Stereophile Interview excerpt below
The Absolute Sound Interview click here
Widescreen Review Interview click here
Stereophile's Editor John Atkinson Interviews Designer Jim Thiel
Stereophile's March '98 issue featured an insightful interview with Jim Thiel written by Editor John Atkinson. Here is an excerpt of that interview.
John Atkinson visited the THIEL factory last year for a tour
and to sit down with Jim to discuss his design philosophy. The
results were published in the March 1998 issue of Stereophile.
After designing and marketing two bookshelf
style speakers starting in 1976, "I started thinking a lot
more seriously about not just making a good product, but making
a product that I hoped would be better than any others available."
Jim Thiel is the founder, chairman and chief designer of Thiel Coherent Source Loudspeakers, a leading and respected designer of advanced time coherent reproducers. A matched THIEL CS7 system is one of the magazines valued reference systems, often used to evaluate other system components. Jim Thiel was interviewed by Editor Gary Reber for this exclusive On Screen Feature.
Gary Reber, Widescreen Review: One of the major things that separate the many speaker designs on the market today is radiation pattern. What are your views on point source, line source, monopole, dipole, and bipole designs? Would you please explain each one of those terms for the benefit of our readers?
Mr. Jim Thiel: A point source radiates sound energy equally in all directions from a small-sized source with no reverse phase energy. Because the source size is small, the intensity decreases proportionately to the distance from the source. Real world speakers referred to as point sources almost always have reduced dispersion of energy at the higher frequencies, so they are not true point sources. But their radiation patterns are still similar to a point source because the sound source is small, resulting in a wide radiation pattern, and because there is no reverse polarity energy.
Both dipoles and bipoles radiate energy both to the front and to the rear. In the dipoles the rear energy is of reverse phase to the forward energy and this causes, in addition to reverse phase energy in the room, a cancellation of energy to the sides of the speaker. In the case of a bipole radiator the rearward energy is of the correct phase, so it can be looked at as an attempt to make a truer point source, one that radiates more uniformly in all directions.
A line source refers to a radiator that is big in the vertical direction. There can be monopole, dipole or bipole line sources.
My view is that a modified point source is the ideal, a point source that's somewhat modified in its radiation pattern so that the higher frequencies are radiated only to the front. I like this radiation pattern because it is most similar to real world sound sources. What I don't like about dipoles is that there is reverse phase energy radiated by the speaker which is completely unlike most real life instruments. The dipole radiation pattern does tend to give a spatial special effect, but I consider it unnatural and therefore undesirable. The bipole designs, not having reverse phase energy, I consider superior to the dipole. But in all the real world implementations of bipole designs there are what I consider to be significant interference problems between the forward and rear radiation which result in uneven and changing off-axis response. I consider the line source also to have the problem of not having a natural real life characteristic, in the case that their loudness does not diminish proportionately with distance as a live instrument would. They also do not radiate as much total energy into the room. So I consider the modified point source radiation pattern to be ideal, most like the sound sources we are reproducing.
WSR Reber: So you're saying bipole is a better design than dipole?
Thiel: I consider the bipole better because I don't like the idea of out of phase energy that the dipole produces. But I consider the modified point source to be superior to both in terms of natural reproduction.
WSR Reber: How these different designs interface with the room also differentiates them. Are your speakers designed to be used at or near room boundaries, away from room boundaries, on the floor or stand mounted?
Thiel: Our speakers are designed to be used away from room boundaries on the floor because this can give you the most natural spatial reproduction. The advantage of designing a speaker to be used near the room boundaries is to get additional bass reinforcement, and therefore, everything being equal, you will get a little more bass output or a little more sensitivity. But the problem with such a speaker is the interaction of the mid frequency energy with the boundary which causes colorations--unnatural tonal characteristics. Here again is an advantage of the modified point source radiation pattern. Not having much rearward energy at the higher frequencies means that this type is much less interactive with the room. It's much easier to get good results. The dipole and bipole designs are much more interactive with the room boundary and the results that you get are much more dependent upon the room and the speaker placement in the room. With point source designs, being away from room boundaries doesn't necessarily mean it must be far away, although theoretically it is more ideal to be far away. Even distances of one foot to two feet can give you most of the benefits of uncolored midrange reproduction.
WSR Reber: So optimally, you would suggest something like three or four feet?
Thiel: Yes, three to four is very good and five feet is even better, but you're really at the point of diminishing returns beyond that.
WSR Reber: What do you do in terms of getting a smooth frequency response and bandwidth?
Thiel: I feel that it is often not appreciated how sensitive your hearing is to minor tonal alterations. It is often assumed that because they may not be noticeable on a cursory listening, that they don't matter. I think that, in fact, even minor colorations significantly impair your long term enjoyment of sound reproduction and, therefore, I place a very high value on achieving extremely accurate tonal balance. Of course, many factors are involved in actually achieving a smooth frequency response, including the willingness to invest the engineering time and the additional costs required. There are three approaches to getting good frequency response. The first is to design drivers that have a very good response, which is more complicated than you might think. The second is to reduce cabinet diffraction which detrimentally alters the frequency response of the drivers. And the third is to engineer the electrical network to correct frequency response problems. We aggressively pursue all three approaches and we probably take the third approach further than anyone else. Our networks are very complex and much of the complexity is to correct small frequency response problems. Some people feel that using so many electrical components can degrade the sound quality, but I think this is only true if the electrical parts are of the usual poor types, like electrolytic capacitors. The down side of correcting these frequency response problems with complex circuitry of high quality parts is the cost.
WSR Reber: Should speakers used in a multichannel system be voiced differently than stereo speakers--for example, flat vs. a warm tonal balance?
Thiel: I don't think so. I subscribe very strongly to the philosophy that the speakers should accurately reproduce the signal that they are fed. Interestingly, Gordon Holt seems to feel that when accurate speakers are used in a multichannel playback array that you then, for the first time, get an experience that is as warm as life rather than with a stereo system where he feels that the playback is not as warm as it should be.
WSR Reber: Do you subscribe to "timbre matching" of the rear or side speakers, the idea that a rear speaker should be voiced differently than its opposite in front, as THX states, to counteract the different head related transfer function the listener experiences when sounds arrive from the side and rear?
Thiel: No, I don't subscribe to that. If the ear hears sounds differently from the rear, then that is true both in real life and when listening to playback. So if the speaker reproduces the signal as it was recorded then your ear will hear the reproduced music the same as it would have heard the live sound. So I don't feel that the speakers should be altered to predistort the signal to be what your ear would hear because your ear is going to hear it from the rear and make the alteration itself.
WSR Reber: Do you subscribe to the idea of identically matched speakers in a multichannel array?
Thiel: Yes, that is ideal. However, that is often not possible, especially for the center channel. What is most important is that the tonal characteristics through the midrange and the radiation pattern be very similar. I feel this is an advantage of our speakers. Since all our models possess very accurate tonal balance, you can mix and match any of our models to satisfy your budget or space requirements and still obtain a well balanced total system.
WSR Reber: Do you favor room treatments to alter acoustics? If so, what type?
Thiel: Well, of course, it depends. In general, I think that sometimes people worry too much about the reverberant characteristics of the room. I think almost all rooms that people are actually comfortable in and live in have quite acceptable reverberant characteristics and so, for the most part, I think it is unusual that people really need treatment. Now in some cases they do, but I think it's rather unusual.
WSR Reber: How about electronic equalization of the in-room response?
Thiel: Well, I'm usually not in favor of it for two reasons. One is that it is practically impossible to equalize your reverberant energy to compensate for the characteristics of the room without also altering the sound that you hear directly from the speakers. And I would go to great lengths to avoid that. And the other reason is that room characteristics are different for different listening positions. So even if you could achieve a correction of the room characteristics for one location, you're likely to make the room characteristics worse at another listening location.
There's an important concept to think about when you're understanding how you hear loudspeakers in a room. When you listen in a room, you're hearing two different signals so to speak. First, you are hearing the direct sound that comes directly from the speaker to your ear which does not encounter any reflections or interactions with the room boundaries. Secondly, you are also hearing the reverberant energy, which is energy that first interacts with a room boundary before it gets to your ear. You hear the reverberant energy later in time because that energy has taken a more indirect path from the speaker to your ear and has traveled a greater distance and has maybe bounced around the walls a few times before you hear it.
The important idea here is that your brain separates the direct sound from the reverberant sound. And what your brain and ear primarily tune into is the tonal character and balance of the direct sound which is not influenced by the room. The only way for that direct sound to have the correct tonal balance is for the speaker itself to have an accurate anechoic frequency response.
The effect of the room is that it will change the character of only the reverberant energy. So if you are in a very live room, the reverberant energy will sound very live, and if you are in a very dead room, the reverberant energy will sound dead. Your ear will be able to interpret what kind of acoustics space you are listening in based on the character of the reverberant energy, but the direct sound will not change because of the room.
So if, for example, you are in a very lively room that accentuated the high frequency reverberant soundfield, and you tried to compensate by reducing the high frequency output from the speaker, you would succeed in achieving more accurate tonal balance of reverberant energy. But you would've also unnaturally altered the tonal balance of the direct sound, making it too weak in the high frequencies. And I consider the value of the direct sound to be much higher than the value of the reverberant sound. So for both those reasons, I don't think it's a practical idea.
If you really have to do something about the sound of the room, it would be better to try to do it acoustically than to try to do it electronically.
WSR Reber: Do you believe that what's right for stereo is right for multichannel in the context of speaker directivity on the precision of imaging when more than two channels are present?
Thiel: Yes. I really don't see any fundamental difference in the requirements of imaging and speaker directivity depending on how many channels there are.
WSR Reber: There is a lot of discussion about bass loading methods. Do you favor second order sealed boxes, fourth order vented, or passive radiator designs, band pass, dipole, bipole, or other methods?
Thiel: Second order sealed boxes have a little tighter, cleaner bass character than the fourth order vented or passive radiated designs which are essentially equivalent. However, the comparison really shouldn't be between a sealed box system and a vented system with the same bass extension because, in the real world, what you really have to compare is two speakers that have equal sensitivity and equal size. Given a sealed system with the same sensitivity and same size as a vented system, the difference becomes that the sealed system will simply have less extended bass than the vented system. Therefore, it's really a question of whether or not you want more extended bass. For that reason, I design vented systems because, for a given sensitivity and size, you get more bass extension.
WSR Reber: What are your views on subwoofers?
Thiel: That they're very difficult to integrate well with the main speakers. There are several problems. The subwoofer is in a different physical location and is also working into a different acoustic environment. It's a different distance from the listener than the main speakers, and then the level matching is a problem also. In addition, there are serious problems with most of the crossover systems used for subwoofers. So these problems make it very difficult to integrate subwoofers well.
WSR Reber: Does that mean to imply that THIEL is not planning on producing any stand-alone subwoofers?
Thiel: Well, these are the reasons we have not to this point. However, I am now working on a subwoofer, and although there still will be integration difficulties, we're incorporating a couple of new ideas to achieve better integration than usually can be achieved with separate subwoofers.
WSR Reber: I have found that one of the advantages of subwoofers, at least in terms of reproducing modern day motion picture soundtracks, is that the level of bass extension is very deep and also the SPL level of the bass is very loud and powerful. Without a subwoofer you're not really able to achieve all of the impact of the soundtrack.
Thiel: That's true, although there's no theoretical reason why a normal integrated full range speaker could not produce the same output level and bass extension as the subwoofer. But, in fact, I agree with you that very few full range integrated speakers produce the output levels and bass extension that some people like for movie soundtrack reproduction. Therefore, there's certainly a practical benefit of using subwoofers.
WSR Reber: Do you recommend a dedicated subwoofer to reproduce the ".1" low frequency effects or LFE channel for maximum impact?
Thiel: If you need the output capability of a subwoofer to get the impact then that's a value judgment that needs to be made, and quite likely it would be the best thing to do.
WSR Reber: Do you think that bi-wiring offers an audible improvement that justifies the cost? Can your speakers be easily bi-wired?
Thiel: Our speakers cannot be bi-wired in the traditional sense of the speakers having two sets of input terminals to feed different frequency sections of the speaker. They can, however, be bi-wired in the sense of using two sets of wires, both sets connected to the same input terminals. And, surprisingly, this will give you most of the advantages of using bi-wiring in a traditional form, because just using two sets of wire in parallel means that you are cutting the resistance in half and the inductance in half.
In general, we don't recommend bi-wiring because I feel people are usually better off taking the amount of money that they would spend on two sets of wires and buying one set of higher quality cable. I think in most cases people get better results by using on set of better cable than using two sets of lesser cable.
WSR Reber: Do you believe that time and phase are audible parameters in loudspeakers? Do you favor first order cross-over slopes or steeper slope designs? Do you physically time align the drive elements in your designs? Is diffraction important? What if anything do your designs use to effect diffraction?
Thiel: Yes, I do believe that time and phase are audible parameters in loudspeakers. Very audible actually. It's not the kind of thing that's going to immediately make you say, "Wow this sounds incredibly different!" But it is certainly the kind of performance that adds a sense of realism and naturalness to the reproduction, and I think it's a major factor in being able to reproduce sounds so that they really do sound live and not reproduced. Correct phase performance contributes to a sense of natural space, enhances clarity of subtleties, and provides a greater sense of realism. I think it's quite audible and quite important, and therefore, I do favor first order crossover slopes in the loudspeaker as the only way of maintaining complete time, phase, and power response accuracy. We utilize first order crossover slopes in all our products.
We also physically time align the drive elements in all of our speakers by either of one of two methods--mounting the drivers along a sloped baffle so that the distance from the listener to each of the drivers is the same, or by using coincidently mounted tweeters that are physically time aligned.
I also consider diffraction to be important in loudspeaker design. It is important in terms of reducing the speaker's cabinet as a source of sound coloration. If the speaker cabinet edges diffract energy, it gives your ear/brain clues to the size, shape, and location of the speaker that's reproducing the sound, and the speaker starts sounding more like a speaker that's occupying it's location rather than the musical instrument's sound that it is trying to reproduce. So we reduce diffraction in our products by rounding the edges of the baffle quite strongly in three dimensions to significantly reduce diffracted energy.
WSR Reber: In addition to test instruments what source material do you use to evaluate loudspeaker? Music? What do you recommend that consumers use? How should they listen?
Thiel: We use music. Most of what I use is female voice because most of my listening and design process involves evaluating the tonal accuracy of the speakers. Some speaker performance parameters don't need to be evaluated subjectively nearly as much as the tonal character does. For example, the imaging capability of the speaker is pretty much predictable from the radiation pattern, the phase characteristics and the diffraction characteristics. If you get all these things right, the speaker will image wonderfully. You don't have to spend a lot of time subjectively evaluating different design options to optimize imaging performance. Whereas, in terms of tonal characteristics of the speaker, your ears are extremely sensitive to very small inaccuracies that are difficult to interpret from the measurements. We spend a lot of time, mostly by using female voice, evaluating the tonal characteristics of the speaker.
In terms of what the consumer should use, and I presume you mean to evaluate speakers that they may be buying, they have to evaluate a broad range of performance characteristics. We like to simplify everything that a speaker does into four performance characteristics; tonal characteristics; clarity; spatial realism; and dynamics. I think the consumer is very well served by focusing on those four characteristics and evaluating each of those with a different type of music. I think the best type of music for evaluating the tonal characteristic of the speaker is voice and that the customer should listen for naturalness of the voice reproduction. For evaluating clarity, you want to use any type of music that's quite complex. Often if you get a recording that's been made, say, in an old cathedral where you have a lot of reverberation from the original acoustic space, you can evaluate the clarity and musicality of the reverberant energy as a good way of zeroing in on and evaluating the clarity of the loudspeaker. In evaluating the spatial naturalness of the speaker, you want to make sure that the recording is made with a simple mic setup. And often some of the smaller jazz groups have recordings where you can best evaluate the spatiality of the speaker. For dynamics, you want to listen for the speaker not becoming strained or distorted at higher output levels. So if the buyer zeros in on those four characteristics, it makes it a lot easier to try to objectively compare speakers.
WSR Reber: In your designs, do you subscribe to a wide dispersion pattern or to a very narrow, focused, controlled dispersion pattern?
Thiel: I subscribe to a wide dispersion pattern. But even more important than wide dispersion is that the dispersion pattern be uniform at all frequencies as much as possible. Often in loudspeakers, if you measure the energy radiation off-axis, you will get some areas of the spectrum that are reproduced weakly, which result in unnaturally balanced reverberant energy in your listening environment. If the reverberant energy is unnaturally balanced tonally and also balanced differently than the direct sound, the sound becomes disconcerting and unnatural. So I think the most important thing is for the dispersion pattern to be even and not change with frequency. That's another benefit of the first order crossover approach: you get a very even dispersion pattern of overall energy. But I do consider it desirable to have a wide dispersion pattern. It gives you more naturally spacious reproduction.
WSR Reber: The Home THX program subscribes to the idea that it should be a very highly focused dispersion pattern to reduce or eliminate any room boundary reflections, particularly ceiling and floor.
Thiel: Right. Well, I think there are a couple of problems with that. First, any method I know of to reduce dispersion, to make the dispersion pattern more tightly focused, causes audible degradation of the natural sound characteristics. Another problem is that you end up with reproduction that has little reverberant energy content so that what you hear is a higher percentage of direct sound, which is an unnatural experience. The reason they give for having a focused dispersion pattern is to increase the intelligibility of dialogue on movie soundtracks. I think the problem there is often that the speakers just lack the clarity or the natural tonal balance they need to make the dialogue intelligible. The real solution is to get speakers that have better clarity.
WSR Reber: What is your view on the suitability of home theater systems for music reproduction and vice versa?
Thiel: This reminds me of the 70's when people would say "Speaker X" is really great for rock music but not good for classical music, and "Speaker Y" is vice versa. And it is a bit like that, I think. The argument about rock speakers and classical speakers was that rock music needed speakers that had good bass and dynamic reproduction and didn't particularly need a speaker that had good spatial characteristics or tonal accuracy. So that if you had a speaker with good dynamics and bass reproduction then it was great for rock, but it was not necessarily great for classical music because the sound character might be unnatural and the spatial reproduction poor.
It's a similar argument here. It's true that in a home theater system, the bass output capability is a lot more important than it is for normal music reproduction. Therefore, a speaker that could be very good for music may not necessarily be practical for home theater reproduction. And very often the speakers that do provide the bass you want for home theater can have poor tonal and spatial performance and poor clarity and so sound bad when reproducing music. But, in fact, you can make a speaker that has bass output capability for movie soundtracks and also has the accurate tonal characteristics and natural spatial characteristics needed for good music reproduction. Therefore, the ideal would be a speaker that can do everything well.
WSR Reber: So your conviction is there should be one speaker for all types of reproduction?
Thiel: Yes, a good speaker is a good speaker, given this one consideration that you may need greater bass output capability for home theater reproduction. A speaker doesn't care where it's signal comes from--whether it comes from a music recording or a movie soundtrack or, for that matter, a computer game. A signal is a signal as far as the speaker is concerned and will reproduce what it's fed.
WSR Reber: Do you advocate using identical speakers in each position in a multichannel system?
Thiel: Ideally, yes. A lot of people won't choose to spend the money it would take to have five great speakers, but ideally it would be good.
WSR Reber: Since you put it that way, do you feel that it would be better for a consumer on a tight budget to invest in five speakers of lesser cost than invest in two main speakers--left front, right front--of much higher cost and three lesser speakers?
Thiel: I guess it depends on what he's planning to use the speakers for. If he's planning on using the system to listen to both music and movies, the second purchase would definitely be better. And this approach can also work well with movie soundtracks if the bass energy is redirected toward the left and right front speakers or to separate subwoofers. Then there's not as much requirement for the other three speakers in terms of bass and output capability. But it's still very important that the other speakers have very similar tonal balance and radiation patterns, but that can be achieved. You can get small, less expensive speakers that are of good quality, but of limited output and bass extension.
Now if the system is only used for movie soundtrack reproduction, then this argument would have less force, and it might make more sense to have five of the same speakers.
WSR Reber: Now, in terms of subwoofer capability in that sense, do you subscribe to a single mono subwoofer, or do you subscribe to the idea of at least two or more?
Thiel: I think more is better--four or five would be best-- and you are getting back to the approach of the speakers being full range. But a benefit of multiple subwoofers is that they improve the spatial experience, regardless of the opinion of some people that low frequency energy is not directional. Also, if you use only one subwoofer, you tend to have more of a problem activating room resonances. The more subwoofers you use in different locations the fewer problems you'll tend to have with low frequency resonances in the room So I prefer more subwoofers--at least two.
WSR Reber: Do you recommend identical time distance relationships of each speaker from the sweet spot listening position in a multichannel system?
Thiel: Yes. Depending upon the source it can be important for maintaining a sense of natural space.
WSR Reber: How do you view speaker placement to optimize phantoms between left front and left back speakers and conversely right front and right back speakers, or for that matter left back and right back? What about along the diagonals between left front and right back and right front and left back?
Thiel: There's a lot of appeal to using the four corners of the square--at 10:30, 1:30, 4:30, and 7:30 positions. Some people argue that you can't get a good enough phantom image behind the 9:00 and 3:00 positions, so you might want to pull the back speaker forward to the 8:30 and 3:30 positions.
WSR Reber: Should speakers in a multichannel system have identical full range capabilities for each discrete channel?
Thiel: Again, that's the ideal.
WSR Reber: Do you subscribe to the Home THX specification for dissimilar speakers using direct radiators in the front hemisphere of the room and dipoles in the rear hemisphere of a multichannel system?
Thiel: No, I do not subscribe to that, although it can work fine for most movie soundtracks. By using dipoles in the rear, you tend to make all the energy coming from the rear speakers diffused and non localized, regardless of what the movie producer intended. Since the movie producer can make the sound non localizable on the soundtrack, you don't need the speaker to impose that characteristics onto everything it reproduces. The problem with having speakers with radiation patterns that diffuse the sound is that you are then limited to hearing only diffused sounds from the rear speakers. And there are sounds, even on movie soundtracks, that should be localized and they are not able to be if you have speakers with such radiation characteristics. So I think the speakers should be point sources for all the playback channels, and if the sounds are meant to be diffused and non localizable, then they need to be recorded that way. Then you retain the capability in your playback system of reproducing localizable sounds, which it otherwise would not.
WSR Reber: Particularly with discrete digital soundtracks now, there is the capability to localize and have phantom images--even diagonally.
Thiel: Yes, and if you have speakers that are diffusing all of the sounds that they are reproducing, then you are losing the capability of experiencing the multichannel music recordings the way they were intended to be experienced.
WSR Reber: Is a different model speaker permissible for the center channel?
Thiel: I think so. But it's important that the tonal characteristics be well matched to the other speakers, particularly left and right front, and for the radiation characteristics to be well matched. Of course, it's ideal to have the same model because you then have a perfect match. But I think that you can get center channel speakers that are well matched, especially if you stick within the same manufacturer.
WSR Reber: To what extent is the digital 5.1 discrete format effecting your approach, both in terms of speaker design and application in room placement in the context of a multichannel multi-speaker system?
Thiel: Well, it's not particularly effecting our approach
at all since our philosophy is to make speakers that accurately
reproduce the signal they are fed. There is nothing in our design
approach that is incompatible with multichannel. In fact, I feel
that our approach is particularly good for multichannel systems
because we have always made speakers that have imaging as a strong
performance aspect. And imaging is what multichannel is all about.