John Dunlavy's posts to rec.audio.* during 1998:
New messages will be added here as I see them on Usenet; hopefully in less than a week. Thanks to Dunlavy for this information.
Last revised 6 December 1998. 



--------------------------------------------------------------------------------


From awrigby@aol.com Thu Mar 26 11:25:15 1998
Newsgroups: rec.audio.opinion
Subject: Loudspeaker Accuracy
From: awrigby@aol.com (AWRigby)
Date: 26 Mar 1998 17:25:15 GMT

I thought that the readership of rec audio opinion might find the following article recently composed by John Dunlavy interesting:

_____________________

Funny, isn’t it, how a simple word like "accuracy" can acquire so many different meanings, interpretations and questionable uses within the audiophile community. The relevance of this observation becomes obvious after examining only a few of the claims for accuracy being made by several loudspeaker manufacturers! It is difficult to imagine how such baseless, untrue claims can serve the best interests of our audiophile community. 

On the bright side, the retail prices being charged for audiophile components such as CD transports, power-amps, and pre-amps have generally fallen during the past five years, while their performance has risen to almost flawless levels. But, with few exceptions, loudspeakers have not followed the same trend with respect to more accurate performance at lower prices.

So, why has the development of more accurate loudspeakers not kept pace, generally, with that of other audiophile components? Certainly, neither a lack of technical/scientific knowledge nor a lack of precision measurement equipment (such as Doug Rife’s MLSSA system) can be blamed. Indeed, the teachings of engineering, physics and acoustics are comprehensive, well-known, and available to all competent designers.

However, despite the availability of technical knowledge and measurement equipment, few loudspeaker designers and manufacturers appear willing to invest the time, talent and money required to design and manufacture products with true, "state-of-the-art" performance. As a consequence, not many of today’s audiophile loudspeakers (regardless of price and claims for accuracy) exhibit truly accurate performance in both time and frequency domains. Indeed, it is the exception to find a loudspeaker with respectable impulse, step, waterfall, energy-time, excess phase, etc. responses. And, despite advertising claims, very few loudspeakers exhibit a frequency response (without "smoothing) better than about plus/minus 3 dB across their advertised range. 

As an alternative to making and publishing measurements, some well known manufacturers (whose loudspeakers measure poorly) have taken to claiming that only "subjective listening evaluations" can establish the "true" accuracy of a loudspeaker. To them, only listening sessions (usually with few or no controls) can reveal whether a loudspeaker is "truly accurate". Indeed, some designers of loudspeakers that are at the most expensive end of the price spectrum, claim that they use a technique called "voicing" to design their products - believing that their ears are a better judge of true accuracy than any set of measurements. But are their ears a "standard" that can be relied upon? And does the listening room within which they voice their loudspeakers have the same acoustical properties as the room within which the purchaser will ultimately listen to them? And how about the designer’s choice of recordings - will their miking, E.Q. and other properties match those of the purchaser’s favorite recordings? Hmmm!

Let’s approach the accuracy question from a different perspective: simple logic reveals that the "true audible accuracy" of any loudspeaker can never "exceed" that predicted by a complete set of anechoic measurements, accurately made and competently interpreted. Such measurements should include, at a minimum, frequency response, impulse response, step response, waterfall, harmonic/IM distortion levels Vs SPL across the spectrum, and vertical and horizontal plane radiation patterns, all made at a distance of from 9 to 12 feet (typical listening distance). Further, measurements of impedance Vs frequency (including both resistive and reactive components) are important for determining whether the loudspeaker represents an easy "load" for the intended power amplifier. (A loudspeaker with an input impedance that drops to 2 ohms or lower at some frequencies might not be appropriate for some power amps - especially tube types.)

But, while a full set of measurements can, properly made and interpreted, predict the potential of a loudspeaker to yield "audibly accurate reproduction", verification of the audibility side of the equation must rely upon carefully-controlled live Vs record/playback sessions using musical instruments, voices, etc. Such sessions are very difficult to control, require a large and expensive anechoic chamber for making the recording (to eliminate room acoustics from the recorded sound), a listening room that will not bias the sound of either the live musicians (during comparisons) or the sound reproduced by the loudspeakers. Although careful planning and suitable controls can yield successful live Vs record/playback evaluations, it is an arduous and time consuming task - though a revealing and rewarding one.

So, if both measurement technology and live Vs record/playback evaluation methods are available, why don’t more loudspeaker designers and manufacturers utilize them both when designing their products?

One answer, perhaps, is that it is much easier and more "cost effective" to design an attractive loudspeaker enclosure with a "hi-tech" appearance, stuff it with expensive looking drivers and advertise it using language and claims with a high-tech flavor. Since there is no shortage of legitimate, meaningful terms and language for describing the properties and performance of loudspeakers, the use of "off-beat" gobbledygook terms and expressions should be a warning that the designer is either not qualified or deserves a Nobel Prize for his discoveries. Hmmm!

Maybe more audiophiles, sincerely concerned with accuracy, should ask whether the manufacturers of loudspeakers they are considering purchasing have the requisite engineering staff and measurement facility/equipment needed to ensure that they have been properly designed and their performance adequately measured and documented. If they cannot provide well-documented, guaranteed performance specifications and complete measurements for their products, perhaps one should question why they are not available.

Having said all of the above, an argument exists for loudspeakers whose reproduction properties satisfy the listening preferences of audiophiles concerned less with "true accuracy" than obtaining reproduction best described as sounding "lush, sweet, musical, involving, etc." And, as long as the manufacturer of such loudspeakers does not claim them to be "accurate", there is nothing intrinsically wrong with catering to this market. The "honesty barrier" (or truth in advertising) is breached only when a designer or manufacturer claims inaccurate loudspeakers to be accurate.

The bottom line is simple: true accuracy can only be said to exist when a loudspeaker measures accurate and has passed a rigorously-controlled, "blind", comparison with live music. The recording of the musical group must be made within a suitably "anechoic" environment, using instrumentation-quality mics, and a precision DAT recorder with ruler-flat frequency response and near-perfect impulse/step responses. The live-Vs-recorded comparison must then be made in a reasonably large room possessing well-damped (but not anechoic) acoustical properties. The musical group (preferably a trio or quartet) must be located directly between the loudspeakers and at a distance from the listener that is approximately equal to the distance of each of the loudspeakers from the listener. When these conditions have been met, truly accurate loudspeakers should provide reproduction that is indistinguishable from the live musicians - even fooling those with "golden ears".

So! What does all of this talk have to do with "accuracy". I believe it indicates a pressing need for the audiophile loudspeaker industry to develop a set of definitions for describing the properties and accuracy of loudspeakers, that can become a standard for use in advertising and marketing.

In this regard, I would suggest that loudspeaker accuracy might be separated into categories such as 1) subjectively accurate (pleasing/satisfying) and 2) fully accurate (measurably and audibly).

"Subjectively accurate (pleasing/satisfying)" would apply to loudspeakers that provide reproduction judged to be subjectively accurate to the average, but non-critical, listener - without any live-vs.-recorded comparison. 

"Fully accurate" would pertain to loudspeakers that exhibit accurate measured performance, with respect to all important measurable properties, and which have passed a rigorously-controlled, "blind", comparison with live music.

I suspect that more than a few readers (and designers) will probably find fault with the above suggestions regarding a rating system for loudspeaker "accuracy". Fine! Lets hear, consider, and evaluate their comments and ideas.

But the time has come to overhaul the audiophile loudspeaker industry (including manufacturers, mag reviewers, ad agencies and dealers) in a manner that creates a higher level of honesty in advertising and promotion of products. (The same can be said of the audiophile cable industry - where "truth-in-advertising" has reached an all-time low!)

By developing more accurate loudspeakers and introducing more accurate advertising claims to sell them, I believe we will succeed in recapturing the imagination of a large segment of audiophiles who have become increasingly disenchanted with high-priced products that do not meet expectations.

Best Regards,

John Dunlavy CEO Dunlavy Audio Labs

_______________________________________________________

From 102365.2026@compuserve.com Tue Apr 14 11:06:29 1998
Newsgroups: rec.audio.high-end
Subject: Ten Commandments: was Loudspeaker accuracy
From: Dunlavy Audio Labs <102365.2026@compuserve.com
Date: 14 Apr 1998 16:06:29 GMT

Please approve the following post composed by John Dunlavy for inclusion on rec audio high end under the thread "The Ten Commandments; was Loudspeaker Accuracy":

My! My! John Otvos is at it again! His posting of 3 April under the subject "The Ten Commandments" was directed at DAL and myself. Much of what he said needs to be addressed in a proper manner.

In this regard, I take no pleasure in correcting the language, expressions and terms used by some people to wrongly describe technical things about which they know or understand little. I believe that those who know me will attest to the fact that I am not, by nature, an arrogant person (as some have called me). Rather, I am merely an engineer/physicist driven by a desire to discover the real truth about how things work and to share the fruits of the knowledge I have gained with others. I simply have no other agenda. So, if I have ever appeared arrogant in any of my postings - mea culpa!

Perhaps, J.O. is merely mistaking my insistance that engineering, physics and acoustics have a lot to teach us about loudspeaker design, performance, measurements, audible properties, and so forth, for "arrogance". But for J.O. to state that measurements are essentially meaningless and that only "blind-A/B" listening comparisons in a darkened room can reveal the true accuracy of loudspeakers is tantamount to expressing belief in the old "flat earth concept" or that "the earth is at the center of the universe". Hmmm!

Indeed, to deny the usefulness of a "full set of accurate measurements" (made at the normal listening distance of 10 feet) for determining the "potential" of a loudspeaker to exhibit audibly accurate performance is like denying the usefulness of a dynamometer for evaluating the performance of an automobile engine.

But, for those lacking appropriate engineering/physics credentials, I suppose it is easy to assume that only subjective listening comparisons can establish the accuracy of loudspeakers. Funny, though, how simple a loudspeaker looks but how complex its operation really is? Equally disquieting is the habit some non-technical persons have of attempting to interpret complex technical things according to a confused view of science and engineering principles, etc. Such deceit is made even worse when they attempt to ridicule those who do possess competent professional credentials.

It may be relevant to remind readers of the old maxim: "It is better to only be thought a fool than to open ones mouth and remove all doubt". Harsh? A bit, perhaps, but let us examine what J.O. and a few others have had to say about myself, DAL and the design properties of our loudspeakers.

J.O. begins his post by stating: "To save a soul, I can't understand why there is such a reluctance on the part of supposedly passionate audiophiles to do side by side comparison tests with matched levels and lights out. It's almost as if many people don't want to know the truth."

No! J.O., it has to do with the fact that most experienced audiophiles (and professional designers) are fully aware that such subjective comparisons reveal little with respect to determining the true accuracy of loudspeakers. Why? Lets examine the issue in an objective manner.

1) It can be readily demonstrated that two identical loudspeakers, located side-by-side within a typical listening environment, will almost always exhibit different audible properties due to standing waves created by room boundary reflections. Therefore, performing A/B comparisons between two identical loudspeakers, physically adjacent to one another within a typical room (even one possessing "good acoustics"), will almost always lead a competent listener to swear that they exhibit totally different audible properties.

2) The acoustics of the listening room and the locations of the loudspeakers within the room can considerably alter the perceived "accuracy" of the loudspeakers. Indeed, simply transposing the locations of two loudspeakers can totally alter which one "sounds the most accurate". This is because the "acoustical standing-waves" that exist in every normally reverberant listening room vary considerably in relative amplitude and phase from one room location to another. Indeed, moving a loudspeaker no more than a foot or two can often alter its audible properties to a considerable degree.

3) Then, there are the natural biases that many listeners have that favor their personal subjective preferences for such such properties as "sweet sound, pretty sound, forward sound, rich sound, full-bass, etc.", that seldom relate to the accurate reproduction of complex musical transients, imaging, etc. Thus, results obtained from such poorly-controlled subjective comparisons within a typical home listening room are usually worthless.

These are but a few of the many reasons that we at DAL believe that a full set of accurate anechoic room measurements made at the normal listening distance of about 10 feet can, properly interpreted, are necessary for predicting "the potential of a given loudspeaker to accurately recreatethe original musical sounds". But, measurements made at distances of only 30 to 50 inches (especially if they are made within a non-anechoic environment) seldom if ever, correlate very well with audible accuracy, except for loudspeakers having small physical dimensions, e.g., less than about 18 inches in any plane.

J.O.'s insistence upon the relevance of what he refers to as "drag racing" between loudspeakers holds little appeal to a competent engineers and physicists. As mentioned briefly above, subjective listening comparisons between loudspeakers within typical rooms can only determine the "subjective appeal" that one loudspeaker possesses compared to another. It can seldom (if ever) determine the "true audible accuracy" of any loudspeaker. Indeed, in the early days of hi-fi, several companies designed and marketed, with considerable success, loudspeakers that exhibited what soon became known as "pizzazz sound"! And they sold by the car load --- until people discovered that their appeal was not lasting and that what they were really searching for was a more accurate and faithful level of musical reproduction.

The same can be said for loudspeakers that have radiation patterns that "spray sound in all directions". Such speakers can enthrall some listeners with their ability to create a sound field that surrounds the listener - but imaging is non-existant and the location of individual instruments cannot be ascertained. As a consequence, although such loudspeakers experienced a rapid rise in fame (and riches for their manufacturers) they never acquired a sustained appreciation among knowledgable audiophiles.

But some such "pizzazz" loudspeaker attributes appear to be making a comeback among those audiophiles more concerned with "effect" than genuine, verifiable, measurable and audible accuracy. Sigh! The old expression, "Vy ist it dat ve grow so soon old und yet so late schmardt?" seems appros! I do, of course, fully agree with J.O.'s comment that, "It is absolutely crucial for enthusiasts to understand the basic underpinnings of established acoustic theory so that they will have a more informed knowledge before commiting hard earned money for a new purchase." It is unfortunate that J.O. does not appear to possess such knowledge, yet seems to infer that he does. Hmmm!

And, with respect to veracity, lets examine some of J.O.'s published comments. For example, it is interesting to note that in the Stereophile magazine review of J.O.'s Waveform Mach 17 loudspeaker (June 1997 issue) Larry Greenhill states that, "... the Mach 17 has been adopted as a recording monitor by the Telarc, Delos, and Dorian record labels".

A check with Craig Dory of Dorian to determine his view of the Mach 17 as a recording monitor or "reference loudspeaker" revealed that Dorian did not consider Waveform loudspeakers to be sufficiently accurate for such use. Indeed, Craig said he will attempt to post his feelings about the matter in the next few days. He also stated that Dorian uses DAL SC-IVs as their only "reference monitors" - not Waveform loudspeakers, as inferred by J.O. (I believe Dorian now owns three pairs of SC-IVs, one pair of which they keep boxed for shipment to domestic and overseas recording venues.)

It is also true that the SC-I (a mini-monitor), the SC-IV and the SC-V are the monitors of choice for many other top recording and mastering studios in the United States and overseas. And, major mastering studios such as Sterling Sound which use SC-Vs as their reference monitors and Absolute Audio, which use either SC-IVs and or SC-Vs as their reference monitor loudspeakers. And the list of major recording labels that use DAL loudspeakers is too long to recount here.

Perhaps, John Otvos can provide readers with a more accurate accounting of just what recording and mastering companies actually use his loudspeakers as their "reference monitors". Hmmm!

Readers might also like to know that Doug Rife, the man behind the design, etc., of the famous and incredibly accurate MLSSA loudspeaker measurement system, now used by nearly all serious loudspeaker designers, uses a pair of SC-Vs, an SC-IV and a pair of SC-Is for his personal stereo and home theater system. Doug could have chosen any brand he wished but decided upon DAL loudspeakers because of their measured and audible accuracy.

And, perhaps, J.O. might wish to compare the many glowing reviews and meaningful awards that DAL (and earlier, Duntech) loudspeakers have earned since 1976 within many of the most prestigious audiophile magazines throughout the world. Can J.O. produce such a list. Hardly! If anyone would like a copy of the multi-page list of these reviews and awards, its free for the asking.

Let's now consider the "Ten Commandants" posted by J.O.:

1) "Flat measured response, 20 Hz to 20 kHz on axis within a variation of only 2 dB, i.e., +/- 1 dB (this is the goal although not always realized)."

This is the "spec limit" that every DAL loudspeaker must pass (at a distance of 10 feet within an anechoic chamber) before being accepted for shipment. By contrast, Waveform loudspeakers, at least the Mach 17, do not appear to come even close - as may be seen from the measurements of the Waveform Mach 17 beginning on page 129 in the June 1997 issue. The frequency response measured by J.A. shows variations of about +/- 2.5 dB from 100-20,000 Hz. Fig. 7 on p. 157 of the June 1997 issue of Stereophile Magazine's review. For comparison, see the measurements of several models of DAL loudspeakes that have appeared in Stereophile, Audio, etc., over the past 22 years.

2) "Smooth room response without large peaks or dips." 

The room response (or total radiated power as J.O. also calls it) is normally taken to mean the total power radiated by a loudspeaker integrated over the surface of a hypothetical sphere of very large radius. The Waveform loudspeakers do not exhibit a smooth room response as inferred by J.O., as may be seen from the off-axis responses measured by John Atkinson in the aforementioned Stereophile review.

3) "Controlled wide dispersion, ie., smooth response off axis, ... with respect to the first arrival, up to 60 degrees horizontally. ... " J.O.'s perception that the radiation of a loudspeaker at 30 degrees off-axis, horizontally, "is most responsible for the reverberant energy one hears in typical listening rooms. Hmmm!

What wall boundaries usually exist only 30 degrees off-axis of a loudspeaker a few feet from a side wall, with the loudspeaker pointed toward the listening position? Very few I suspect!

4) "THD must be negligible, i.e., less than 1% below 85 dB continuous..". 

I agree with this spec and all DAL loudspeakers meet it.

5) "The employment of an active electronic crossover is essential, to reduce intermodulation distortion, ...".

Nonsense, properly designed passive crossovers using high quality components (no iron/ferrite cored inductors or electrolytic caps) do not create non-linear distortion products at any reasonable SPL's.

6) "A flat listening window, ..."

I agree, except that I would add the need for vertical and horizontal radiation patterns to be precisely symmetrical about the intended listening axis, with the vertical plane pattern being narrower than the horizontal to minimize the often undesirable reflections from the floor and ceiling surfaces. It is worth noting that Waveform loudspeakers do not meet this criteria, as may be seen from the measurements appearing in the aforementioned Stereophile review. (All DAL loudspeaker models provide symmetrical patterns about the listening axis in both horizontal and vertical planes.)

7) "The vertical dispersion of the speaker system must also have a similarly smooth response characteristic within a 0 to 30 degree envelope"

I generally agree, but would add the desirability of the radiation to be symmetrical about the listening axis. The Waveform Mach 17 exhibits an assymetrical radiation pattern in the vertical plane, with a null of about 24 dB at only 10 degrees above the horizontal axis. Hmmm!

8) "To manufacture a loudspeaker with high power handling and high sensitivity. All Waveform systems must have an anechoic sensitivity of 90 dB for 1 V in at 2M as a maximum."

J.O.'s rating of sensitivity relates to the "input voltage into the loudspeaker's integral power amps required to achieve the rated SPL at 2M as a minimum" (as J.O. calls it). This is misleading because it does not tell the purchaser how much actual power is required for the array of drivers to attain the 90 dB SPL - an important spec for determining the power-handling of the loudspeaker.

Further, the 24 dB/octave Linkwitz Riley crossover (I thought they were usually 12 dB/oct., with the driver polarities reversed to obtain an optimum vertical pattern) cannot correct for the mis-alignment of the drivers in the "time-domain" along the listening axis, as claimed in the Stereophile review.

9) "The Waveform loudspeaker must be a benign impedance load for a wide range of amplifiers. ... We use ported systems, the most effective damping is provided by solid state amps for those who want true accuracy and are not interested in screwing up the impedance of the speaker with the high output impedance of a tube amp."

The input impedances presented to the separate bass, mid and tweeter amps required for the Mach 17 are simply typical impedances exhibited by most loudspeaker drivers. Hmmm! (It is also doubtful if any amp can accurately control the damping of a bass driver at its 15 Ohm resonant peak at 60 Hz, a rather high frequency for a port/woofer resonance.)

10) "The room. Perhaps as important in some respects as the first commandment, although not a part of the speaker they constantly work in tandem with each other, and are considered in the design protocol."

Since all loudspeakers interact with rooms by virtue of reflections from the walls, floor and ceiling, their radiation patterns are usually an important aspect of a loudspeaker's audible performance. However, it appears that the use of a 24 dB/octave crossover in the Waveform Mach 17 hardly represents a slope capable of reducing the levels of off-axis radiation that especially might create problems with respect to floor and ceiling reflections. This seems to be evident from the curves published in the Stereophile review of the Mach 17. Hmmm!

Anyway, its now 6 PM and I need to head for home. Perhaps, time will permit further comments related to the current thread early next week. Keep the faith! Don't be confused by pseudo engineering and physics - they seldom work as claimed by some designers. Search for the "time-tested" truths available from the teachings of legitimate engineering, physics and acoustics.

Best of listening,

John Dunlavy

From 102365.2026@compuserve.com Wed Apr 29 10:07:14 1998
Newsgroups: rec.audio.high-end
Subject: Loudspeaker Accuracy, Part 2
From: Dunlavy Audio Labs <102365.2026@compuserve.com
Date: 29 Apr 1998 11:07:14 -0400

Sorry that I have not had the time to participate in the "accuracy" discussion the past few days but running a good-sized company can be bit demanding at times!

Anyway, I have enjoyed reading some of the comments about loudspeaker "accuracy" and what it means to different people. Most of the comments seem to focus on either "objective" or "subjective" definitions of accuracy, with little middle ground.

Another controversial subject being discussed concerns the radiation patterns of loudspeakers, their interaction with room boundaries and how they affect imaging and our perception of realism. In this regard, there seems to be a lot of divergent opinions as to what horizontal and vertical patterns might be best for obtaining the most audibly accurate reproduction of the original musical event. So, perhaps a few comments might help bring key aspects of the discussion into sharper focus.

As has been previously suggested, the terms "objective accuracy" and or "true accuracy" should probably be reserved to describe accuracy that has been properly documented by a complete set of precise anechoic measurements (competently interpreted), coupled with carefully-controlled listening comparisons between the loudspeakers and a small group of musicians (quartet, quintet, etc.) playing within an acoustically suitable environment.

"Subjective accuracy" might be reserved to denote the level of "perceived accuracy" determined by a typical audiophile while listening to familiar music within a familiar listening environment possessing reasonably good acoustical properties.

But neither of these definitions may satisfy those who have their own notion (or agenda) as to what accuracy means or should mean.

Personally, I believe that a complete set of measurements interpreted interactively with rigorously-conducted listening sessions (preferably involving comparisons with live musical instruments), are required to reliably establish whether a given loudspeaker is "truly accurate". Thus, if a loudspeaker is truly accurate, a listening session will verify the accuracy predicted by measurements and vice-versa. However, if inaccuracies exist, measurements should reveal them and listening should further confirm their presence. This, I believe, represents both good science and good engineering.

With respect to claims that loudspeakers should exhibit broad (perhaps even omni-directional) radiation patterns to simulate the radiation patterns of most musical instruments, so as to yield reflections from listening room boundaries that more closely replicate those of the original live music, the following observations are relevant.

1) Virtually all musical instruments, including most horn and string types, do not exhibit the nearly omni-directional radiation properties envisioned by those inexperienced in this field. Instead, their radiation patterns reveal reasonably significant directivity over most of their frequency range. In fact, the -3 dB beamwidth (in degrees) of the major radiation lobe of any radiating aperture (assuming a uniform distribution of energy over the radiating area) can be determined with useful accuracy by dividing 50.8 by the width of the radiating area expressed in wavelengths at the subject frequency. To say or imply that most musical instruments exhibit a radiation pattern that is essentially spherical is simply not true.

2) If loudspeakers possess wide or omni directional properties, a large percentage of their radiation will reflect from walls, ceiling and floor boundaries, creating multiple signal paths which, when they combine with the direct-arrival sound at the listening position, create a frequency spectrum comprising a series of peaks and nulls (comb filter type response). Our ear-brain combo sub-consciously processes this spectrum and uses the information to interpret the audible properties of the listening room. When this spectrum is superimposed upon the spectrum of the recording venue, the two add vectorialy, often producing a combined spectrum of peaks and nulls that the listener may find either confusing or un-natural (although some listeners may hear such a combination as audibly pleasing).

3) Most "live sounding" recordings being made today have been "stereo-miked" to preserve the highly-valued acoustical ambiance of the recording venue, be it a concert hall, a dance hall, large recording studio, etc. Of course, some recordings are multi-miked within a relatively dead studio environment, with "ambiance" introduced later by electronic reverb, etc. during the final mix by engineers at the mastering studio. As a consequence, most accurate stereo recordings have the acoustical attributes (and proclivities) of the recording venue "imbedded" within the recording. The most accurate reproduction of such quality recordings within a typical home environment can only be accomplished by using a stereo pair of loudspeakers that exhibit symmetrical radiation patterns in both vertical and horizontal planes, with beamwidths that minimize the amplitude of reflections from walls, ceiling and floor boundaries.

4) A company known as Sigtech makes a digital room-EQ system which has achieved commercial success for use in "eliminating" the effects of reflections from wall, ceiling and floor boundaries within troublesome listening rooms. If such reflections were a desirable attribute, it is doubtful that Sigtech's success would have been realized.

5) There are also a number of successful companies that design and sell various damping and or dispersive materials for treating both home and studio listening rooms, with the object being to reduce audible reflections that might overlay the acoustics of the concert hall, etc.

6) During the late 1970's, when I had Duntech Labs, we conducted several experiments related to listener's perceptions of loudspeakers with a wide variety of vertical and horizontal dispersion patterns. This resulted in the design of a loudspeaker we called the DDRL-3, a time-aligned, minimum phase system which had remotely switchable radiation patterns, permitting a listener to switch between a horizontal beamwidth of about 45-60 degrees and a nearly omni-directional pattern, while retaining precisely the same on-axis frequency response. We exhibited the DDRL-3 at the Showboat Hotel during the 1979 WCES in Las Vegas. Acoustically, the listening room was reasonably good, probably better than most home rooms. The audible differences between the directional and omni-directional modes were apparent to everyone who participated. The conclusion: the omni-directional pattern produced a blurred stereo soundstage, with relatively poor imaging, along with a shift in spectral balance induced by the additional room reflections.

With the above in mind, why would anyone truly serious about accurate reproduction suggest that a loudspeaker with very wide radiation patterns might create a more accurate reproduction of the original music than a loudspeaker whose radiation patterns have been carefully engineered to reduce listening room reflections to levels that will not normally override or audibly conflict with those of the original recording venue (or those added by the mastering engineer)?

A number of recent posts have questioned claims I have made regarding our ability to record an 85-piece symphony and compare the reproduced sound to that of the live symphony. I continue to stand by my claims for the accuracy we were able to achieve. Several well-qualified and discerning audiophiles, including an audiophile magazine editor, two symphony conductors, two Ph.D. EE's etc., have witnessed the recording and playback demo - first-hand. I believe that at least two of them, John Pessetto (of HP) and Mike Guyote, PH.D., have posted their first-hand observations - having been present during both recording and playback sessions. But, I suspect that if a thousand competent listeners confirmed what I have said, it would not be sufficient to convince those whose minds are made up and who want to believe that such accurate reproduction is impossible. Hmmm!

And - yes, I am well aware of the claims being made that AR performed the same stunt with their AR-3's, etc. back in the 1950/60's. But I was present at one of their sessions and can attest to the fact that the venue did not permit any meaningful comparisons to be made because of the lousy acoustics (within NYC's noisy Grand Central Station). And, at the session I attended, there was never any switching between the live and recorded sound - just an inference that we were listening to live music when we were really listening to the AR speakers, which (as I recall) surrounded the musicians. "A neat trick", but hardly a suitable demonstration for making an informed decision regarding true accuracy.

With regard to claims for obtaining accurate stereo imaging over a wide listening area with loudspeakers exhibiting a very broad or near omni-directional radiation pattern, the laws of physics and acoustics seem to deny the possibility. The reason is simple: thus far, no loudspeaker has been designed that exhibits the asymmetrical off-axis radiation patterns, with the required coherent variations in phase and amplitude Vs azimuth, necessary for achieving such a result. While it seems possible to design a loudspeaker with a skewed radiation pattern possessing the necessary off-axis amplitude-frequency Vs azimuth properties, it is beyond the present state-of-the-art to also achieve the off-axis variation in phase Vs azimuth needed to achieve accurate and stable off-axis stereo imaging. (Any competent designers want to comment on this?) The reason that both phase and amplitude variations Vs azimuth are important is that we use both components to determine the direction of a given sound, not amplitude differences alone. Indeed, experiments where the phase of the two stereo channels was "digitally dithered" resulted in listeners hearing a very diffuse and imprecise soundstage, with a center-channel vocalist appearing to emanate somewhere within a fairly wide arc between the two loudspeakers. When the original relative phase components were re-introduced, pin-point imaging was restored for a listener located equidistant from both loudspeakers.

Further, the use of loudspeakers employing 4th-order networks results in a very broad vertical directivity pattern, causing high-amplitude reflections from floor and ceiling surfaces that can considerably alter the perceived spectrum heard by a listener. Such reflections also typically blur the sound of musical transients, etc. However, some listeners (without a standard for comparison) interpret such blurring as pleasant and pleasing. 

By contrast, a loudspeaker employing a properly designed, vertically-symmetrical array of drivers, fed by a 1st-order crossover, exhibits a relatively narrow lobe in the vertical plane that minimizes the effect of floor and ceiling reflections by reducing them to virtually inaudible levels at normal listening distances. By reducing or eliminating the effect of such reflections, the audible accuracy of the reproduced sound is considerably enhanced - although some less experienced listeners might find it less appealing. Gulp!

With their one-cycle, 360 degree delay at the crossover frequency, 4th-order crossover networks may also create audible waveform distortion of complex musical transients whose fundamental frequency is below the crossover frequency and its harmonics fall above it. This seems to be confirmed on page 203 of the book "Hearing - Its Psychology and Physiology" (published by the American Institute of Physics for the Acoustical Society of America), which states: "Not only does the phase of a harmonic present in the stimulus have an effect upon the threshold for distortion, but it may also influence the subjective effects of a complex tone." And, on page 229 of the same book, we find, "Now, the role of phase turns out to be unexpectedly crucial in these considerations, despite the well-documented doctrine that the ear does not take account of phase-relations. We have already encountered in chapter 7 instances in which changes in the phase-relations of harmonic components produced noticeable effects, but here we have even more dramatic evidence that the ear may be extremely sensitive to the relative phases of the components of a sound."

For those who may not have read an earlier post of mine on the subject of audible time-domain and waveform distortion created by higher order crossovers (2nd-order and higher), I mentioned a peer-reviewed paper I presented several years ago before a joint meeting of the AES and the E.E. Dept. at the University of Adelaide (Australia), complete with a demonstration, that clearly identified the audible differences heard within a typically-reverberant room when the crossover of a two-way loudspeaker was switched between a 1st-order and 2nd-order slope, while preserving identical on-axis response. Some of the audibility can be attributed to the much wider vertical dispersion/directivity properties introduced by the use of higher-order crossover networks which tend to create higher amplitude reflections from floor and ceiling boundaries. These time-delayed boundary reflections, combining with the direct arrival sounds, tend to arrive sufficiently late to create audible alterations of spectral balance, blurring of transients, etc., (as I mentioned above in an earlier paragraph).

There are a few truisms that we can take comfort in and believe. One of these, and perhaps the most important, is that no loudspeaker can ever reproduce a level of "true audible accuracy" that exceeds that predicted by a competent assessment and interpretation of a complete set of anechoic measurements, including frequency response (modulus of amplitude Vs frequency), radiation patterns in both vertical and horizontal planes at numerous frequencies, phase response, impulse response, step response, waterfall (cumulative spectral decay), energy-time, squarewave reproduction at several frequencies, non-linear distortion at various levels of amplitude at several frequencies, impedance (both resistive and reactive components), etc.

The same applies to the usefulness of measurements for determining the potential audible accuracy of all other components within the audiophile chain, including amplifiers, CD players, etc. Indeed, would any competent audiophile seriously consider purchasing an expensive power-amp that could not accurately reproduce square waves at relevant audio frequencies? If not, why would the loudspeaker be an exception - since, in any "linear system", comprised of several active components connected in series, the overall accuracy cannot exceed that of the least accurate component, whether it be the amp, CD player or loudspeaker. (Would anyone care to refute this?)

Hopefully, what I and others have recently posted on the subject has helped satisfy skeptics that a full set of accurate measurements, properly interpreted, can predict the "potential" a loudspeaker possesses to accurately reproduce complex musical waveforms - not how accurately it actually will reproduce them. This is why I have frequently said that it requires the interactive use of both a complete set of competently-made measurements and carefully-controlled listening comparisons with live music, etc. to establish "true accuracy". 

But a loudspeaker that does not exhibit good measured properties, especially with respect to impulse response, step response, amplitude/frequency response, good radiation patterns and low levels of non- linear distortion, can never reproduce complex musical waveforms as "accurately" as a loudspeaker that does - no matter who says otherwise. And, a loudspeaker with radiation patterns intended to create significant reflections from distant room boundaries (back walls, side walls, ceiling, etc.) begs the question as to what room dimensions, etc. the designer chose - because different room dimensions will result in different perceptions of what is heard. 

However, I suppose that no amount of "proof" will suffice to satisfy diehards who are convinced that it is proper (and even scientific) to design loudspeakers whose properties cannot be described by words, terms and expressions found within the lexicon used by competent designers possessing credentials recognized by their peers. In this regard, it is not always easy to respond in a professional manner to postings here on the NET that use words, phrases and explanations that are alien to engineers and physicists.

But those in pursuit of genuinely accurate products, based upon the serious application of well-known theory and principles taught by legitimate science and technology, will continue their search for audio components offering ever more accurate reproduction - that is substantive and can be proven by competent means.

Best of listening!

John Dunlavy

From 102365.2026@compuserve.com Mon May 11 15:15:01 1998
Newsgroups: rec.audio.high-end
Subject: Loudspeaker Accuracy, Part 2
From: John Dunlavy <102365.2026@compuserve.com
Date: 11 May 1998 16:15:01 -0400

Once again Gary Eichmeier/Susan Andrus seem intent upon demonstrating their apparent inability to grasp the true meaning of accurate reproduction.

"Accurate" has but one real meaning: output equals input. No ifs ands or buts! No attempts to out-guess what the recording and mastering engineers heard in their respective listening rooms and which was approved by the musicians. The intent of most of the better recording and mastering studios is to achieve a product that best represents the efforts of the musicians=. How do I know this? Simple! Most of the better known and better quality recording and mastering studios use our SC-IV's and SC-V's as their principal reference monitors. Ask a few if you don't believe it! You might also ask Doug Rife, the man behind the MLSSA loudspeaker measurement system, what loudspeakers he listens to and considers to be the most accurate? The list goes on-an-on!

Anyway, the idea of radiating sound in virtually all directions or in such a manner as to limit the direct sound and increase the amount of sound reflected from room boundaries to achieve ultimate accuracy is almost too absurd to justify serious comment. Indeed, while speakers that exhibit such radiation patterns can create the effect of sound "arriving from everywhere, this is hardly a facsimile of what is heard when listening to live music in most venues. Indeed, the relatively short distances traveled by reflected sound, relative to direct arrival sound, within a typical home listening room largely results in a listener merely hearing a frequency spectrum distorted by numerous peaks and nulls caused by cancellations and additions created by the vector sum of the direct and reflected components heard at the listening location. Frequently, these relatively broad peaks and nulls are perceived as a "boomy bass", strident sounding mids, and blurred, unnatural-sounding highs and transients. Yup!

The purpose of a "truly accurate" loudspeaker is to reproduce the signals fed to it and transport them to a listener with as little time and frequency domain distortion as possible. In doing so, a properly designed loudspeaker, with controlled dispersion in both vertical and horizontal planes (to minimize the amplitude of room reflections), will have the best chance of conveying to listeners the music as it was heard by the engineers and musicians that created it. Ask any competent recording and mastering engineers who produce audiophile recordings, e.g., those with Reference Recordings, Dorian, Sterling Sound, Absolute Audio, Sony, etc., all of whom depend upon DAL loudspeakers for producing their product.

Here, it seems appropriate to ask the question, "how can a loudspeaker whose radiation patterns (and room reflections) result in grossly distorting the original complex musical waveforms be expected to produce an accurate representation of what the recording and mastering engineers heard when the recording was made - and which they intended to be heard as they heard it in their studios?" Hmmm! Perhaps Gary/Susan need to recognize that there is no way for the recording and mastering folks to EQ a recording to take into account the almost unlimited variations in response, etc. that would occur within all of the different kinds and sizes of listening rooms used by audiophiles. Gulp!

With respect to Gary's earlier comments regarding the paper I delivered before a joint meeting of the AES and the EE/Physics Departments of the University of Adelaide on 19 July 1983, entitled "The Importance of Phase Linearity On The Perceived Quality of Loudspeaker Reproduction", I spoke of it in one of my postings as a "Peer Reviewed Paper" because "peers" (numerous professors, assoc. profs, engineers, physicists, etc.) were present to make comments and judge its veracity, etc., although it was not later published (even though I had the best intentions of doing so). I have, of course, delivered a number of peer reviewed papers covering many different subjects that were published in national and internationally read technical journals.

It seems that Gary also spoke of what he referred to as a "Peer Reviewed Paper" he presented before a meeting of the AES which he also mentioned was never published. Hmmm! Recently, I send a number of copies of my paper to those on the NET that requested it. I would be happy to send a copy to Gary/Susan if they will E-Mail me their address.

Some weeks back, I sent a copy of my CV to Jeff Adams of GTE in California, who mentioned it on the NET. If anyone doubts my credentials, technical competency, etc. I would be happy to forward a copy to them. In this regard, it would be interesting to see a copy of Gary's CV - perhaps he will condescend to send some of us a copy.

Thanks to all for waiting so long to receive my answer to Gary's/Susan's post - but I do keep very long hours attempting to keep three companies in tow and out of trouble.

Best of listening!

John Dunlavy

From 102365.2026@compuserve.com Sat Jun 06 13:11:02 1998
Newsgroups: rec.audio.high-end
Subject: Loudspeaker Accuracy, Pt. 2
From: Dunlavy Audio Labs <102365.2026@compuserve.com
Date: 6 Jun 1998 18:11:02 GMT

Well, in his latest posting Gary Eichmeier once again provides ample evidence that he simply does not understand acoustical theory and what it teaches about the different spectrum of sound heard within a large concert hall Vs that heard within an audiophile listening room of typical size.

For example, within most large symphony halls, the time it takes for a sound radiated by a musical instrument on stage to reach a listener seated in row 6 (center) is typically shorter by several hundred milliseconds than the average reflected sounds traveling over paths that are longer by several hundreds of feet. Don't forget, Gary, sound travels at about 1,100 feet per second (or about 1.1 feet per millisecond). Thus, if one of the musicians located at approximately center-stage is replaced with a swept-frequency sound source and this is picked up by a microphone at 6th row center, the direct sound will arrive in perhaps about 50 milliseconds. But the first sounds reflected from the sides and back walls of the hall will have traveled an additional average distance of perhaps 250 feet (or about 250 milliseconds). Thus, the average time difference between the arrival of the direct sound and the reflected sounds will be about 200 milliseconds. When the direct and reflected sounds add at the listening location, a spectrum of peaks and nulls is created with peak-to-peak and null-to-null intervals occurring about every 5 Hertz over most of the audible spectrum.

Let us now compare the concert hall acoustics with those of a typical audiophile listening room, where the direct sound radiated by each loudspeaker of a stereo-pair arrives at the listener's within about 10 milliseconds. But the first sounds reflected from the side and back walls will typically arrive at the listener's ears delayed by about another 10 milliseconds (corresponding to average reflected paths about 10 ft longer than the direct paths). Thus, the spectrum of peaks and nulls created within such a home listening environment will reveal peak-to-peak and null-to-null intervals occurring about every 100 Hertz over much of the audible spectrum.

I believe it should be evident to everyone that the difference between a spectrum of peaks/nulls with a periodicity of about 5 Hertz (in the concert hall) will be audibly quite different than the spectrum of peaks/nulls with a periodicity of about 100 Hz, heard within a typical audiophile listening room.

Indeed, it is well-documented that blind persons use this complex spectrum of partial peaks and nulls to aurally assess the size of a room they enter and the distance to some objects such as tables, chairs, etc., using ambient noise as the signal source. When such noise, etc. does not exist, some blind persons resort to making a "ticking or clucking" sound with their tongue or mouth.

To repeat what I and several other competent posters have said many times before, "Accurate has but one real meaning: output equals input". No ifs ands or buts! Nor any attempts to out-guess what the recording and mastering engineers heard in their respective listening rooms! The intent of most of the better recording and mastering studios is to achieve a product that best preserves the individual and collective efforts of the musicians, coupled with the desirable acoustical artifacts of the symphony hall, etc..

Therefore, it seems logical that a "truly accurate" loudspeaker should exhibit well-controlled radiation patterns (to minimize the amplitude of undesirable room reflections) and be able to acoustically reproduce the complex electrical signals fed to it with as little amplitude, time and frequency domain coloration/distortion as possible. Such a loudspeaker should have the best chance of conveying to a listener the music as it was intended to be heard by the recording and mastering engineers.

And here, it also seems appropriate to ask the question, "how can a loudspeaker, whose radiation patterns (and room reflections) result in grossly distorting the original complex musical waveforms, be expected to produce an accurate representation of what the recording and mastering engineers heard when they made the recording - and which they intended to be heard as they heard it in their studios?"

Oh sure, it is easy to design a loudspeaker that radiates sound in many different directions to produce what many in the industry refer to as "pizzazz sound" - but this is hardly an accurate representation or acoustical reconstruction of the original musical event! Accurate? No, no, no! Pretty sounding? Perhaps, to some listeners who prefer euphonic embellishment over true accuracy!

In this respect, perhaps Gary needs to recognize that there is no way for the recording and mastering folks to EQ a recording to take into account the almost unlimited variations in listening room acoustics, etc. that will be encountered within the myriad of different listening rooms with different acoustical properties used by audiophiles. All they can hope to intelligently accomplish, in pursuit of true accuracy, is to preserve the desirable acoustical properties of the original recording venue that will allow listeners with accurate equipment and loudspeakers to experience what persons attending the live performance heard and felt.

How do I know this? Simple! Most of the better known, audiophile-quality recording and mastering studios use our SC-IV's and SC-V's as their principal reference monitors. Ask a few if you don't believe it! Try Master Disc, Absolute Audio and Sony Mastering (to name but a few mastering groups). You might also ask Dorian, Millennium, Reference Recordings, etc2E And, ask Doug Rife, the man behind the MLSSA loudspeaker measurement system, what loudspeakers he listens to and considers to be the most musically and measurably accurate? And then there are the many amplifier designers/manufacturers who demand the most accuracy possible from a loudspeaker to evaluate their products. And, how about all of the many product reviews within the better audio magazines that give loudspeakers with true measurable and audible accuracy the highest ratings? And, what about all of the many competent magazine reviews, product of the year awards, etc., given DAL loudspeakers within major audiophile publications over the past five years. The list goes on-an-on!

Anyway, the idea of radiating sound in virtually all directions to achieve ultimate accuracy is almost too absurd to justify serious comment. Indeed, while speakers that exhibit such radiation patterns can create the effect of sound "arriving from everywhere," this can hardly be considered an accurate facsimile of the original sound heard by a listener at the original concert, etc.

Yup! The sound of loudspeakers that exhibit broad or quasi-omni directional radiation patterns may appeal to the tastes of those who favor spectral and/or time-domain coloration, but this should never be referred to as "accurate reproduction". Indeed, only properly-designed loudspeakers that exhibit true pulse-coherent, flat-spectrum response, with well-controlled radiation patterns in both vertical and horizontal planes can be considered capable of simulating the original live musical experience with an accuracy that prevents most competent audiophiles from discerning any audible difference during competently conducted live Vs reproduced comparisons. And, yes, we have conducted such comparisons on several occasions with experienced audiophiles participating. No bull! Just the facts Gary.

Perhaps, this is also why well-designed loudspeakers, with accuracy determined by both complete measurements and properly-conducted listening evaluations, continue to garner most of the competent magazine awards - year-after-year!

Best of listening, 

John Dunlavy
CEO
Dunlavy Audio Labs
www.dunlavyaudio.com

From 102365.2026@compuserve.com Thu Jun 25 11:01:19 1998
Newsgroups: rec.audio.high-end
Subject: Loudspeaker Accuracy
From: Dunlavy Audio Labs <102365.2026@compuserve.com
Date: 25 Jun 1998 12:01:19 -0400

I have resisted replying to Gary Eichmeyer's recent postings, especially those targeting John Ongtooguk, Coyote, "uh OH!", and others, until now. But I believe the time has come to try and set the record straight by stating well-known and well-understood facts attributable to competent scientists, acoustical engineers, recording engineers and physicists.

However, it appears that Gary does not wish to be swayed by such facts, preferring instead to trumpet his own incorrect ideas as though they were "gospel". Anyway, lets once again explore why Gary's "teachings" are not in consonance with teachings known and understood by those well-versed in relevant fields.

For example, when an audiophile quality recording is made of a symphony orchestra playing within a hall noted for its outstanding acoustical properties, most experienced recording engineers attempt to capture (1) the complex inner-detail of individual instruments, (2) an accurate stereo rendition of the orchestra's soundstage, (3) acoustical separation of different instruments, (4) the "ambiance" of the hall, and (5) a "natural" spectral balance that takes into account the acoustical properties of the venue. The "bottom-line" intent of a competent recording engineer is to capture the "true acoustical presence" of the orchestra's performance within that particular symphony hall so that an audiophile with a suitably accurate system can hear the music much as the engineer heard it while making the recording. 

How do I know this? Very simple! I was a member of a team under the supervision of the well-known Musicologist Dr. Robert Winter of UCLA when he was in Colorado Springs to make an interactive CD of the Colorado Springs Symphony playing the Appalachian Springs. He spent weeks at our facility listening to the recordings in our well-equipped listening room to determine what alterations to make in the recording process, which mics better captured the essence of the music, which mic locations yielded the most accurate soundstage, e.g., location of individual instruments, separation of instruments, relative distances of different instruments, etc.

Shortly after this project was completed, the Symphony's Board of Directors asked me if I would record future monthly performances for re-broadcast over KCME, a local classical FM station. I did so for the next two and one-half symphony seasons. Making these recordings provided a rare opportunity to experiment with mic types, mic placing, recording equipment, etc. The most accurate of these recordings, in my opinion, were those I made with a perfectly matched pair of instrumentation-quality, omni-directional mics with ultra-low noise/low distortion FET pre-amps. The recorder was a nearly flawless 24-bit DAT machine. My favorite mic location was 10 feet above the "third row center" seat, directly in front of the orchestra. At this location, the first-row musicians were about 18-22 feet from the mics.

Monitoring the recordings in "real-time" through a pair of Sony MDR-7506 Professional Studio Monitor headphones, while seated directly below the mics, there was literally no audible difference between the live music and that heard through the phones =96 for at least the first few minutes of listening. However, after several minutes of comparing the live with the recorded music, very small differences became audible, mostly consisting of being able to distinguish between sounds reflected from the back of the concert hall Vs "direct arrival" sounds from the orchestra. But the differences were very subtle and required several minutes to discern.

Listening to these symphony recordings over measurably and audibly accurate loudspeakers within our well-equipped listening room and comparing the sound (in real-time) to that heard through the same Sony MDR-7506 headphones, virtually all persons who have participated have been unable to discern any audible difference for the first several minutes. Thereafter, some of the more discerning listeners begin to pick up on the same subtle differences heard while making the recording, i.e., the inability of the headphones to provide a listener with audible means for differentiating between sounds arriving from the "rear" of the hall from the direct arrival sounds from the orchestra. Surprisingly, however, the loudspeakers seem to restore some of the original concert hall ambiance, perhaps by providing the pinna with an opportunity to pre-process some front-back directivity cues introduced by reflections from the rear wall directly behind the listener.

Carefully-controlled listening experiments, using a pair of "side-firing" speakers located adjacent to the main loudspeakers to introduce reflections of various amplitudes from the side walls of the listening room, seemed to add additional "ambiance" to the reproduction. However, our hearing processes appear to be very hard to fool and such "time-delayed subsidiary radiation" has never been able to provide more than momentary listening satisfaction. Usually, only a few minutes of listening are necessary for a competent listener to reach the conclusion that the perceived accuracy of the reproduction has been degraded.

The most important lesson we have learned, I believe, is that the most subjectively-accurate reproduction of well-recorded music is obtained with loudspeakers determined to be accurate based upon a complete set of anechoic measurements (at a distance of about 10-12 feet) of all measurable performance parameters, including time-domain, frequency-domain, directivity, non-linear distortion, impedance, etc. We have concluded, after much experimentation, that the most accurate reproduction is obtained with loudspeakers possessing well-controlled directivity patterns that minimize reflections from side walls, the wall behind the loudspeakers, the floor and the ceiling surfaces. It is important to reduce such reflections because they typically possess delays amounting to only a few milliseconds =96 compared to the delays of hundreds of milliseconds encountered within typical symphony halls, as mentioned earlier.

So, technically speaking, how are the reflected sounds within a typical home listening room heard and perceived differently from those heard within a symphony hall? Well, although human hearing is not terribly acute in discerning the arrival times of reflected sounds differing, perhaps, by as much as tens of milliseconds, we are able to readily discern the "spectral differences" created by the reflected sound combining with the direct sound. For example, the frequency spectrum created by the "vector addition" of the direct and reflected components of a sound within a room consists of a series of peaks and partial nulls that persist over a wide range of frequencies, with amplitudes dependent upon the absorptive properties of the reflective surfaces. Amplitude peaks occur at those frequencies where the direct and reflected sounds arrive "in-phase" (multiples of one-wavelength). Likewise, nulls (or partial nulls) occur at frequencies corresponding to those where the two paths differ by an odd-multiple of half-wavelengths (180 degrees, 540 degrees, etc.). 

Let's briefly examine the difference in the spectrum of peaks and nulls that might occur within a typical home listening room versus the spectrum that might be heard within a typical concert hall. For a listener seated about 10 feet from a loudspeaker within a typical listening room of modest size, the extra distance traveled by a reflected wave, compared to the direct-arrival sound, probably averages about 7 feet. This is equal to approximately one-half wavelength at about 79 Hertz, the frequency at which the first "aural null" will be heard by a listener (assuming the null is not partially "filled-in" by in-phase reflections from other room surfaces). This null (or partial null) will be repeated at each of several odd-harmonic frequencies, e.g., 237 Hz, 395 Hz, 553 Hz, etc. 

Likewise, "peaks" in response will be heard at frequencies where the direct and reflected sound components "add in phase". In our example, these will occur at approximately 157 Hz, 314 Hz, 471 Hz, etc., depending upon the dimensions, reflective properties and reflection angles of the wall, ceiling and floor surfaces.

How do these typical listening room spectrums compare to those encountered within a typical concert hall for a listener located, perhaps, at a seat in third-row, center? Well, our own symphony hall in Colorado Springs, would probably yield a first reflection arriving about 100 milliseconds after the direct arrival sound. This will create a spectrum of "partial nulls" at frequencies corresponding to about 5.5 Hz, 17 Hz, 28 Hz, 39 Hz, 61 Hz, etc. Peaks in response will be heard at frequencies of about 11 Hz, 22 Hz, 33 Hz, =85 121 Hz, 132 Hz, etc. Acute listeners will perceive the spectrum comprised of such closely-spaced peaks and nulls quite differently from the spectrum produced by a pair of loudspeakers within a typical home listening room.

Indeed, it is these differences in the spectrum of "peaks and nulls" that many blind persons utilize to determine the dimensions of a room they are within and their approximate distance from walls, etc.

Now, lets return to audiophile systems and examine what loudspeaker radiation patterns might reproduce well-recorded music with the greatest perceived accuracy within a typical listening room. Assuming that one is listening to music that has accurately captured the "ambiance" of the recording venue, e.g., reflections from relevant reflective boundaries, etc., will the much different spectrum of peaks and nulls created by reflections from listening room surfaces add any relevant or realistic ambiance to what is heard. No! At least, not when heard by competent, experienced audiophiles. Why? Well, the significantly different spectrum of peaks and nulls introduced by reflections within the listening room will tend to obscure those within the recording being heard, resulting in "blurred cues" as to the ambiance, etc. of the recording venue. 

Thus, although an omni-directional loudspeaker (or one which accentuates the level of sounds reflected from room boundaries) may sound "expansive" and "encompassing" for a relatively short time, an experienced audiophile will soon discover that such sound quality does not coincide with the real-world "concert hall experience". This is because our "ear-brain combo" is really a lot more perceptive and smarter than many people believe. Thus, while it may take time for a person to discover they have been "fooled" by a "cleaver loudspeaker design", they will eventually reach that conclusion and will no longer enjoy listening to such enhanced "make believe" reproduction. 

The "bottom line" is simple: long-term, it is hard to beat the reproduction provided by a stereo-pair of truly accurate loudspeakers with "controlled-dispersion" radiation patterns that minimize the undesirable spectral components created by wall, ceiling and floor reflections. In consonance with this assertion, I am not aware of any major recording or mastering studios that utilize loudspeakers with wide vertical and/or horizontal directivity patterns. All of the ones with which we are familiar and regularly speak to, use monitors with directivity patterns optimized to minimize reflections from wall, ceiling and floor surfaces. This permits them to monitor the sound of their recordings in a manner that lets them judge how they will sound and hopefully appeal to potential purchasers. Hmmm! 

So! If listeners want to hear what the recording and mastering engineers probably heard when making their recordings =96 why not use loudspeakers with similar properties?

Best of listening, John Dunlavy



--------------------------------------------------------------------------------


Subject: Loudspeaker Accuracy<br>

From: Dunlavy Audio Labs 102365.2026@compuserve.com

Date: 1998/07/16 Message-ID: <6olps3$ifp@news01.aud.alcatel.com> 

Newsgroups: rec.audio.high-end [More Headers] [Subscribe to rec.audio.high-end]<p>

Please approve and post the following article composed by John Dunlavy for inclusion on rec audio high end under the thread "Loudspeaker Accuracy":<p>

Some recent claims for the "audible superiority" and "better soundstage accuracy" of loudspeakers with more-or-less omni-directional radiation patterns, compared to conventional loudspeakers with much narrower horizontal directivity patterns, beg further examination and informed comment. Further, a few of the definitions and descriptions given for binaural and stereo recording and reproduction also need to be compared to what well known texts have to say about the subject.<p>

For example, in one recent posting the author spoke of what he believed to be the differences between "binaural" and "stereo" recordings and their reproduction. Unfortunately, much of what the author said does not appear to agree with what has been published on the subject within credible books and journals.<p>

With respect to defining the "differences" between binaural and stereo recording and playback systems, on page 181 of the book "Hearing - Its Psychology and Physiology", by Stanley S. Stevens, Ph.D., published by the American Institute of Physics, we find, "There are two ways of reproducing sounds in true auditory perspective. One is binaural reproduction, in which there is led to the observer's ears, by means of earphones, an exact copy of the sound-waves which would stimulate his two ears if he were listening directly. We can do this conveniently by picking up the sound with two microphones, placed in the position of the ears on a man-shaped dummy, and connecting one earphone to the amplified output of each microphone. Then, if someone walks around the dummy, talking as he goes, a person wearing the earphones has a compelling illusion of someone walking around him. The other method uses two or more microphones and a corresponding number of loudspeakers, and aims to reproduce in a second room an exact copy of the pattern of sound-vibration that exists in the original room." And, "Ideally, an infinite number of microphones and loudspeakers of infinitesimal dimensions would be needed to make the reproduction perfect, but, in practice, as few as two microphone-loudspeaker combinations (channels) have been found to give fair auditory perspective. Extensive tests were carried out with two and three channels in various combinations in order to determine the adequacy of such methods (Steinberg and Snow). With three-channel reproduction, there is reasonably good correspondence between the actual position on the stage and the apparent position on the virtual stage, both as regards right and left, and front and back. Thus, the system affords depth as well as angular localization." and, "With two-channel reproduction, the virtual stage tended to appear wider and less deep than with three-channel reproduction."<p>

Also, in Howard Tremaine's book, "Audio Encyclopedia", the following questions and answers appear, beginning on page 82: Question - 2.111: "What is 3-D sound? - The term applied to three-dimensional or stereophonic sound." Q - 2.112, 2.112: "What is a monaural sound system? - A sound system consisting of one source of sound, such as a radio, using a single loudspeaker for reproduction. It is termed monophonic." Q - 2.113: "What is a binaural sound system? - A system consisting of two microphones at the point of pickup, placed in the same relationship to each other as the ears of a listener. The microphones are connected to separate amplifier systems and transmit the program material to the listener through headphones. True binaural sound cannot be achieved with loudspeakers, only headphones. When loudspeakers are used it is two channel stereophonic sound." And, Q - 2.114: "What is a stereophonic sound system? - A system using two or more microphones with a separate amplifier and loudspeaker for each microphone channel. With such an arrangement of equipment, the sound travels from one speaker to the other as the principals move across the stage. Such a system permits an orchestra to be reproduced closer to its proper perspective."<p>

While some aspects of these definitions might have been a little clearer had the authors used slightly different wording, I believe they adequately describe "binaural reproduction" (through headphones) Vs "stereo reproduction" through loudspeakers.<p>

The belief held by some that large amounts of time-delayed reflections from listening room boundaries are beneficial and or necessary for accurately reproducing the original sound field of an orchestra by a stereo playback system is simply not true! While loudspeakers with near-omni directivity can sometimes create the illusion of the listener being "surrounded by sound", this can hardly be described as an accurate recreation of the original recorded soundstage or the "stereo imaging" as it was intended to be heard by competent recording and mastering engineers (which seldom, if ever, use omni-directional loudspeakers). Indeed, having maintained close contacts with several of the top recording and mastering companies in the U.S. during the past twenty-plus years, I cannot recall a single one that has used loudspeakers with omni or near-omni radiation patterns while engineering their products.<p>

(However, it must be admitted that a wide and diffuse sound field, created by omni or quasi-omni loudspeakers within a typically reverberant listening room, can appeal to many kinds of listeners - until they eventually recognize that such a sound field is not realistic and often becomes tiring after a relatively long listening session.)<p>

Further, the diffuse spatial qualities produced by loudspeakers with quasi-omni directivity should never be likened to the spatial qualities of a "binaural recording" heard through headphones. They simply are not the same - period! Nor should the diffuse soundfield created by nearly omni-directional loudspeakers ever be referred to as being more "accurate" than the "stereo soundstage" attainable within a good listening room when listening to truly accurate loudspeakers possessing symmetrical, relatively narrow radiation patterns in both vertical and horizontal planes.<p>

Anyway, my time limit for internet chatter has been reached for today. Perhaps, a few others will find time to post their own references, experiences and opinions regarding the audible differences between very wide and relatively narrow loudspeaker directivity patterns and how each impacts upon our perception of "realism" when reproducing well-recorded music within a typical audiophile listening room.<p>

Many thanks to John Ongtooguk, Robert Salvi, Thomas Nulla, Stewart Pinkerton, Paul Macca, et al, for their many cogent and informed comments.<p>

Best of listening,<p>

John Dunlavy<p>

From 102365.2026@compuserve.com Thu Dec 03 16:58:23 1998
Newsgroups: rec.audio.high-end
Subject: Re: speakers - neutrality and dynamics mutually exclusive?
From: Dunlavy Audio Labs <102365.2026@compuserve.com>
Date: 3 Dec 1998 22:58:23 GMT
The 25 Nov. posting of Dan Kirkland, which states:

>Hmmm, the recent test done by Stereophile showed this Dunlavy to be >down about 6dB at 30 Hz. > >They also said the woofer resonant frequency was about 37 Hz.

deserves informed comment.

First, I sincerely believe that Stereophile magazine does an excellent overall job of reviewing loudspeakers of relatively small to moderate size. I also believe that John Atkinson does a very good job of measuring the important performance properties of loudspeakers with modest dimensions. However, John's ability to perform accurate measurements of large-size, full-range loudspeakers employing a symmetrical array of drivers is often severely limited by his lack of an anechoic environment with suitably large dimensions. This was certainly the case with respect to his review of the SC-IV/A, which seems to significantly conflict with Bob Deutsch's "listening" section of the review. (Bob bought the SC-IV/A review pair.)

Indeed, the size of John's measurement room and its non-anechoic properties, precludes taking meaningful measurements of large, multi-driver loudspeakers at distances exceeding about 50 inches. This is because a loudspeaker with a symmetrical array of drivers, e.g., a tweeter surrounded above and below by mid and woofer drivers, is much like a multi-element camera lens: it can only be sharply focused a specific distance but retains reasonably accurate focus over range of distances known as the "depth of field". 

This comparison applies to a loudspeaker like the SC-IV/A, which uses a symmetrical vertical array of five drivers, "time-path aligned" to produce true pulse-coherent performance over a distance range from about 8 to 15 feet. Thus, it does not make sense to measure its performance at 50 inches, a distance at which no speaker with a symmetrical multi-element driver array can provide optimized performance. Although John attempted to take measurements at a distance of 98 inches, it is clear that his room reflections at this distance were destroying the integrity and accuracy of his measurements. (Sob!)

The difference between Stereophile's loudspeaker measurement capabilities and those of DAL are significant. We have two large anechoic chambers (24 ft long by 20 ft wide by 16 high), with all interior surfaces covered with high-density open-cell acoustical foam in the form of 4 ft. square blocks with 1.5 ft. deep alternating wedges. Both chambers provide measurements accurate to within plus/minus 0.2 dB from about 300 Hz to 20 kHz at a measurement distance of 10 feet. By using a highly accurate technique, both chambers have been calibrated down to 20 Hz to yield measurements accurate to within about 0.5 dB.

John's assessment of the low-end response of the SC-IV/A, based upon its "impedance resonance" at about 37 Hz and an assumed roll-off of 12 dB/octave, is highly simplistic. It assumes a single woofer within a more-or-less rectangular-shaped, closed-box type enclosure, having little or no internal damping material. Such designs emphasize the realization of higher efficiency at the expense of wider bandwidth and optimum damping/impulse-response properties. 

By contrast, the SC-IV/A has two woofers sharing a common large enclosure, filled with a high-density, open-celled foam, which not only "over-damps" the response of the woofer-box combination, but also increases the apparent volume and lowers the rate of amplitude roll-off below resonance. Based upon many years of research, we have concluded that our non-conventional design criteria yields the most audibly (and measurably) accurate bass reproduction. Our design approach yields a woofer response with a low-end roll-off of amplitude Vs frequency that is typically less than about 4-6 dB for the first octave below the impedance resonance. Of course, the roll-off ultimately approaches the well-known, classical 12 dB/octave.

Another property of our unique design that must be weighed in assessing its low-end response is that it uses a symmetrical, vertical array of drivers, consisting of a "center" tweeter, two mids and two woofers. Because the separation of the two woofers yields an on-axis "directivity/amplitude gain" of 3 dB, which increases at frequencies above the crossover frequency to a maximum approaching 6 dB, the crossover network was designed to reduce and flatten the woofer SPL curve at higher frequencies, resulting in a slower roll-off in on-axis amplitude below the crossover frequency.

Likewise, a near field measurement of woofer response, with the mic close to one woofer (as John did), will not reveal the effect of the aforementioned directivity gain on the overall, far-field response of the woofers and their roll-off below resonance.

John Atkinson's somewhat incorrect assessment of the SC-IV/A's low-end response should not be held against him for readers should not expect an editor to have the ability to fully incorporate all of the rather complex design algorithms and measurement requirements associated with various loudspeakers properties. Likewise, he is terribly limited by not having a suitable anechoic environment of relevant size for obtaining accurate measurements of complex loudspeaker performance parameters. My hat goes off to John A. for his many fine reviews of so many loudspeakers of different design, complexity and size. Even though his measurements may not always represent true laboratory accuracy, they are still a good guide and the best among all of the current audiophile magazines. 

Best of listening,

John Dunlavy

From 102365.2026@compuserve.com Sat Dec 05 16:41:48 1998
Newsgroups: rec.audio.high-end
Subject: Re: speakers - neutrality and dynamics
From: Dunlavy Audio Labs <102365.2026@compuserve.com>
Date: 5 Dec 1998 16:41:48 -0600

Randall Bradley (randy@rdrc.rpi.edu) writes: >AFAIK, you can get about 10% increase in box size by the use of a good
>stuffing material (wool for example).

The widely accepted assumption that wool is an efficient damping material for use within woofer enclosures is not true - even a fairly dense stuffing of wool has little effect on a woofer's properties. A fairly dense, open-cell (non-flammable) foam is the medium of choice among knowledgeable designers.

>In any event, except for rooms that measure a boost in LF response (in
>which case it is there for any speaker measured) speakers I've measured do
>not have extra response below rolloff. This would tend to indicate that JA
>should have seen this response in his room too. In fact I've found a very
>strong correlation between the 1m measurement and the room measurements.

The use of the Stereophile review measurements of the SC-IV/A to prove that its bass response is not what we claim is somewhat ill-informed. Unfortunately, although John Atkinson is a superb product reviewer, Stereophile's measurement capabilities, including a room requiring measurements to be made at a distance of only 50 inches, are not adequate for accurately measuring the performance properties of loudspeakers having the physical dimensions and complexity of the SC-IV/A, whose design distance was 10 feet. This especially true of loudspeakers that employ several drivers within an "array" whose size is large with respect to the measurement distance. (It is a limitation well known to antenna engineers tasked with measuring the gain and directivity patterns of antennas.) And his "attempted" measurements at a distance of about 98 inches became further distorted by room reflections (as is evident from the "ringing" observed on the square-waves he measured, which show no such ringing in our own excellent anechoic chamber at a distance of 10 feet.)

(Indeed, a large, multi-driver loudspeaker is also much like a multi-element camera lens in that it can only be sharply focused at a specific distance, but retains reasonably accurate focus over a range of distances known as the "depth of field".) 

>Of course, a presumption of a 12dB/oct roll-off below the resonant
>frequency of the woofer is reasonable for a sealed box enclosure. The
>ability to "hang" the response for another octave below the resonance as a
>first order (ie 4-6dB) because of the inclusion of a stuffing of open cell
>foam would represent a departure from TS theory. Or so it seems on the
>surface.

In reality, the 12 dB/octave "roll-off" is merely the eventual slope approached at frequencies well below the system resonance and where the medium within the enclosure is air. Indeed, within the first octave below resonance, the drop in amplitude of a closed box system is typically only about 8-10 dB, depending upon whether the enclosure is internally damped by means of acoustical absorbing material. Also, with a vertical array of two woofers connected in parallel, whose combined SPL is more than 3 dB in excess of that needed, the woofer crossover circuit can be designed to begin its low-frequency roll-off at a higher frequency than the crossover point, yielding a much lower slope than 12 dB/octave within the first octave below resonance.

And, a woofer's rate of roll-off can be considerably reduced by using internal damping materials having appropriate acoustical properties that significantly affect the velocity and absorption of waves propagating through them. In engineering circles, this is referred to as the Velocity-of-Propagation (VP) Factor of the material or medium, which expresses the velocity of a wave through the medium divided by its velocity through air. Lossy materials generally exhibit a VP Factor that decreases as their density increases. Likewise, the loss incurred by a wave propagating through a lossy medium is not constant with frequency and generally decreases as the frequency is lowered. (High-density, open-cell foams can yield VP factors as low as about 0.5 at bass frequencies.) Indeed, physics and engineering teach that when a wave, be it acoustical or electro-magnetic, propagates through a medium with "dielectric properties" (or its acoustical equivalent), its velocity is reduced by a very relevant amount. (Ask any knowledgeable college prof.)

Additionally, the use of two woofer drivers in a "vertical broadside array", versus a single woofer, typically yields a minimum, on-axis gain of 3 dB, if the input signal voltage is held constant. However, at frequencies above where the two woofers are separated by a vertical distance greater than about one-quarter wavelength, this gain rises slightly above 3 dB because of the additional "directivity gain" achieved by what is called the "array factor". (Again, something well-known to antenna engineers and those who work at electromagnetic frequencies.)

For example, the approximate "minus 3 dB beamwidth" of a uniformly illuminated aperture (about the same for a two-element array of drivers, neglecting the directivity of the drivers themselves) can be found by the expression 50.8 times the wavelength divided by the separation between the radiators (expressed in wavelengths). Thus, two drivers physically separated by a vertical distance equal to a wavelength would exhibit a main lobe with a -3dB beamwidth of about 50.8 degrees in the vertical axis (at a reasonable distance in front of the array). This represents an "on-axis power gain" of almost 6 dB, compared to one of the drivers by itself. Thus, at this frequency, a designer attempting to achieve a "flat on-axis" frequency response would have to reduce the signal being fed to the two drivers by a full 3 dB. By doing so, the roll-off in amplitude within the first octave below the "system resonance" becomes much less than the eventual roll-off of 12 dB/octave.

Likewise, a near field measurement of woofer response, with the mic close to one woofer (as John did), will not reveal the effect of the aforementioned directivity gain on the overall, far-field response of the woofers and their roll-off below resonance.

No, Randy, the design and measurement of large loudspeakers, like the SC-IV/A is not a trivial matter and requires as much knowledge and measurement capability as a competent company can afford. Without appropriate knowledge and an adequate measurement capability, loudspeaker performance becomes an exercise in guesswork - where the accuracy of the product is a matter of personal opinion, based solely upon subjective opinion.

Best of listening,

John Dunlavy



End of archive. New posts will be added as I become aware of them. 



--------------------------------------------------------------------------------


I hope you find this information as interesting and useful as I have. The other contributions to these threads (and many others) can be found in the rec.audio.high-end archives; arguably the best audio resource extant. 

These and other topics are also discussed in a series of 'white papers' available from Dunlavy Audio Labs, as well as in the August 1996 'Stereophile' interview with John Dunlavy. 

I have no connection with DAL except customer.