustics, and so on. How, then, to design for this? It totally breaks the rules of what I know about designing, especially from a measurement point of view. These different technologies do one or two things well, but they are not a fully balanced design. A line source that is made of multiple smaller drivers has huge power-handling capability and can be explosively dynamic. But it still presents line-source problems, and, with multiple tweeters, introduces phasing effects between the tweeters. You can’t get them close enough at the highest frequencies (unless you have a true, continuous source). So a discrete line source is not as good as a continuous line source. But then the continuous line source (like a single ribbon) can’t go down low enough in frequency. So now you have to stack them horizontally (the tweeter, midrange, and bass), and it is back to more compromise. You don’t put a tweeter alongside a midrange (which is why center-channel speakers are fundamentally wrong). But all discrete, multi-driver line sources do that. That’s not right either.
We examine all these theoretical ideals, but real-world implementation never matches the ideal. It’s always a set of compromises. Do those compromises outweigh the theoretical benefit of what you were trying to achieve? Very often they do, and a distinctive kind of sound results, which is attractive to some people. I’m not saying all these speakers are bad. Some of them have qualities that are really nice. But do they do everything well? No. Do they do enough well for a lot of people? Yes. But for me -- and I’m pushing to do everything well -- it can’t quite do that. It doesn’t mean I don’t try to find solutions as to how I could take one of those design approaches and transform it into what I want to do. I’m always thinking of that idea.
As to horn-loaded speakers, it goes back to the efficiency and dynamic capability of compression drivers. It is a good thing to get the raw drive-unit as efficient as possible. You minimize heat in the voice-coils -- heat is always detrimental. But the horn must be implemented correctly. Horns got a bad name in the past because they were lacking controlled directivity. They can also have high distortion levels, although there is a debate as to how audible that is. The materials of the horns themselves are usually not so great -- they can vibrate and color the sound. If they go for extreme efficiency, you get reflections from the horn’s mouth. Often, they are incorrectly terminated in the transition from the throat to the mouth. They used to be designed solely for maximum efficiency. Now many are backing off, as they sacrifice some efficiency in exchange for controlled directivity, for better control of in-the-mouth reflections.
These days you can achieve much better sounds from horn drivers, arguably better than with regular drivers. I do employ horn-loaded drivers -- a concentric driver is essentially a horn-loaded tweeter in a horn that is moving -- but because so many have a negative reaction to the word horn, we instead refer to a waveguide. With every tweeter, I pretty much put it in a short waveguide to help control the directivity. Whenever I’ve been forced to put a tweeter directly on the surface of the box, I’ve made the measurements and then struggled with the design of the rest of the system. It is just the wrong thing to do, from my viewpoint. Controlling the directivity with a waveguide of the correct shape is a good thing. You can get a very dynamic sound, low coloration, and other benefits from a properly designed horn driver these days. But you still have this bandwidth limitation. So how do you overcome that? It may now mean you are crossing over at 1000Hz. Is that better than crossing over at 2000-3000Hz? Maybe. But you are still typically crossing over to a bigger bass driver, and so still have the same issues of interference off axis, so you’re back to design compromises. I think horns have gotten a bad rap. They can be so much better these days. You can do some really good hi-fi systems with correct choices of drivers and matching wavelengths.
PR: In the balancing act of speaker design, what other features do you favor with your dynamic drivers, relating to both ultimate performance and the speaker-room interface?
AJ: Sealed cabinets vs. ported cabinets vs. transmission lines: Nearly all of my speakers are ported/vented, in large part because of the advantage it gives in efficiency vs. box size vs. the bandwidth matrix. Does venting present disadvantages? Certainly, people recognize that it has a faster rolloff, so the transient performance is theoretically worse than in a closed box. But, in practice, is that is what you hear? Have naysayers really done the experiment of listening to the theoretical difference in impulse response by having the faster rolloff rate? You can’t just build a speaker that is closed-box or reflex and listen. That’s wrong. Proper experimental procedure changes only one thing at a time. Venting affects the linear response of the speaker, but you never listen to the linear response of the speaker because, at practical listening levels, there is always distortion. The driver is not moving linearly, and as it moves in a nonlinear manner, it is not just about creating distortion products, but changing the tuning of the speaker box. As the driver moves farther, the "L" product is changing, and so the alignment of the reflex system is changing. If you plot frequency response vs. level, you’ll see that the box tuning is peaky and just changing a lot. That is because of driver nonlinearities.
So it’s not just reflex vs. closed-box, but the interaction with the driver. And you could say that that doesn’t happen in a closed-box system, because the box is more of a determining factor to the low-frequency alignment; the driver won’t change that alignment as much when it starts moving farther, as it would in reflex. So, OK, if you are going to do a reflex system, design a better driver. And that is what I feel I have done with TAD’s bass driver. This idea of reflex not having fast bass, not having attack, not being as in control? I don’t think that has ever been a criticism of my reflex speakers. A lot of people have had to ask, "Is that reflex?" They don’t feel it sounds like what they’ve come to expect from a reflex design. There are theoretical advantages to reflex, but not if you don’t implement it properly. Making a reasoned judgment about the theory, taking into consideration the practical realities, is different from following conventional wisdom, which is handicapped by what other people have been able to do, leading to possibly wrong conclusions.
I’m not a fan of transmission lines. It is a kind of a badly designed vented box. Transmission line, in theory, means the backwave continues forever. But it doesn’t. So a transmission-line design is not truly a transmission line, but rather a badly terminated transmission line. In theory, you load the end of the line with a resistor equal to the impedance characteristics of the line, so the wave gets absorbed at the end and never gets reflected. In a practical speaker transmission line, it does get reflected. You can see it in the line resonances, which dictate that they stuff (typically, overstuff) the cabinet. And they don’t close the end of the line; rather, they open it to get some bass out. Now, however, that bass is more delayed than in a reflex system. And now you have this long resonant line rather than just a compact box, and it has anti-resonances in it. By the time you’ve stuffed those, you’ve gotten nothing out of the end, and so it’s a balancing act between how much you get out of the vent vs. how you handle these anti-resonances. You’ve got a huge suckout at the first anti-resonance in the line, which gives you a sense as to why they put the driver one-third of the way down the line: to try to minimize that one. Yet they give the impression of woofing, because they can be tuned quite low.
From a theoretical analysis and the practical implementation, a transmission line is not for me. To me, the ported cabinet with superior drivers offers a better real-world balance of efficiency, box size, and bandwidth. It is the implementation of a conceptual approach that can make all the difference. Good designs don’t all have to start with the same approach, but they have to maximize their inherent benefits and minimize their limitations, with a real-world overlay to identify how much of the ideal can actually be realized.
PR: Your fascinating Uni-Q experiment certainly informs your two-channel speaker designs. With TAD’s buildout of its new listening room, did you come up with ideas for an optimal listening room, or -- because you know your designs will not be used in an optimized room -- did you take a different, more end-user representative approach?
AJ: That is an important thing! To acoustically outfit my listening room, the question was, How far to go? We are designing for playback in people’s homes, not studios. Studios are acoustically treated, and if I take a speaker designed for the home and put it in a studio, you will always get a different result -- particularly in the bass, because studios always have bass traps. Studios don’t have the modal problems we face in real-world rooms because they tune them out. In a home, you nearly always have low-frequency issues. Corner traps, in reality, do very little to properly treat room modes at low frequencies. You are going to have low-frequency room modes, and that will change the character of the sound. So if I am designing a speaker and listening to it in my room, what does that say? It tells me how the speaker sounds in my room. Then how much do I treat my room? I’d like to have a really good sound, so I can have a reference point. But if I tune it just to sound good in my room, how will it sound in all the others?
In Tokyo, we have several different listening rooms. Many of them are heavily treated acoustically, to the extent that I walk in and realize I can’t even listen in them, because they simply are not representative enough. I always insist that, before I finish my sound tuning, I take it around to different places. I need to listen on different equipment, because that will influence the sound, and to have different rooms and feedback from different people. I am going to get those different opinions and environments to make my ultimate judgment, to get a type of averaging of what it will perform like. There are certain aspects that will come through whatever the room does, but there are also aspects that are variable.
PR: It’s returning to the concept of real-world performance along multiple parameters!
AJ: Yes! With such a strong, broad parameter baseline, the end user can then effectively tweak the system to taste.
PR: Every time I’ve been to one of your show setups, whether at a Consumer Electronics Show, Rocky Mountain Audio Fest, Newport Beach, wherever, you’ve always challenged for Best Sound of Show! Even with a changing lineup of supporting equipment, you apply your Philosophy of Sound.
AJ: That’s right. As we are establishing ourselves, I think we have to focus on the sound experience. Some manufacturers may feel that they are at a stage where they don’t have to worry about sound at shows. For them, it is all "You know who we are. You know what we can sound like. Let’s just do business." That is an important part of being a hi-fi company. I’m not on the business side, however. Don’t ask me to handle any of the business. What I care about is the result. I set up the show, I choose the equipment, I choose the music, I work with the studios for the music, I’ll work with manufacturers for equipment. In short, I will do the best sound that I can.
I remember, from when I was young and going to annual hi-fi shows, that this was my one chance. I just wasn’t going to be able to walk into a store and listening to a $100,000 system when I could buy only a several-hundred-dollar rig. But I always wanted to go into a hi-fi show and be blown away. We built a reputation of always making good sound, doing a good demo. So I can choose the best, as manufacturers have confidence that they can loan me equipment and know that it will be presented in the best light. The demonstration room is my showcase, every year. Even if I haven’t changed the speaker, I’ve changed the electronics, I’ve found better recordings. All I want, every year, is for everyone to come back and go, "Boy, it sounds better! What did you do? It sounds so much better than last year." I’m showcasing what is possible in a sound system, whether it is due to my speakers or to what I’m partnering with. It has to show that I know what good sound is. Plus, it shows that they have a choice of associated gear. They are not thinking, "TAD only works with Pass Labs, or it only works with Ayre," but rather that TAD speakers sound great with all kinds of electronics -- although now we have the TAD electronics, which gives us total control over the result.
PR: Still, however, you have to contend with the room.
AJ: When setting up to do a demonstration at a show, what do you do? You have to deal with the room you’re given. I may have to make decisions about how to set it up for the best sound, but I know my speakers can handle it. That is also why, for example, at the Venetian (CES), I always get the same room. I’ve learned that particular room’s characteristics, and each year I can work to make it slightly better, because I already know what I’m going for. It is like a Formula One team returning to the same track year after year . . . you know the basic tuning parameters to start with, but this year you’ve learned a bit more, so you can tweak it for that extra 100th of a second.
. . . Peter Roth