In praise of AR3a's

[Zilch]

Zilch, if you are looking for my review of Howard's books, you are out of luck. I never read any of them. I never even saw the magazine he usually wrote for (was it Absolute Sound) even once. If I read an article he wrote for High Fidelity Magazine 30 or 40 years ago, I've forgotten it. I didn't read most of the articles in most magazines I subscribed to and I rarely if ever take note of who the author is.

Indeed, I was merely expecting a reiteration of your earlier, cogent synthesis:

I never heard of him.

[soundminded]

Yikes! It certainly was not Absolute Sound. Rather, it was Sensible Sound, which, although sometimes certain of its contributors were capable of heading off into tweako territory, generally tried to steer a rational line. Two other contributors (out of a group of maybe ten others) were David Rich and David Moran, who are anything but members of the lunatic fringe. The magazine also had some very good record-reviewing columnists. Heck, one of them was me.

The Bose 901 review was not supposed to be a real review at all. It was never stated as such in the magazine. I did try to get a pair to review at my place (and would have followed manufacturer guidelines for placement), but the company wanted me to fly up to their headquarters for a conference first, no doubt to get a proper dose of brainwashing before they turned me loose with a pair of their speakers. They did this, even though some time before they sent me photos of the 901 for my first book and in that volume I discussed the Bose approach and pretty much gave it a thumbs up.

Yes, the owner of the 901s I reviewed for The Sensible Sound did indeed have them pulled out from the front wall five feet. He swore up and down. after doing a lot of listening and speaker moving, that this made them work better than the factory suggestions, primarily because of his large listening area. He had even presented a paper at one AES convention outlining his "image model" approach to soundfields in home listening rooms. (I think that Soundminded would get along with this guy very well, by the way.) I did run a room/power curve on the speakers (uncorrelated pink noise as a source, with me moving my microphone SLOWLY over a 1 x 1 x 5 foot area at seated ear height at his listening couch, while my AudioControl RTA did a cumulated, 20-second average measurement) and the result was fair. The 901s in his room did not do a particularly good job of delivering flat response to my measurement microphone.

Yes, I know that pulling the speakers out five feet would have an impact on the sonic impression, but it would not have much of one in terms of power input to the room above the mid-bass range. The power input from the speakers to the room (and reflected off of walls) would remain the same no matter how the speaker was located. I have discovered this after moving a lot of speakers and measuring the results. Pulling the things out would reduce mid-bass, quarter-wavelength cancellation artifacts that would happen with factory-mandated placement, actually.

However, measurements notwithstanding, I was impressed with the ability of the speakers (even just two of them, although the owner also had two more handling surorund-channel duties further back into the room) to generate a realistic soundfield with small-ensemble (like jazz groups) input. He did want suggestions, and the only thing I could come up with would have been for him to move his "surround" 901 units, which were hanging from the ceiling, high up and aimed at the rear wall, so that they were high up to the sides of the listening couch and aimed at the side walls. I also suggested that he get a decent subwoofer - a large one.

This article did not adhere to my usual speaker-reviewing style, simply because it was as much a review of the huge room (21 x 31 feet, with a high ceiling) and the philosophy of the owner as it was a speaker-system analysis.

The article in High Fidelity (my first, actually) involved analyzing sales techniques used by dealers hoping to push certain products that might not have been the best of their kind. It was anything but a technical exercise. By the way, I do not consider myself an audiophile. Enthusiast, yes, but not an audiophile.

Howard Ferstler

901 was the first and among the few attempts to ever market a speaker which is deliberately intended to be integrated into listening room acoustics by the end user. Even though this approach was only partially successful in the way the concept was executed in 901, the advantages it offered made it an overwhelming commercial success. It's hard to think of another electronic product that has been around for so long in a form that is relatively similar to the way it was when it was first introduced.

I'm not surprised that the power response did not look particularly flat. I assume that the measurement was a steady state measurement which is the time integrated response that includes the room acoustics and not an FFT response curve. The reflected sound will be strongly influenced by the absorption/reflection characteristics of the room, especially the surfaces behind the speaker to a greater degree than other designs will. When 901 was first introduced, low cost graphic equalizers were not available to home users. Therefore you were stuck with whatever the manufacturer gave you in the way of control. One shortcoming of this design is that the front and rear speaker levels and equalizations cannot be adjusted independently to minimize the different effects different environments create by compensating for them electronically. The equalizer affects all of the drivers simultaneously including the front driver. Therefore the only effective control is placement.

The conversion from an AS design to a ported design with series III marked a transition from an audiophile targeted product to a mass market targeted product. IMO this was done to make the speaker much more efficient so as to be usable by consumers who could not afford high powered amplifiers. It also changed the relative cost/profit ratio in favor of the speakers by increasing the price of the speakers at the expense of the cheaper amplifier. It became suitable for use with low to medium powered receivers. The performance cost was probably the capability to reproduce the lowest octave of sound. Speakers with this capability such as AR9 have other problems such as systems being prone to acoustic feedback, enhanced turntable rumble and audibility of other recorded low frequency disturbances and can challenge many amplifiers by forcing them to the point of clipping. Original Bose 901 does this in spades for all but very high powered amplifiers and well isolated low rumble turntables.

One aspect of this speaker Gordon Holt pointed out in his review in Stereophile Magazine is that the drivers we now would call "midwoofers" do not make good tweeters because of their inertial mass. I agree with him on this point. My own criticism of it includes its failure to reproduce the top octave of sound satisfactorily. I heard series VI in a Bose store on a business trip in California last year and it seemed to me to have the same defect.

Another reviewer in e/e magazine and I think in one or two others found a mid bass peak which they said was inaudible. I find it very audible. I don't know if the later versions had that problem. I think at least one version had an equalization adjustment in that range. In the small room I have them in now, about 14 x14 the bass cuts off pretty high as I would expect and as a result the low bass needs considerable boost to sound flat. It's easy to push the 138wpc receiver I'm using to drive it to clipping...with an extra 8 db of boost at 30 hz. The "below 40hz" 10b shelf attenuator switch on the equalizer becomes mighty handy.

As I've reported elsewhere, I had a lot of fun re-engineering these speakers and I'm very pleased with them now. With careful equalization for each recording they can be very accurate reproducing individual instruments and small groups as they would seem to me to sound in that room. It's the best I've done or heard so far.

I don't consider myself an audiophile but a music lover and an engineer. As a hobby, I engineer sound fields. The equipment is a means to an end, not an end in itself for me so I'm rather circumspect about it, I don't get very excited about any of it, no more than someone would get excited comparing an Allen Bradley resistor with an Ohmite resistor or a Texas Instrument transistor with a Motorola transistor of the same type. I find it cheaper, easier, and less expensive to re-engineer existing speakers other people usually don't want such as these according to my own theories than to start from scratch.

Bose has been very generous. When I took these out of storage after many years and inspected them, I noticed that the putty used to seal the drivers to the cabinet had dried out. I called their service department for advice and they offered me a trade in for a brand new pair of series VI at half price. Rather remarkable after 35 years and no proof of original ownership. I opted to keep the ones I have and I have no regrets about it.

[Carlspeak]

Howard:

It looks like you've reached the limit of your upload space allotment. You can either contact Mark (admin.) for more space or go to your "My controls" area and consider erasing some of your oldest uploads. That is, of course, if you will want to upload anything bigger than your miniscule remaining 245 KB allotment in the future.

Another option you could try is to ask Zilch for some of his upload allotment since he hasn't used much yet to show you his listening space

[Zilch]

Howard:

It looks like you've reached the limit of your upload space allotment.

Please, Howard, post more high-resolution pics of your rather big listening space....

Sure. Why not....

(Heh, heh.... )

[Carlspeak]

Interesting. I thought that therer were only limits to each thread and not to what an individual would post on every thread.

Well, I have probably posted enough pictures, anyway. I am sorry that the three graphs I wanted to post will not show up. Many people might be interested in how the 901 behaved, and there are probably many Allison fans out there who would want to see what Roy's very best speaker could do in the way of a room/power curve. And of course some might care to know just what kind of results I got with my from-scratch work with those smaller systems of mine, not that I would intend to make any more to sell.

I might go to the "my controls" area if I can figure out how to do it. The problem is that I would not know which items to delete.

Howard

It isn't hard. Once in the my controls area, look down toward the left corner of the page. There is an optioin there called "manage attachments". Click that and a listing will come up with detailed info. on each upload. Size, date, topic. No. of views, etc. which should help you decide.

[Zilch]

My ploy to shame you into posting pictures of your installation by posting many of my own obviously did not work.

Bah.

With each bloated upload, you relished the thought that, being on dialup, I would never see it....

[kkantor]

Mark,

I have a great idea: you should ask Howard and Zilch to buy banner ads to reply to each other here. Even if you give them a highly discounted rate schedule, I sure it would support the site, and probably let you retire early!

-k

[Zilch]

There's the name for our new band, Ken....

[speaker dave]

Well, I go off hiking in Newfoundland for a week (see attached) and you guys get into an interesting discussion of room acoustics!

As best I can see, while a "reverberant" field in the large-hall sense you "postulate" cannot form over a long period of time, it is impossible for the off-axis sound generated by any loudspeaker to not be heard in that room, and since most of the reflections (first, second, third, etc.) will be getting to the listener fairly soon after the direct-field signals (and certainly sooner than they would compared to reflections in a large concert hall, where it is universally acknowledged that the hall shapes the orchestral sound), it has to be assumed that they impact the sense of spectral balance put forth by a speaker. We do indeed listen in a sea of reflections, even in small rooms. If that speaker's wide off-axis signals are choppy, it will negatively impact the spectral balance of the speakers as heard by the ears. Even Toole will admit that, which is why he indicates that the off-axis output of a speaker should be smooth. The reason he wants it to be smooth, is because it is ultimately heard in that room, and not just as something coming at the listener from a bounce or two.

You need to carefully read Toole. He explains that a smooth off axis response is important because it is the best indicator of resonances or "colorations". He specifically states in several papers (although he vacillates a little on the subject) that the shape of the power response curve or the system's directivity index curve is not a strong indicator of quality. His initial two-part paper shows that in the rank ordering of the 16 systems that major holes in power response don't lower the rankings of the top 4 speakers. Conversly, some speakers with excellent power response are poorly ranked.

Typical "reverberant chambers" are not all that much larger than typical home listening rooms (although, admittedly, since they are unfurnished they reverberate more), and I do not think that even you would say that reverberant fields do not exist in reverberant chambers.

I'd agree that rooms of any size can have a reverberent field above a certain frequency (sorry Zilch). This is, of course, the basis of acoustical scale modeling. If a 1/10th size model of a concert hall is tested with 10 times the usual frequencies, then the results are presumed (and by experience, proven) to be valid. Reducing the size of the model doesn't prevent a diffuse field from forming at proportionally higher frequencies. I've both measured and modeled my living room and above about 160 Hz the field appears to be sufficiently diffuse. That means that, above that frequency, measured RT is consitent around the room. This transition is called the Schroeder frequency and is determined by the density of standing waves in a given frequency band. Standing wave quantities always go up with frequency (viewed in octave or 1/3 octave bands.)

A statistical approach to acoustics (the Sabine equation) has served us well for a century. It is well known to be only a general approximation but it lets us define RT, level vs. source distance, etc., with only a few parameters defined such as room volume and mean absorbtion coefficient. The shortcoming have long been known: Sabine acoustics doesn't do well with wide, low ceiling office spaces for example. Baron and others have modified the equations for better accuracy, but the usual approach today is to do complete modeling with a ray tracing programs such as CATT acoustics. Still, most any typical room can be considered a diffuse field above a certain frequency.

To say that we listen in a mix of finite paths is certainly true. However, you seem to believe that just because they are finite there are not many of them. You appear to believe that after a few bounces the sound just disappears and has no impact on the speaker's overall spectral balance. You may believe this, because you have never had a chance to hear good, super-wide-dispersing speakers performing in good rooms.

It isn't the quantity of "sound hits" but the timing, relative to the direct sound. Adams gave an intersting paper that showed that the general shape of the LF response was largely determined by the initial boundary bounces, that adding a longer time window added "fuzz" to the curve but didn't change the shape. Papers by Lipshitz and Vanderkooy, and Soren Bech showed that early reflections impact spectral balance but later reflections had no impact. the ear judges balance through a variable time window that is quite short at high frequencies, just admitting the direct sound and cabinet reflections. In this regard, the sound does "disappear" after a few bounces. Note that this is perception of spectral balance, not spaciousness, that we are talking about.

I also noticed that one of the criticisms of the omni design in the explanatory text is that its behavior is "a bad thing for imaging." That remains one of the problems for many audio buffs, who treasure imaging because they have not attended enough live concerts in good halls to appreciate just how poorly imaged the soundstage is when live acoustic instruments are involved and the listening distances are greater than that of the conductor. Note that in Toole's book he was bowled over when he attended a concert at the Musikverein in Vienna, where the sense of spaciousness (and I am sure the lack of precise imaging) caught him rather off balance (page 51). He had probably never heard anything like that sense of large-hall space from any of the speakers he lionizes. From there, he obviously had to move on to heavily promote surround sound and additional speakers, because he never did grasp how well some stereo pairs could the trick all by themselves. Note that I am big on surround-sound, too, but I also believe that super-wide dispersing speakers make surround sound work even better.

The Musikverein is considered a breakthrough as one of the first larger halls with a good sense of spaciousness. It was the stimulus for a lot of the study and subsequent understanding of the roll of early lateral reflections in a good hall sound. I don't know that it is at odds with direct sound imaging? The goal is to get more energy to the ear prior to 80 msecs, and to have as much of it arriving laterally as possible.

I'll usually shut my eyes for a few moments at a live concert to evaluate the "imaging". It is true that it isn't precise in a "Hi Fi" sense, although the more directional instruments such as the trombones, can be well pinpointed. Strings can be determined at least by group. In the absence of a picture, is it wrong to want a sharper sonic image of the orchestra?

Regards,

David

[soundminded]

Yes, I like the room integration idea. Allison did that with his models, too, particularly as his placement suggestions related to mid-bass cancellation artifacts. Maybe the only "electronic" product out there that has outlasted the 901 in terms of the design concept is the classic Klipschorn.

It would be difficult to make sense of an FFT measurement of a 901, I should think. In any case, I continue to believe that what matters is what the listener gets in the way of the direct- and reverberant-field mix at the ears. Different speakers can do this in different ways, of course, but certainly a smooth sound at the ears has to be very important, indeed, no matter how the sound is propagated or what we have with the spaciousness issue. As I have noted before, I do a "room" curve that is the result of me doing a 20-second integration (continuously averaging) measurement with my AudioControl SA-3051 RTA, while I move the microphone slowly over a roughly 1 x 1 x 5 foot area at seated ear height, about 14 feet from the front wall of my main room.

Regarding the response smoothness of the 901, I had intended to attach a graph of the curve I ran at my friend's home, but the program here kept saying that 99.76 MB of the 100 MB limit had been used, with only 244.45 KB of space left, and so I was not able to do so. I do not know what attachments are elsewhere on this thread that would use up so much previous space, but maybe there is a malfunction somewhere. I might try later. For comparison purposes I had also wanted to show a curve I had run on my Allison IC-20 speakers, as well as one I had run on those home-built speakers of mine that I use in my smaller system, but obviously that was also out of the question, given the attachment limits I have encountered. Funny, I have never had this problem with previous attempts. Maybe Zilch has taken over the attachment program. Hey, just kidding, Zilch.

Frankly, I think the 901s would benefit as much from subwoofer assistance as a lot of other speakers. A sub would also offset any bass limitations with later series entries, and would free up main-channel amp power to handle things from the midbass on up.

Holt made a good point about the speaker when it comes to driver mass, but another issue with the drivers involves diameters. Obviously, the front-aimed driver is going to beam more than a tweeter, and so the speaker is fairly directional in the treble over what we would ordinarily call the treble range. However, each of the four-group arrays in the back probably each behave as a single larger driver, and so the energy they are aiming at the front wall at slight angles has to be quite directional, too. If the wall is fairly smooth that bounced energy will remain just as directional. I have pointed out to a number of people in the past that the 901 is actually a fairly directional loudspeaker, with the sound beamed in three directions.

I think that Bose offered you the same trade-in deal they did to my friend, but he took them up on it.

Howard Ferstler

The problems you cited were from a practical consumer standpoint insurmountable in 1968 except in a well equipped laboratory. Today, their solution is straightforward with off the shelf hardware so common and cheap as to be no problem for anyone who understands them and is persistent enough. Even so, it took me about 3 1/2 years of experimentation. The results retain the advantages of the direct/reflecting principle but there are no 8 foot wide pianos. The orlginal 901 and series II had enormous low frequency reach with usable output down to 23 to 26 hz with 10% THD about twice the distortion of AR3a but still inaudible. The hitch was the power requirement and maximum output capabilities of a single pair.

The results not only vindicate the basic concept but the overwhelming superiority and adequacy of the Redbook Digital Compact Disc system. Every musical instrument can be reproduced with such accuracy that their beautiful rich tones, mellowness and clarity are easily evident without the noises, acoustic feedback, other distractions, and deterioration of vinyl phonograph records. The 192 khz 8x oversamply Toshiba chip which proved indistinguishable from the 1 bit 8x oversampling JVC chip of 16 years earlier is not merely adequate but ideal with DVD players incorporating it costing as little as $30. I would consider anything that sounds different from this chip flawed. Right now I'm listening to "Guitar Favorites" played by Norbert Kraft on Naxos 8.553899. Most surprising is how beautiful and mellow the sound of a trumpet can be. Pianos, especially the large Steinway and Baldwin grands, superb violins and well trained human voices retain all of their power, purity, and beauty. Trouble to adjust the system for each recording individually but the results are usually well worth the effort.

[Zilch]

I'd agree that rooms of any size can have a reverberent field above a certain frequency (sorry Zilch). This is, of course, the basis of acoustical scale modeling. If a 1/10th size model of a concert hall is tested with 10 times the usual frequencies, then the results are presumed (and by experience, proven) to be valid. Reducing the size of the model doesn't prevent a diffuse field from forming at proportionally higher frequencies. I've both measured and modeled my living room and above about 160 Hz the field appears to be sufficiently diffuse. That means that, above that frequency, measured RT is consitent around the room. This transition is called the Schroeder frequency and is determined by the density of standing waves in a given frequency band. Standing wave quantities always go up with frequency (viewed in octave or 1/3 octave bands.)

A statistical approach to acoustics (the Sabine equation) has served us well for a century. It is well known to be only a general approximation but it lets us define RT, level vs. source distance, etc., with only a few parameters defined such as room volume and mean absorbtion coefficient. The shortcoming have long been known: Sabine acoustics doesn't do well with wide, low ceiling office spaces for example. Baron and others have modified the equations for better accuracy, but the usual approach today is to do complete modeling with a ray tracing programs such as CATT acoustics. Still, most any typical room can be considered a diffuse field above a certain frequency.

See Toole regarding these points. Small rooms such as are typical of home listening environments are not Sabine spaces, RT is all but meaningless, and what diffuse soundfield is present is ansiotropic. Once spectral balance and spaciousness are analyzed independently, there being both a physiological and psychoacoustic basis for doing so, it is easily seen as foolhardy to incorporate a reliance upon room characteristics into loudspeaker design, and wide dispersion consequently fails.

PS: right now I am reading Antonio Santosuosso's "Barbarians, Marauders, and Infidels: the Ways of Medieval Warfare." A very interesting read.

Building a trebuchet and lobbing stuff like refrigerators (I ain't sayin' what else, but it begins with "A") around would be WAY more fun:

http://en.wikipedia.org/wiki/Trebuchet

[Zilch]

Rather, some do a few response-curve measurements at a one-meter distance, often on and off axis a few degrees (usually just horizontally, not realizing that moving the microphone up a few inches and then measuring horizontally again will give completely different results), and then assume that this approach will determine just what a speaker will do when listened to at typical - and obviously much greater - distances in typical rooms.

Yah, presume we are stupid, Howard:

http://www.pispeakers.com/misc/Vertical_Nulls.wmv

[Zilch]

Your problem is that you read Toole and Geddes and assume they are correct....

Toole, Geddes, and Parham, now, and the list is ever-expanding.

I can make an ultra-wide dispersion loudspeaker in less than a minute, actually, and have suggested to others how to do it here.

(I use reflected and not reverberant to keep from offending you)....

Alas, you're still missing the "early" part.

Regarding that video (how, I have to wonder you can watch with that dial up connection of yours)....

Call the Zilchster "Resourceful."

[The "Edit" button still MIA on your screen, is it, Howard? ]

[kkantor]

I have read audiophile excuses before indicating that recorded sound (with no pictures) should have "better" imaging than a live performance because of the lack of visual clues. While that may seem logical (and is a way to justify making beamy speakers for listeners who sit locked into the sweet spot), it still stands against the concept of high fidelity sound reproduction that supposedly works to simulate a live-performance "sound" in home-listening environments, notwithstanding a lack of visual clues.

Howard Ferstler

I've said this many times in print, and I stand by it. I don't think it's an "excuse," I think it is a really interesting point that invokes important questions about what "accuracy" means in a cognitive sense. Further, it suggests, (as did Einstein in so many words...) that Logical Positivism is a poor framework around which to try and justify a philosophy of science.

But, let's keep it simple for a moment, and stay with easily digestible, Positivist assertions about imaging:

A speaker's job is to reproduce accurately what is fed to it. A recording engineer's job is to provide the speaker with signals that capture the original listening experience. A speaker with poor imaging sets an upper bound on what can be achieved. It forces the loss of information that cannot be recovered. On the other hand, a speaker with excellent imaging allows the recording engineer to deliver as much, or as little, localization information as they want to.

-k

[soundminded]

I've said this many times in print, and I stand by it. I don't think it's an "excuse," I think it is a really interesting point that invokes important questions about what "accuracy" means in a cognitive sense. Further, it suggests, (as did Einstein in so many words...) that Logical Positivism is a poor framework around which to try and justify a philosophy of science.

But, let's keep it simple for a moment, and stay with easily digestible, Positivist assertions about imaging:

A speaker's job is to reproduce accurately what is fed to it. A recording engineer's job is to provide the speaker with signals that capture the original listening experience. A speaker with poor imaging sets an upper bound on what can be achieved. It forces the loss of information that cannot be recovered. On the other hand, a speaker with excellent imaging allows the recording engineer to deliver as much, or as little, localization information as they want to.

-k

It seems to me based on the kind of music you've said you like best, you'd do well to listen to the same kinds of speakers used to produce it in the first place. Community Sound, JBL, Altec, even some of those old Crowns sound men dragged around from rock concert to rock concert. I'll bet they didn't use AR3a, AR303, or NHT3.3 at the Paradise Club.

[speaker dave]

Hi Howard,

Thanks for your thoughtful reply. Here are a few comments on your points.

I would imagine that one would be able to detect resonances or colorations at less than wide angles, too. What Toole seems to dodge is the impact of VERY wide off-axis performance. (One reason that he and others may dodge the issue is that when radiation gets really, really wide it becomes more difficult to measure the output of the speakers in a way that satisfies those obsessed with direct-field performance; one example would be to try to characterize the behavior of the Bose 901 by employing Toole's standards.) As best I can tell after reading much of what he wrote on the topic, his "ranking" issue still basically involves listener taste. Given the variables with rooms, listener position, recordings, and the way certain people prefer certain kinds of speaker radiation-pattern behaviors over others, a ranking-based paradigm has certain value, but is most definitely not going to be the end all as it relates to what is and is not absolutely important with speaker design. Toole is looking to have one "immutable" standard for speaker performance in a world where taste, once certain reasonable parameters are adhered to, plays a huge and important role.

I believe that Toole uses power response as a benign means of smoothing. He states that frequency smoothing (1/3 octave, etc.) will obscure resonances, while spatial averaging will allow abberations common to all radiation angles to stand out. The hemispherical curve averages are essentially as revealing as the spherical, and probably a sufficent number of curves over any directional sector would suffice. I'm not totally comfortable with dismissing angle related aberations, i.e. reflection effects, as Toole does.

Regarding his test methodology, I feel he has been consistent and very thorough over the decades. Subjective rankings, by definition involve listener taste. Early NRC tests showed that a variety of listeners came to repeatable rankings of loudspeakers in their blind tests. Later tests showed that mono vs. stereo was not a strong variable in the rankings. (The ESL 63 was one exception. It was ranked poorly in mono tests and improved to below average in stereo tests.) Other tests showed that room acoustics weren't a strong variable. If I'd come up with any test where a wide range of listeners consistently prefered certain speakers and disliked others, and if I found that the room wasn't a factor in the rank ordering, I'd be pretty happy with that test.

As I noted in my review of his book, there is no way that I can tell that he ever did A/B comparisons between good fairly directional speakers and good super-wide-radiating speakers. Most of the comparisons he notes involve speakers that have fairly basic design similarities when it comes to cabinet shapes and driver-angle arrangements. The one exception would be dipoles, where he does at least bring various Quad models into the picture.

He addresses general directivity in Chapter 7 of the book, clearly stating that wide dispersion speakers that encourage lateral reflections, other things being equal, were prefered by listeners. He also addresses dipole speakers with the ESL 63, and bipole speakers in discussions of surrounds. I'm not sure the Allison system offers a unique case. Unlike the Bose 901, with a clearly negative d.i. the Allison falls somewhere in the middle with a low but positive d.i. The effect of its "super wide dispersion" is just to reduce the ratio of direct to reflect sound, along the order of what a somewhat further listening seat would. I just don't believe that a special class of speakers was omitted from his tests.

As for Toole's indicating that the directivity index is not a proper indicator of quality, that gets us right back into the "taste" issue all over again. Anybody who has seen pictures of the room Toole used for his listening sessions in Canada before he built his turntable system (and I assume the room he used at Harman was similar in approach) will note that the front wall was lined completely with drapery, with that drapery also extending down the side walls some distance. (Admittedly, I am not sure that he retained the drapes with his later turntable arrangement.) While this was not specifically a pure LEDE layout, it certainly favored speakers with certain (and let's face it, narrow) dispersion patterns over others. In rooms like that it would be all but impossible to compare, say, a pair of Bose 901s against some of his favored models. The big, triangular-shaped Allison models would also not do particularly well, what with the drapes absorbing their wide off-axis output, nor would speakers like the Ohm Walsh units or the dbx Soundfield models do well, and certainly planar-magnetic and electrostatic types would have problems.. When your "ideal" room is set up as Toole prescribes, all bets are off as to what does and does not sound good in the way of speaker performance. And remember, too, that with his turntable system the arrangement still works against speakers like the against the wall Allison models.

I never have said that power response was the end all of design (something Allison implied in his early papers), but I have indicated that for a speaker to get into the ball park it needs decently flat power response. (In Toole's semi-padded rooms this might not apply, which may explain the problems he had with certain flatf-power designs.) Beyond that, radiation pattern behavior (paying homage to Mark Davis, when he was at dbx) becomes very important. Ironically, apparently Toole did not consider Davis's research into time/intensity trade offs with speaker design all that important, either, which explains why there are zero references to that research in his book. As I recall, Toole also liked to do mono speaker comparisons, which while interesting, would put certain designs at a distinct disadvantage when it comes to the goal of delivering a realistic soundfield. There is no way to happily separate spectral balance from overall soundfield performance, in my opinion. Different factors may be involved, but the merge to create a whole. When I measured the Bose 901 in my friend's room (sorry, but as we know the graph could not be downloaded) its response was not particularly flat. However, as a stereo-soundfield system its radiation pattern did some wonderful things, and with some recordings the speaker was impressive and as realistic sounding as hell.

Surely you don't mean that a speaker should have flat power response if that entails a rising on axis response? That is contrary to every modern study that I am aware of. You really should read the Lipshitz and Vanderkooy paper on modifying the power response of a combined system (ESL63 atop a KEF 104/2). They clearly found that flattening power response at the expense of axial response sounded wrong and that even flattening power response while retaining flat axial response sounded "bright" Another example of this is the Sean Olive experiment from chapter 20 of Floyd's book, where listeners are asked to draw the perceived response of the speakers they are evaluating. Where the power response is flat and the axial anechoic response is rising they drew the response as rising. Where the anechoic response is flat and the power response falling, they perceive (draw) the response as flat.

If you just mean that dispersion should be wide and so axial response and power response should be close to each other (directivity index low), then I have no great quible with that. We are just back to the legitimate listener preference of which do you prefer, imaging precision or listener envelopment? If we need to make a choice between flattening axial response or flattening the power response the conclusion is inescapable that axial response takes precedence.

The more I read the more I am convinced that this debate Zilch and I have been having about reverberant fields involves semantics, and I thank your for doing some clarifying in this area. In a technical sense he is partially correct (as you indicated), but the bottom line involves the strength of the reflections and their impact on the perceived spectral balance. And if those reflections are a mess the speaker will not perform as well as a speaker that generates smoother reflections. There are ways to deal with this, of course. One is to build speakers that have smooth response at all angles off axis, including those that are very wide. Another is to build directional speakers that have smooth off-axis behavior, even at wide angles. I certainly experienced this with the Dunlavy SCII and Cantata models I auditioned, and maybe those Geddes speakers Zilch likes exhibit this characteristic. A third is to listen in rooms with plenty of lateral absorbing capability (LEDE types, and beyond), so that speakers with erratic off-axis performance have those offending signals absorbed.

(In one article in Stereo Review that I wrote years ago, plus another I wrote for The Sensible Sound some time later, I pointed out that the AREA covered from 45 degrees off axis to 90 degrees off axis in front of a speaker was almost 2.5 times as large as that covered from zero to 45 degrees off axis; this means that a LOT of sometimes erratic energy can be heading into the room at angles much wider than either Zilch or Toole seem to believe are important to spectral balance. And this energy can be coming from the drivers themselves and not just from cabinet related resonances.)

I will admit that at high frequencies the direct field may dominate in terms of spectral balance (although Allison pointed out that the further you sit from the speakers the more the reflected energy increases in impact, and may enthusiasts tend to snuggle up to their speakers) in spite of the speaker having super-wide dispersion. However, this still leaves the midrange and upper bass as determiners of spectral balance at those frequencies (even perhaps at those closer listening distances). Also, while spaciousness may not have a major impact on spectral balance, it does have a huge impact on the way the soundstage is perceived, and in my experience super-wide dispersing speakers have a more realistic soundstage with a majority of good recordings than speakers with narrower dispersion, even if their off-axis behavior is smooth enough to satisfy Toole.

No quibles here. Note that the Soren Bech experiments showed that for midrange frequencies, generally only the floor bounce was sufficient in strength and early enough in time to be audible. At low frequencies the steady state room curve seems to be fully relevant.

I have read audiophile excuses before indicating that recorded sound (with no pictures) should have "better" imaging than a live performance because of the lack of visual clues. While that may seem logical (and is a way to justify making beamy speakers for listeners who sit locked into the sweet spot), it still stands against the concept of high fidelity sound reproduction that supposedly works to simulate a live-performance "sound" in home-listening environments, notwithstanding a lack of visual clues.

Howard Ferstler

Regards,

David

[speaker dave]

See Toole regarding these points. Small rooms such as are typical of home listening environments are not Sabine spaces, RT is all but meaningless, and what diffuse soundfield is present is ansiotropic. Once spectral balance and spaciousness are analyzed independently, there being both a physiological and psychoacoustic basis for doing so, it is easily seen as foolhardy to incorporate a reliance upon room characteristics into loudspeaker design, and wide dispersion consequently fails.

Hi Zilch,

I've re-read the Toole chapter on this.

First, I'm not comfortable with the term "Sabine spaces" or non-Sabine spaces". I believe he and Shultz are refering to the completeness of diffusion of the room and whether or not the Sabine equations give accurate RT calculations. Clearly, even a small room has a reverberent field (a complex sound field that builds up as a sound source is turned on and, when the source is turned off, dies at an approximately constant rate). Above the Schroeder frequency, RT is not meaningless, it just might not be that close to the number you get from Sabine's equation.

For a sound field to be anisotropic is not of concern either. At my last job we spent a lot of time designing lateral reflectors into concert halls to bathe concert goers with early lateral reflections. This is an instance of anisotropism(?) by design and it was considered very desirable.

Comparison of listening rooms to large office spaces is also troubling. In acoustics, office spaces are singled out as an important but very different case. If a concert hall can be simplified to a 6 sided box (certainly the rectangular halls can), then a large office is approximately a two sided space with no side walls. If the office is wide and deep enough then only the floor and ceiling surface matter. Studies show, as Floyd points out, that a constant reverberent field is never achieved. Level in draw-away tests may hold for a bit and then fall off with distance as in a free field. However, in a typical listening room, unless the ceiling is very low or the width and depth huge, you never get to 3 or more ceiling heights away from your speakers. (In my room I can just get to 2 room heights from one speaker with my nose in the far corner.) I believe that most listening rooms have dimension ratios closer to a typical concert hall than to the large office space Toole is refering to. Side wall reflections are very important in our living rooms and comparisons to "infinite" office spaces don't apply.

The point is that RT may not be simply calculated by the Sabine equations, and reverberent field level may be non-constant and may have a distance component, but that doesn't mean that some rooms are very reverberent while others are dead, and that quantity and quality of absorbtive materials, and room size, determine the measure of that reverberence. I'm not arguing that RT of a small room is primary in your perception of reproduced sound, like Floyd I feel that standing waves plus early reflections, their timing and spatial location are the important small-room issues. I just feel that the general acoustics of rooms are independent of room size (scaleable).

Regards,

David

[soundminded]

Hi Zilch,

I've re-read the Toole chapter on this.

First, I'm not comfortable with the term "Sabine spaces" or non-Sabine spaces". I believe he and Shultz are refering to the completeness of diffusion of the room and whether or not the Sabine equations give accurate RT calculations. Clearly, even a small room has a reverberent field (a complex sound field that builds up as a sound source is turned on and, when the source is turned off, dies at an approximately constant rate). Above the Schroeder frequency, RT is not meaningless, it just might not be that close to the number you get from Sabine's equation.

For a sound field to be anisotropic is not of concern either. At my last job we spent a lot of time designing lateral reflectors into concert halls to bathe concert goers with early lateral reflections. This is an instance of anisotropism(?) by design and it was considered very desirable.

Comparison of listening rooms to large office spaces is also troubling. In acoustics, office spaces are singled out as an important but very different case. If a concert hall can be simplified to a 6 sided box (certainly the rectangular halls can), then a large office is approximately a two sided space with no side walls. If the office is wide and deep enough then only the floor and ceiling surface matter. Studies show, as Floyd points out, that a constant reverberent field is never achieved. Level in draw-away tests may hold for a bit and then fall off with distance as in a free field. However, in a typical listening room, unless the ceiling is very low or the width and depth huge, you never get to 3 or more ceiling heights away from your speakers. (In my room I can just get to 2 room heights from one speaker with my nose in the far corner.) I believe that most listening rooms have dimension ratios closer to a typical concert hall than to the large office space Toole is refering to. Side wall reflections are very important in our living rooms and comparisons to "infinite" office spaces don't apply.

The point is that RT may not be simply calculated by the Sabine equations, and reverberent field level may be non-constant and may have a distance component, but that doesn't mean that some rooms are very reverberent while others are dead, and that quantity and quality of absorbtive materials, and room size, determine the measure of that reverberence. I'm not arguing that RT of a small room is primary in your perception of reproduced sound, like Floyd I feel that standing waves plus early reflections, their timing and spatial location are the important small-room issues. I just feel that the general acoustics of rooms are independent of room size (scaleable).

Regards,

David

"For a sound field to be anisotropic is not of concern either. At my last job we spent a lot of time designing lateral reflectors into concert halls to bathe concert goers with early lateral reflections. This is an instance of anisotropism(?) by design and it was considered very desirable."

Beranek agrees. Moreover, he is quite explicit that the quality of the lateral reflections are also important. In his paper on his web site, the number one correlation between 20 different measured performance variables of 59 concert halls and preference by "golden eared" conductors and others familiar with some of them was Binaural Quality Index or BQI= (1-IACC.) AFAIAC, all reverberant fields are literally anisotropic because they are composed of reflections that individually have direction but after the first few milliseconds of reflections near the source, they are subjectively isotropic, that is the directions of any of the reflections as distinct from the source cannot be determined by listening to them. If this is not the case, then there is a serious defect in the acoustics. This was my first clue that there was something very wrong with Philharmic Hall at Lincoln Center the first time I heard music there shortly after it opened. There was a distinct reflection of the trumpet off the left procenium arch. Later it came out that there were far more defects. Beranek has a very long and interesting explanation of why the hall was the fiasco it turned out to be (naturally not his fault.) Ultimately he was fired. I'm not sure it is much improved to this very day despite so many attempts to improve it. It's main problem seemed to be that it was too big because the managers specified the minimum number of seats they would accept in the design. Acoustics is in very large measure a matter of geometry.

In his Gugenheim lecture at the Mechanical Engieering department at Georgia Tech in 2001, sadly no longer available for free on the web, Beranek said that the maximum width of a concert hall should be not much more than 72 feet. This relates to the delays based on the geometry of the room and the speed of sound. It isn't clear why lateral reflections are important in enjoyment of music or if they play the same role in small rooms although experience with Bose 901 suggests it. He also said there should be 12 reflections reaching the listener within the first 100 milliseconds for good acoustics. Fortunately that number and many more are probably inherent in just about every commercial recording available. Experience shows that the shoebox geometry generally gives the best results (but not always as the hall at Oberlin College proved.) Fan shaped halls are usually less desirable because they provide fewer early lateral reflections.

I have found generalizations about acoustics can often be very misleading. But there aren't yet satisfactory methods of measurement to adequatly characterize them. I think the state of the art of measurement has a long way to go.

In large open landscape offices, transmission is usually the main problem, not reflections. Acoustic partitions made by hundreds of different companies have both absorbing material on the outside and transmission barriers internally. The typical ceiling is Armstrong 758 which provides fair absorption. Floors are usually carpet squares or low pile broadloom glued to concrete. Absorption of this combination leaves much to be desired. Metal stud and sheetrock walls for conference rooms or manager's offices where privacy is considered critical enough to justify added expense often get a second layer of sheet rock on firring strips and fiberglass batting inside the walls and even above the ceiling. An interesting product in this regard is something called "Quietrock." It has an unusually high resistance to transmission through it because of its unique patented construction. There's a web site for it.

[speaker dave]

This was my first clue that there was something very wrong with Philharmic Hall at Lincoln Center the first time I heard music there shortly after it opened. There was a distinct reflection of the trumpet off the left procenium arch. Later it came out that there were far more defects. Beranek has a very long and interesting explanation of why the hall was the fiasco it turned out to be (naturally not his fault.) Ultimately he was fired. I'm not sure it is much improved to this very day despite so many attempts to improve it. It's main problem seemed to be that it was too big because the managers specified the minimum number of seats they would accept in the design. Acoustics is in very large measure a matter of geometry.

Beranek is a bit of a hero of mine. I prize a signed copy of his latest book and studied his earlier "Music, Acoustics, and Architecture" religiously in college. The whole story of Philharmonic Hall is fascinating. Plenty of blame to go around, but ultimately it seems as though Beranek dropped the ball. All the drawing revisions were sent by the architect to BB&N and apparently ignored.

In his Gugenheim lecture at the Mechanical Engieering department at Georgia Tech in 2001, sadly no longer available for free on the web, Beranek said that the maximum width of a concert hall should be not much more than 72 feet. This relates to the delays based on the geometry of the room and the speed of sound. It isn't clear why lateral reflections are important in enjoyment of music or if they play the same role in small rooms although experience with Bose 901 suggests it. He also said there should be 12 reflections reaching the listener within the first 100 milliseconds for good acoustics. Fortunately that number and many more are probably inherent in just about every commercial recording available. Experience shows that the shoebox geometry generally gives the best results (but not always as the hall at Oberlin College proved.) Fan shaped halls are usually less desirable because they provide fewer early lateral reflections.

I'm aware of the maximum ideal width but haven't heard the "12 reflections in the first 100 msec" rule. Do you know what the derivation of that is?

Regards,

David

[kkantor]

It seems to me based on the kind of music you've said you like best, you'd do well to listen to the same kinds of speakers used to produce it in the first place. Community Sound, JBL, Altec, even some of those old Crowns sound men dragged around from rock concert to rock concert. I'll bet they didn't use AR3a, AR303, or NHT3.3 at the Paradise Club.

Agreed. However, the speakers I should use are studio monitors, and especially those used for mastering. The 3a, of course, employed frequently for this, as was the 3.3. There were a few studios that liked the 303a, too, but not nearly as many as the other two.

Clearly, I have been a speaker addict for much of my life. As an audio hobbyist, I was quite promiscuous and tended to have at least three pairs up at once. I've owned 2ax, 3a, LST, 10pi, 12, 9, 4ax, 18, M1, 48, 91, others. And Advents and BA's and Wharfedales and KEF's and Quads and 801's and 901's and 824's and Dynaudios and Gelelecs and Infinity's and TOA's and NS1000's and 4312's and A20's and University's and Altecs and and and. Never found true love. Had to try and build it.

Over the last 40 years, I've heard a hell of a lot of speakers, both on the market and in R&D labs. At AR and NHT, we were always bringing in competition. When I was writing for mags, I got to hear things in that context. Working for a driver supplier, Tymphany, (Peerless, Vifa, ScanSpeak), meant I was exposed to many different products from our OEM customers. Add in trade shows, dealer visits, friends, etc, and I think I have heard a good range of what is out there. Still haven't found true love. But I have had lots of eargasms.

-k

[kkantor]

Regarding his test methodology, I feel he has been consistent and very thorough over the decades. Subjective rankings, by definition involve listener taste. Early NRC tests showed that a variety of listeners came to repeatable rankings of loudspeakers in their blind tests. Later tests showed that mono vs. stereo was not a strong variable in the rankings. (The ESL 63 was one exception. It was ranked poorly in mono tests and improved to below average in stereo tests.) Other tests showed that room acoustics weren't a strong variable. If I'd come up with any test where a wide range of listeners consistently prefered certain speakers and disliked others, and if I found that the room wasn't a factor in the rank ordering, I'd be pretty happy with that test.

In my opinion, there is a subtle and important issue about Floyd's landmark "preference" work: he sorts things into ranked preference >groups<, not a discrete preference >list<. Within a preference group, which may contain a fair number of very different speakers, Floyd does not attempt to rank listener preference. In other words, several very different design approaches could all wind up in the top preference group, but still sound very different from one another. The implications of this are not trivial. Floyd shows us a way to assure a good sounding speaker, but also leaves wide open the designer's and listener's comparative preferences between speakers within this class.

It's possible that I am incorrect in my memory of the test protocol, since it has been a long time since I read the original paper, or discussed this with Dr. Toole in detail.

-k

[soundminded]

Beranek is a bit of a hero of mine. I prize a signed copy of his latest book and studied his earlier "Music, Acoustics, and Architecture" religiously in college. The whole story of Philharmonic Hall is fascinating. Plenty of blame to go around, but ultimately it seems as though Beranek dropped the ball. All the drawing revisions were sent by the architect to BB&N and apparently ignored.

I'm aware of the maximum ideal width but haven't heard the "12 reflections in the first 100 msec" rule. Do you know what the derivation of that is?

Regards,

David

I got the sense from Beranek's lecture that the 12 reflections rule was based on experience.

If there was one person I'd like to meet it's Leo Beranek (I don't even know if he's still alive.)

Beranek's lecture was the most interesting I've ever heard about sound and music related to acoustics. He explained how Sabin invented the modern science of acoustics at Harvard where he taught days and did research at night to develop his famous formula. In those days, the only thing he had for measuring time was a stopwatch.

He also spoke at length about many concert halls and opera houses around the world. According to him, the best concert hall is Musikverein in Vienna. Concertgebeouw in the Netherlands and Boston Symphony Hall vie for second place. In the survey he took, BS hall was third behind Concertgebeow. There's also a very fine one in Tokyo. The best Opera House is in Buenos Aries. The criteria for Opera Houses are somewhat different. He didn't think much of Carnegie Hall or La Scala. Someone I knew who was familiar with Carnegie Hall told me the acoustics were ruined when the NYC fire marshall told the owners to replace the oil based paint with latex paint. That could do it.

It seems to me based on a book I own published by ASA which has much information of both before and after data for a couple of hundred concert halls around the world that were renovated between 1962 and 1982, that the trend is towards less reverberant halls. Personally, I prefer the greater reverberation times but modern music often sounds blurred in such halls. Typical mid frequency RT targets are now 1.8 to 2.0 seconds for concert halls, 1.6 seconds for Opera Houses when occupied. Some are even less. Targets for RTs at 8 khz are 1.2 to 1.0 seconds or even less.

I used to be a member of ASA and to become one, of AIP as well. I had considered presenting several papers on my theories but thought the better of it. I'm not in the publish or perish mode and don't see what good it would do me except to give away intellectual property.

[Zilch]

Let's do a "True Loudspeaker Love for Ken" poll....

[speaker dave]

If there was one person I'd like to meet it's Leo Beranek (I don't even know if he's still alive.)

He still is, as far as I know, living in a waterfront condo in Cambridge across from Boston. I don't know him other than being iintroduced once. A friend of mine who is involved in the Acoustical Society sees him fairly regularly.

David

[soundminded]

He still is, as far as I know, living in a waterfront condo in Cambridge across from Boston. I don't know him other than being iintroduced once. A friend of mine who is involved in the Acoustical Society sees him fairly regularly.

David

From what I can tell, Beranek is the number one acoustician in the world. He has as much first hand knowledge and experience as anyone else I've heard of, probably more. I wonder if he'd be interested in my unique method of acoustic measurement and the equations behind the theory that rationalizes it.

If someone handed me a one hundred million dollar project and said you are in charge of building a concert hall for our organization, I'd go to the number one man and ask him to design an exact copy of the number one hall right down to the last detail. For some reason, every architect and I suppose every acoustic architect wants to make a "statement" with a unique innovative design. While some are striking to look at, they can turn out to be acoustical disasters. Boston Symphony Hall was modeled after Leipzig I think, a very successful design. From what I gathered, this hall was damaged or destroyed in the war but has been rebuilt and restored. How many times have I heard an architect talk about "statements?" Given that acoustics is still a relatively primative science, every new deisgn is an experiment and many of them fail. Even the traditional shoe box design can fail as Oberlin demonstrated.

I had wondered why the lecture was given under the auspices of the mechanical engineering department at Georgia Tech. It only hit me recently that whether those who study acoustics know it or not, it is a branch of mechanical engineering, not electrical engineering. It is a specialized area of fluid dynamics, the fluid of course being air. How fortunate for me to have taken a very rigorous course in this field. One day I will dig out my old textbook and see if my equations dovetail with the classical equations such as Newton's laws as partial differential equations (mine are in the integral form.) They should.

Another related area that is not well understood is psychoacoustics, a specialized area of clinical psychology. How fortunate I took a rigorous course in psychology too as a humanties elective. Those I paid for an engineering education did a particularly fine job, I got my money's worth. I still haven't managed to finish Dr. Oliver Sach's book "Musicophilia" which describes anecdotally the relationship between music and human biology. The effects are many and are being documented. Understanding of how and why these effects occur is not understood yet. I'll bet FMRI (functional magnetic resonant imaging) will be a useful tool to study it.