John Dunlavy's posts to rec.audio.* during 1997: 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 13 November 1997. -------------------------------------------------------------------------------- Newsgroups: rec.audio.high-end Subject: More on loudspeaker performance From: John Dunlavy <102365.2026@compuserve.com> Date: 23 Jan 1997 17:44:58 -0500 A recent comment stated that I, "... regularly veer off into the time domain and phase matters without offering any new (or even old) evidence of why distortion (or the absence of distortion) in these areas should really much matter to the human ear auditioning playback in an enclosed space, especially a domestic-sized one.", is good and deserving of a detailed answer. First, the time and frequency domains are related to each other, mathematically; sometimes in a simple way and sometimes in a more complex manner. For example, events in the time domain (such as changes of signal amplitude with time) always produce events in the frequency domain, with components that are generally related by the reciprocal of the variations in amplitude with respect to time. Further, a short impulse in the time domain contains implicit information regarding frequency response, phase response, step response, etc., all of which may derived by FFT analysis of the impulse. (Doug Rife's now famous MLSSA system is an excellent example of using FFT analysis of an impulse to determine most loudspeaker performance attributes.) Once understood, the step response (merely the integral of the impulse response) provides a quick visual evaluation of several time and frequency domain properties of a loudspeaker. Of course, most audiophiles, without an appropriate technical and math background, probably cannot visualize how so many of these seemingly diverse performance properties can be so directly related to each other - but they are! The only exceptions are the impedance, radiation patterns and non-linear distortions (THD and IMD). Linear signal distortions, such as a poor impulse/step, large variations in the modulus of amplitude and phase Vs. frequency, etc., can directly affect the "perceived realism" of complex musical waveforms reproduced by any component within the audio chain, including the loudspeaker. The acoustical reflections from the floor, ceiling, walls, etc. of the listening room add another set of "time and frequency domain distortions" onto those of the audio equipment, though with a very different signature in the time domain. The difference is that most time-domain distortions created by equipment, including loudspeakers, typically occur within a time window of less than a mSec, compared to several milli-seconds for an average "floor reflection" and even longer for wall and ceiling reflections within most rooms. Also, a property of human hearing, known as the "fusion time" (which is the interval of separation between short transients required to perceive whether one or two transients are present), permits a "critical listener" to discern between room reflections (normally arriving more than about 5 mSec after the direct sound) and time-domain distortion created within audio components, including the loudspeaker (which occur within a time window typically much less than 0.5 mSec.). Thus, fusion time helps us to become familiar with and ignore most reflections from room boundaries, while letting us discern time domain distortions produced by equipment/loudspeakers as a blurring/smearing of musical transients or an alteration of spectral balance, etc. With respect to your question as to what "... blind music-playback tests you and your colleagues have performed with the ***only*** variable being changes in "phase" behavior, with results showing a preference for one kind of behavior over others?", we have conducted quite a wide spectrum of tests during the past 20-plus years. We spent the time and money to do the research for two reasons: 1) it appeared to have been ignored by other investigators, and 2) it represented a unique technical challenge, was intellectually interesting and potentially commercially rewarding. (Commercially rewarding in the context of permitting us to better understand certain design goals and performance constraints, leading to better performing and more salable products.) With regard to the audibility of changes in phase Vs frequency of a loudspeaker (or any other audio component, for that matter), our investigations have tended to show, pretty conclusively, that small changes of less than about 30 degrees per octave are probably not audible - except when listening to certain complex signals such as impulses and square waves. However, phase changes exceeding 180 degrees within an octave (or less), created by a loudspeaker crossover network, non time-coherent drivers, etc., are usually audible with complex waveforms and musical transients, when compared to the unaltered original signal. This is especially true if the 180 degree phase shift occurs within a octave or less within the frequency range from about 200 Hz to 5 kHz (often referred to as the mid-range region of most loudspeakers). I could go on and on about this interesting subject but time presently is not available to do so. It is shame that more investigators have not explored this generally neglected but very important ground. Last, but not least, I would like to remark that if I had to choose but one measurement of a loudspeaker from which I would be required to determine and describe most of its performance properties, including frequency response, phase response, impulse response, etc., it would be "step response". For those familiar with step response and what it has to reveal about so many important properties (at a mere glance), it is a true gem in disguise. Best regards, John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Re: Speaker efficiency vs. sensitivity From: John Dunlavy <102365.2026@compuserve.com> Date: 5 Feb 1997 16:40:50 -0500 Recent postings on the NET have raised questions regarding the "efficiency Vs SPL" of loudspeakers. Since there appears to be some misunderstanding concerning the proper definition and usage of the two terms, I thought a few words on the subject might be in order. To begin, the terms efficiency and SPL have very different meanings. Also, there is no direct connection between the two terms without adding the requirement for additional information such as the input impedance, distance of measurement, etc. For example, "efficiency" is a term that implies the ratio between the total power delivered to a "load" by the device divided by total power fed to the device. It does not really matter much what the device is so long as the two quantities, input and output powers, can be measured. In the case of a loudspeaker, the absolute efficiency is taken to mean the total acoustical power radiated (into a hypothetical sphere surrounding the loudspeaker) divided by the total power delivered to the input terminals of the loudspeaker. By comparison, SPL is a "relative quantity" that defines the measured sound pressure level produced by a given loudspeaker at a distance of one meter (along the intended listening axis) for a 1 watt input (usually taken to mean an input signal of 2.83 volts, RMS, across the input terminals). The important quantities that are usually missing are the frequency of measurement and the input impedance of the loudspeaker. (In most cases, the input impedance is taken to be 8 Ohms but this is often not observed as a requirement since most solid-state amplifiers have a very low output impedance, usually below a few-tenths of an ohm.) Since the question really reduces, in a practical sense, to how loud one loudspeaker will sound compared to another, when using the same amplifier and signal source, SPL provides a good "relative reference" for making simplistic comparisons. If anyone would like more detailed information, let me know. Several books, going back to the 1934 edition of Loudspeakers, by Mc Lachlan (Oxford), provide more detailed information on the subject. Best regards, John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Re: Speaker efficiency vs. sensitivity From: John Dunlavy <102365.2026@compuserve.com> Date: 11 Feb 1997 10:28:49 -0500 My Feb. 4th posting regarding efficiency was written in a hurry and failed to address a few important issues. In the meantime, on Jan. 31st, Dick Pierce posted an excellent explanation regarding the subject; one that leaves little to add. Dick made the statement that the efficiency of a loudspeaker is related "only" to the efficiency of the drivers and not the enclosure (box). This is certainly true with respect to what I might call "absolute efficiency", defined as the ratio of the total power radiated (integrated over the surface of a hypothetical sphere surrounding the loudspeaker) to the total input power at the input terminals of the loudspeaker. However, this definition, taken alone, might be a bit confusing to those lacking Dick's technical background and expertise. For example, a listener comparing two loudspeakers with identical "efficiency" ratings might perceive one sounding much louder than the other. This might be explained on the basis of the two loudspeakers possessing different radiation patterns, with one radiating more acoustical energy than the other in the direction of the listener. A good example might be a small "box" loudspeaker and a "horn" loudspeaker, both with drivers having equal "efficiency". The horn loudspeaker, with a relatively narrow radiation pattern will probably deliver as much as a 10 dB louder sound to an "on-axis listener" compared to the box loudspeaker with a nearly "spherical directional pattern", whose dimensions are small compared to a wavelength (throughout the frequency range being evaluated). Another problem in relating the SPL to efficiency of a "box loudspeaker" may be encountered at frequencies below which the system (a closed-box plus driver) is resonant. This occurs at a frequency that is approximately equal to the square-root of the product of the resonant frequencies of the box and the driver. Below this system resonance, the total acoustical energy radiated by a "closed box design" drops at the approximate rate of 12 dB per octave (assuming that the resonance of the box is much higher than the free-air resonance of the driver). However, if the box is vented by means of a "port", the system "Q" will rise (along with a rise in the level of radiated energy) at the low frequencies affected by the port radiation. Thus, one might conclude that the "efficiency" of a ported enclosure is higher at some bass frequencies than that of a non-ported design. And, this might be a valid conclusion if one ignores radiation patterns, "stored energy" (time-domain response) and the rate of low-frequency roll-off (often exceeding 24-30 dB/octave below resonance). From the above comments, it might be seen that the terms SPL and efficiency convey quite different information concerning performance and must be well-understood if accurate comparisons are to be made between different loudspeaker designs. This includes the effect of "Q" (both of driver and enclosure) on the total "system" performance, including efficiency, SPL Vs frequency (on-axis), input impedance (which contains terms related to loss resistance, radiation resistance and reactance). I suspect that for many readers, all of this seems a bit complicated and beyond their interest (if not grasp). With this in mind, I would suggest that the use of SPL (measured on axis, anechoically, and referenced to a specified distance, e.g. 1 metre), using some pre-agreed upon signal such as white or pink noise (at a given RMS voltage level at the loudspeaker input terminals), probably represents the best means for comparing loudspeakers with respect to their relative "efficiency". As I said in my last posting on the subject, if anyone would like more detailed information, let me know. My earlier suggestion for consulting the 1934 edition of "Loudspeakers" still stands. This truly excellent book, authored by N.W. Mc Lachlan, D.Sc. (published by Oxford at the Clarendon Press) provides one of the best reference sources on loudspeakers and their properties that I have ever come across. Best regards, John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Re: Time Alignment - was Re: 96kHz/24bit: WHY? From: John Dunlavy <102365.2026@compuserve.com> Date: 29 Apr 1997 08:32:00 -0400 My! My! My! Here we go again - back to the subject of impulse and step responses and how they relate to the audible accuracy of loudspeakers - and other components such as amps, CD players, D/A converters, etc. After reading some of the comments from those who claim that time-coherent loudspeaker performance (good impulse and step response) is irrelevant to audible accuracy, I can only echo the words of Tom Morley, Craig Dory and others who have already posted their views on the NET to Gary Eichmeier/Susan Andrus and others holding similar opinions. And, I cannot resist adding a few observations based upon well-known teachings of physics, math, and engineering - and upon our own research at DAL. In general, the notion that "time/phase alignment" of tweeter and mid-range drivers along the listening axis of a loudspeaker has no audible affect on the reproduction of complex musical transients, all other factors being equal, is simply not true. Neither is it true that significant time and phase errors introduced by a crossover network are not audible. For those who believe otherwise, ponder the following question: how many discriminating audiophiles would consider purchasing an amplifier that distorted the shape of square-waves at mid-range frequencies, by linear processes, so badly that their shape became unrecognizable? I suspect that not many could be convinced to do so! But if not, why would a knowledgeable audiophile consider purchasing a loudspeaker that could not accurately reproduce a square-wave - or an impulse, step, tone-burst, etc.? A measurement of frequency response (a curve of amplitude Vs frequency), taken alone, provides little more than an indicator of "spectral-balance" which, although an important audible parameter, is hardly an accurate indicator of overall reproduction accuracy. By itself, frequency response yields virtually no reliable information about how accurate a given audiophile component, including the loudspeaker, can reproduce the complex waveforms of many musical transients. To qualified engineers and mathematicians, the subject is simple: Fourier Waveform Analysis (FWA), Fast Fourier Transforms (FFT), and "Linear Systems Theory" (required courses for E.E.'s at many universities) provide the mathematical means for determining and analyzing all of the spectral properties of waveforms distorted by time-domain anomalies occurring within an audio reproducing system. This includes waveform distortion created by loudspeakers, crossover-networks, amps, etc. In particular, Linear Systems Theory teaches us that within an audiophile system, where one component (e.g., loudspeaker, amp, etc.) distorts the original waveform, the "end result" is the same - regardless of which component created the distortion. (For those who missed taking a course in Fourier Waveform Analysis, a good source to consult would be the ITT "Reference Data For Engineers", 5th Edition, pp. 42-1 thru 42-14.) Indeed, Fourier Waveform Analysis can be used to demonstrate that an "impulse" (a short duration pulse with a rapid rise-time) consists of amplitude components covering a spectrum of frequencies, over which variations in amplitude and phase are determined by the rise-time, shape and duration of the pulse. The "impulse-response" of any network or component, including a loudspeaker, may be used to determine, by Fast Fourier Transform analysis (FFT analysis), measurable parameters such as frequency-response, phase-response, step-response (the integral of the impulse), energy-time response, waterfall, etc. (Doug Rife's MLSSA system is a good example of the use of a "known sequence" of well-defined impulses to obtain an accurate measurement of virtually every measurable parameter of a linear system, including a loudspeaker.) While most "musical transients" probably do not generate as broad a spectrum of frequencies as a short-duration rectangular pulses used for measurement purposes, their accurate reproduction usually requires a bandwidth extending from an octave below their lowest frequency component to an octave above the highest component. Over this bandwidth, the entire reproducing system must exhibit an excellent "amplitude" response in both the frequency-domain (amplitude Vs frequency) and the time-domain (amplitude Vs time), to reproduce them with little or no audible waveform distortion. This requires a "system", including the loudspeaker, that exhibits what is frequently referred to as "time-coherent" performance - along with a flat "frequency response". Further, if a loudspeaker consists of an array of tweeter, mid and bass drivers that are not physically (or digitally) aligned to be time-coherent along the desired listening axis, the impulse and step responses will be "smeared" and a "minimum-phase", 1st-order crossover network will not suffice to yield a "flat" frequency-response curve (amplitude Vs. frequency). With a non time-coherent loudspeaker, obtaining an amplitude Vs frequency curve that is flat over the audio spectrum will usually require the use of a relatively complex crossover network, usually consisting of "high-order" hi-pass and low-pass sections, along with one or more single-frequency hi-Q sections to smooth-out sharp peaks and valleys in response introduced by phase anomalies caused by time delays between drivers along the listening axis. High-order waveform distortion created by such a design approach possesses a high probability of altering the audible accuracy of such a loudspeaker during the reproduction of complex musical transients. Much the same applies to crossovers that employ 2nd-order, 3rd-order, 4th-order (or higher order) networks which, because of the significant time-domain delays that accompany substantial phase-shifts, also alter the waveforms of transients, tone-bursts, etc. (A 4th-order network, with a 360 degree phase shift, stores energy at the crossover frequency by a full cycle, etc.) Lets now examine, in a little more detail, a hypothetical loudspeaker using non time-aligned drivers. For our example, let us assume that the "effective radiation-center" of the tweeter is 2 inches forward of a mid-range driver (mounted on the same surface). If a crossover frequency of about 7 kHz is chosen, at which 2 inches is approximately one-wavelength, the length of a tone-burst will be lengthened by about a full-cycle and a transient (impulse) will be lengthened by about 150 microseconds (0.15 mS). For most listeners, this might not be audible as a smearing in the time-domain but as an alteration in the "spectral or frequency" domain. For example, a "tic" sound might be heard more as a "tac" or a "toc". Likewise, a 10 cycle tone-burst at the 7 kHz crossover frequency, would have its normal "fundamental frequency" (1/t of the burst length) of about 700 Hz altered to about 636 Hz (a difference of approximately 11% introduced by the addition of a "phantom cycle", etc.). I suspect that such a lowering of the fundamental frequency of a burst would probably be quite audible to most listeners! Further, an audible difference in "spectral balance" might be perceived when listening to wide-band, white or pink, noise. Thus, it may be seen that alterations in the "amplitude-time response" of a system can significantly alter its "amplitude-frequency response", with the potential for changing the perceived sound of impulses, tone-bursts and other complex waveforms. Since musical transients often produce complex waveforms, one would expect that audible artifacts would be produced by a poor time/path-alignment of drivers and/or by the use of a high-order crossover network. In simpler words, an alteration of the amplitude Vs time components of a complex waveform virtually always results in an alteration of its amplitude Vs frequency components, the latter probably being the more audible to most listeners because of the presence of "new frequencies" that might not be perceived as possessing a "musical quality". Speaking of measurements, the old saw, "Measurements don't lie, but measurers do" is certainly a truism with respect to claims often made by loudspeaker designers and manufacturers. And all too many published measurements are the "fruits" of a graphic artist hired by marketing personnel - rather than the honest work of competent engineers and technicians. Sadly, the loudspeaker field seems to abound with self-appointed "experts" possessing few (if any) appropriate academic/professional credentials. And few companies have chosen to invest in expensive measurement equipment, a large anechoic chamber and a competent technical staff - to say nothing of the time, effort, knowledge and skills required to research, design and engineer a truly accurate loudspeaker. Certainly, measurements alone are not an infallible means for assessing the "audible accuracy" of loudspeakers! Likewise, so-called "trained ears" cannot be depended upon to ascertain whether "true" accuracy exists - unless the original live source (complex music) is available for a "direct comparison" within a suitable listening environment. It really requires a combination of a full set of accurate measurements, properly interpreted, and intensive comparisons with live music within an acoustically correct environment to critically assess the accuracy of a new loudspeaker design. However, in the absence of proper listening comparisons with live music, a full set of competent measurements (made within a good anechoic chamber at a distance of 10 to 12 feet) can, if properly interpreted, provide a reliable "indication" of a given loudspeaker's ability to accurately reproduce complex musical transients, etc. Likewise, the potential of any loudspeaker to accurately simulate a live musical performance "can never exceed that predicted by a full set of accurate anechoic measurements" (which must include impulse-response, step-response, frequency-response, phase-response, waterfall, energy-time response, harmonic and I.M. distortion, radiation patterns , a curve of input resistance and reactance, etc.). Here at DAL, we require a full set of accurate measurements, coupled with "critical listening comparisons with live music - in real-time", before certifying a new design for production. In regard to comparisons with live music, we regularly record live voices, quartets and quintets within one of our large anechoic chambers (using virtually-flawless instrumentation mics). We then compare the live music, played within a suitable listening environment, with the recorded music reproduced by our loudspeakers, located on the left and right sides of the musicians. A judgment as to the accuracy of the music reproduced by the loudspeakers is left to a panel of competent audiophiles and professional musicians. (To reduce the probability of biases attributable to participants knowing whether they are listening to the loudspeakers or the musicians, the musicians "mime" when they are not playing - thus eliminating any visual cues that might have altered listener perceptions.) A new loudspeaker design is only accepted for production when no audible or discernible differences are heard. During season, we also record the 85-piece Colorado Springs Symphony (playing within a concert hall that possesses some of the finest acoustics of any hall within the United States), which permits us to critically compare the reproduction accuracy of loudspeakers with very complex and demanding musical waveforms. Yes, we take "truly accurate reproduction" very, very seriously. In view of the fact that many of the most respected recording and mastering studios in the U.S. have chosen to use our loudspeakers as their "reference monitors", I am surprised that one Net Poster considers them to be "... run of the mill direct firing loudspeakers." Hmmm! I wonder if that person(s) can name any other loudspeakers (at any price, size, etc.) that come even close to exhibiting the measurable and audible accuracy of every loudspeaker model we make: e.g., true plus/minus 1 dB response (without "smoothing"), flat phase response, near perfect impulse and step responses, near-perfect waterfall and energy-time responses, very-flat resistive impedance curve (with less than about 30 deg. of reactance), symmetrical radiation patterns in both vertical and horizontal planes (with low side-lobe levels, very smooth "off-axis" response, etc.), near-zero edge diffraction (by use of efficient absorbing material between drivers and enclosure edges - for which I hold the patent), almost immeasurable levels of harmonic/I.M. distortion at levels up to about 95 dB SPL (re: 1 meter.), etc., all measured at a distance of 3 meters. (Yes, we do have one of the best-equipped loudspeaker measurement facilities anywhere, including two very large anechoic chambers that permit us to make precise, accurate measurements of virtually all measurable loudspeaker performance parameters.) While we have measured and are well aware of the performance deficiencies of many competing loudspeakers, we prefer to market our products on the basis of their own merits rather than comparing them to those designed and made by others - preferring to let magazine reviews and satisfied customers do our marketing for us. Having had three of our loudspeaker models honored as "Products of the Year" by the editors of Stereophile Magazine and having repeatedly had the sound of our loudspeakers at the yearly Consumer Electronic Shows in Las Vegas (and earlier in Chicago) chosen by several magazine editors as "The Best Sound At The Show", we believe that our efforts speak well for the importance of accurate measurements. However, serious audiophiles who sincerely care about the validity of claims made by many designers for the accuracy of their loudspeakers should demand better proof, in the form of competently-made measurements of relevant performance attributes. The sad truth is that very few loudspeaker designers possess relevant academic/professional credentials and fewer possess adequate equipment and facilities for competently measuring even the most elementary properties related to accuracy. Yep - its true! Lets all keep searching for the truth with regard to what constitutes "truly accurate reproduction" and the directions we must take to improve our chances of coming ever closer to an ideal system - perhaps before the arrival of the 25th century (or the 26th, 27th - Hmmm!) Best regards, John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Time Alignment From: Dunlavy Audio Labs <102365.2026@compuserve.com> Date: Mon, 5 May 1997 19:20:54 -0400 [Moderator's Note - This article was composed by John Dunlavy. Sorry about the MIME encoding in places, but I have only so much time to edit this stuff -- bt] With regard to recent postings concerning the audibility of time/phase-coherent loudspeaker performance, it appears that my comments on the subject have created quite a stir among those who choose to believe that competent measurements of impulse, step, tone-bursts, etc. do not correlate with the audible accuracy of a loudspeaker. It is good to see that so much interest exists with respect to such an interesting and important topic. In the end, after all the dust has settled and the truth begins to rise to the surface, I hope that the real winner will be the average audiophile - a person frequently bombarded with the subjective opinions of self-appointed experts who want everyone to believe the subjective opinions they hold as "truths". To many of them, science and engineering are disciplines practiced in a dank and dusty lab by creepy old kooks whose research cannot be trusted to reveal truths about "real-world performance". (Shades of the Frankenstein era - Ugh!) To many of those who distrust measurements and/or their interpretation, the "subjective approach" to assessing loudspeaker accuracy is the only means available that can be trusted. And, considering all of the false and/or misleading technical performance information used in the past to market new loudspeakers, perhaps some of their gripes have some legitimacy. But not all who pursue the design of loudspeakers are money-hungry pseudo-engineers - some are well-educated audiophiles who combine a serious love of music with a serious desire to reproduce it with the greatest "audible accuracy" possible. Their sincere desire is to narrow the gap between the sound of the original live performance and its reproduction by electronic means. (This, of course, includes the art and technology involved in making audiophile-quality recordings.) But every true designer, faithful to his academic learnings and real-world experiences, recognizes that his quest will always be one that can never achieve a "truly perfect" reconstruction of the original live performance. Indeed, with respect to the postings of several who have been quick to chime in with their opinions regarding past "live Vs recorded" demonstrations by AR and others, I would like to add a quotation that pre-dates them by several decades. The following quotation regarding the old Edison Phonograph (circa1915) may be found on page 202 of the book, "From Tin Foil to Stereo" (by Oliver Read, et.al., The Bobbs-Merrill Co.,1976): "... a distinguished group of visitors ... gathered to witness the first demonstration of the new Official Laboratory Model Edison phonograph. The beautiful opera soprano, Anna Case ... was also present. To make a long story short, Miss Case sang for the assembled guests in the library. More than this, she sang in direct comparison with her recorded voice as reproduced on the new Edison instrument. ... To the amazement of all, they were unable to detect any difference between the voice of the singer and that coming from the phonograph." Gosh, it seems that times don't change - only the venue, the equipment and the "caliber of audible discernment". Hmmm! Perhaps, it might be instructive to examine how most (if not all) of the "live-Vs-recorded" demos were accomplished. For example, the AR demo I attended back in the mid-t-late 50"s (to the best of my recollection) was in New York's Grand Central Station - where the acoustics and the background noise level hardly permitted a very meaningful comparison to be made. Other such comparisons, of which I am aware, have been made under similarly poor listening conditions. To have any meaningful value, I believe that a live Vs recorded comparison must take place within "good acoustical space", where ambient background noise levels are at or below audibility. The live musical group (or vocalist) should be located mid-way between the left and right channel loudspeakers and at the same distance from the back-wall. The demo room should be symmetrical and the loudspeaker/musicians should be positioned symmetrically with respect to the room boundaries (walls, etc.). The recording should have been made with a pair of nearly perfectly-matched, "instrumentation-quality" microphones within a room of fairly large size, possessing anechoic or near-anechoic properties. I could add several other requirements, which some readers will probably want to mention, but these should suffice to indicate the necessity to control as many audible variables as possible within the record-playback chain. Our own live Vs recorded research sessions have been conducted in strict observance of the above - plus some! By making the recording within a good anechoic environment and comparing the live Vs recorded sound played back within a different but acoustically-good room with appropriate dimensions, an opportunity exists to critically examine how close the live and recorded sounds are matched. In observing the above criteria, we have been able to demonstrate that several of our loudspeaker models are capable of reproducing musical groups, such as a string quartet or quintet, with an accuracy that prevents seasoned audiophiles from achieving a score no better than random guessing. Yep, that's the truth. (And, we have performed such comparisons on many occasions to ensure that our measurements were meaningful and trustworthy.) In a private E-mail to me (and I believe on RAHE), Howard Ferstler, a well-known writer in audio journals, noted the above-mentioned AR and other live-Vs-recorded sessions as proof that such sessions reveal little about the "accuracy" of loudspeakers. In this context (venues with poor acoustics), I would certainly agree. I must also agree with those that hold that "phase linearity", taken alone, does not provide a reliable means for judging the audible accuracy of loudspeakers. But impulse, step, tone-burst, etc. do, collectively reveal a great deal about the potential of a loudspeaker to accurately reproduce complex musical transients. Equally important, as I have often said, is the directivity of a loudspeaker over the surface of a "hypothetical sphere" (4-pi space). A reasonably-broad, on-axis radiation lobe, symmetrical about the listening axis (of more or less constant beamwidth Vs frequency) and possessing side-lobes of reasonably low intensity, is also essential to achieving "realistic" reproduction within most rooms. Anyway, I fail to understand how "live-Vs-recorded" sessions, conducted 30-plus years ago under poorly-controlled circumstances, have any relevance in our present, much more sophisticated world, where much more sophisticated expectations exist with respect to "true accuracy". Properly implemented and interpreted, a "complete set of measurements", made under proper anechoic conditions at the normal listening distance of about 10 ft. (on the designated listening axis), with a full set of lab-quality measurement equipment, by a competent engineer, can reveal - quite accurately - the "potential" a loudspeaker possesses to accurately emulate a live musical performance. Indeed, how can the "true audible accuracy" of any audiophile device, be it an amp, pre-amp, C.D. player, loudspeaker, etc. ever exceed its "measured accuracy"? Please, please - tell me how? HMO! I must go for now! It's quitting time and I still have a pile of work to accomplish before heading home. Lets keep the dialog going - but let's also keep it at a reasonably informed level, with a minimum of noise. Best regards, John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Time Alignment From: Dunlavy Audio Labs <102365.2026@compuserve.com> Date: Thu, 8 May 1997 19:07:26 -0400 A few posters on r.a.h.e. have recently questioned whether some loudspeakers employing certain design criteria, such as "time/phase aligned drivers" can provide accurate or "life-like" stereo-imaging within typical room environments. I would like to take this opportunity to share the results of our research at DAL and the conclusions we have distilled from a combination of theoretical work, extensive measurements and numerous critical listening sessions - often using several local audiophiles possessing proven discernment capabilities and a lack of biases. First, two quite different design goals can be defined: 1) sound quality and imaging that replicates, as closely as possible, the audibly important elements of the original musical instruments and soundstage (the acoustical properties of individual instruments, their angular location and distance from the listener, the ambiance, soundstage, acoustical-space and other audible properties of the recording venue); 2) sound quality and imaging that is not truly accurate but which provides soundstage, ambiance and other properties that are audibly pleasing and/or satisfying to the intended listener. While I personally prefer adhering to the first set of design goals, I do not think it entirely fair to criticize the motives or tastes of those who might choose the second. After all, the purpose of music is to appeal to and satisfy the "personal tastes" of many different listeners - even those who wish to add their own alterations to the tonal-balance, imaging, soundstage, etc. And, being realistic, it is the 2nd set of design goals that is most often chosen by loudspeaker designers - because it yields products that appeal to the tastes of the majority of listeners, who are concerned more with "effect" than "true measurable/audible accuracy". But we at DAL have chosen "objectively determined accuracy" as our design goal and it has proven most successful among knowledgeable audiophiles who want their music reproduced (within the limitations imposed by their listening room) as it was heard by a competent recording engineer who tried to capture all of the audible attributes and nuances of the original live performance. I have said it many times before and I will say it again: a complete and accurate set of anechoic measurements made by competent engineers using the latest measuring equipment can accurately predict the "potential" of any piece of audio equipment, including loudspeakers, to accurately simulate the original live performance (properly recorded). Notice that I used the word "potential" - for one must also consider the acoustics of the listening room and the accuracy of each of the other components comprising the chain of equipment used to reproduce the sound. But if measurements reveal that a given piece of equipment does not possess all of the relevant performance properties required for truly accurate reproduction of complex musical waveforms, that equipment will never possess the same potential for providing accurate reproduction as one that does. Simple logic reveals the truthfulness of this statement! It should be obvious that the maxim "a chain is as strong as its weakest link" is equally true of the chain of components that comprise a typical audiophile system. And, sadly, the loudspeaker is virtually always the "weakest link" in the system (amps, pre-amps, etc. generally offering minimal differences in audible accuracy). Thus, it is usually true that the biggest improvement can be achieved by choosing the most accurate loudspeaker within one's budget limitations. So how does the typical audiophile discern which loudspeaker is the best one for his/her particular system. I would suggest beginning by reading reviews of loudspeakers in credible magazines such as Stereophile. I recommend Stereophile in the U.S. because I sincerely believe most of their reviewers try very hard to assess and report on the audible merits and/or deficiencies of most loudspeakers. But readers must always "read between the lines", questioning the relevance of expressions like "sweet-sounding", musically-satisfying", "full-sounding", "big-soundstage", "great dynamics", etc. because these are subjective qualities that vary from one reviewer to another and may not correspond to the priorities of the potential purchaser. And give proper attention to the measurements made by John Atkinson; especially impulse, step, amplitude response Vs frequency, waterfall, etc. Compare them with those of other speakers in the same price range before making a decision. But the final decision must be made by listening to the loudspeakers being considered, within a good listening environment and with the same good program material. While listening, don't let a hungry salesperson "coach" you to "hear" "differences" that may not be relevant to "accurate performance" or differences that might not even exist. (Remember that we humans are very vulnerable to suggestion and cleaver persuasion.) Also, differences in efficiency is an important consideration when performing listening comparisons between loudspeakers. For example, a loudspeaker with an efficiency only 1 dB higher than another will frequently sound more "detailed" and with "greater presence", even though its accuracy might be the poorer of the two. But the ear is virtually incapable of detecting a 1 dB difference as added loudness when listening to music (remember, 1 dB is the average, minimum detectable difference in loudness when listening to a fixed-frequency sinewave at most audible frequencies, within a quiet listening environment). The bottom line might be expressed as follows: measurements don't lie but measurers often do. If a full set of measurements do not exist - perhaps you should question why? By the way, Gary Eickmeier accepted my challenge a few days ago to produce a full set of competent measurements (made anechoically at a distance of 10 feet, or more) for a loudspeaker of any other manufacturer (regardless of its price) that would prove their loudspeaker to be as accurate as ours. Accordingly, I have forwarded a full set of "final Q.C. measurements" to Gary by UPS (ones made of a production pair recently shipped to a dealer) and await his reply. I have also sent Gary a copy of my 1983, peer-reviewed paper on the audibility of loudspeaker phase-distortion. Our present engineering staff and myself are presently up-dating the paper, using the latest measuring equipment, a large number of local audiophile listeners, a good listening environment, etc. to determine if there are any additional observations to report that were not mentioned in my 1983 paper. Lets keep the dialog going - in favor of discovering the real truth about the meaningful performance properties of loudspeakers - devoid of the "high-powered rhetoric" and "questionable claims" so often encountered in advertisements and here on the NET! Caveat Emptor! Best regards, John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Accuracy From: Dunlavy Audio Labs <102365.2026@compuserve.com> Date: Thu, 15 May 1997 19:07:26 -0400 Sorry that I have not had an opportunity to participate on the NET the past few days - but business chores must come first! However, I have managed to read several recent postings related to the topic of "loudspeaker accuracy". Its good that this thread has gained so much attention and solicited so many comments from interested audiophiles. Some of the postings, though, have expressed viewpoints that need to be examined and discussed with respect to what we know about the different types of linear and non-linear distortion products and the correlation between their measurability and audibility. For example, a few posts continue to propagate the belief that waveform distortion in the "time-domain", attributable to physical mis-alignments of drivers along the listening axis, is not audible. The same applies toloudspeakers using a high-order crossover network that alters the length of a tone-burst or an impulse. Such time-domain alterations of a waveform create new frequency components that may or not be audible, depending upo n their duration and amplitude. To proclaim otherwise is to deny the validi ty of Fourier analysis, accurate laboratory measurements and competently conducted blind and double-blind listening comparisons within an acoustically suitable listening environment. Let me restate this using slightly different phrasing: if the different frequency components of a complex waveform undergo different delays in th e time-domain (due to the mis-alignment of drivers along the listening axis ), potentially audible artifacts are created by the production of "new frequencies" that were not present in the spectrum of the original waveform. Ask any competent professor of electrical engineering or mathematics for verification! I would also like to repeat a rhetorical question I have frequently posed to serious audiophiles: "If you seek true accuracy in the reproduction of music, would you choose an amplifier, pre-amp or CD source that could not accurately reproduce a square-wave, tone-burst or an impulse? If not, wh y would you choose a loudspeaker that could not accurately reproduce such signals?" In this regard, the old maxim "a chain is as strong as its weakest link" might apply equally to the components of an audiophile system! A good Uni course in "linear systems theory" certainly teaches th is to be true. (And, don't forget the value of a good course in Fourier Analysis!) Another loudspeaker performance property frequently ignored is the radiation pattern over the full audible spectrum (measured over the surfa ce of a hypothetical sphere at a distance of about 10 feet). A proper radiation pattern is extremely important for attaining accurate reproduction of complex musical waveforms within most listening rooms. Fortunately, if a sound reflected from a nearby wall, ceiling, etc., exceeds the arrival of the direct-path sound by more than about 5 milliseconds, the human hearing process usually recognizes it as a reflected sound and it does not directly contribute to a perceived alteration in "system frequency response". An exception to this, however, is often evident at lower frequencies - where such reflections are frequently perceived as adding a "boomy" quality to music. The problems created by reflections from the floor and ceiling can be mitigated by using a "symmetrical array" of drivers in the vertical plane , i.e. tweeter in the center, surrounded above and below by mid and woofer drivers, located symmetrically with respect to the tweeter. But this alon e is not enough - the spacing between the drivers, along with appropriate crossover frequencies, must be carefully chosen to ensure a reasonably constant "beamwidth" in the vertical plane. Otherwise, more energy will be reflected from relevant floor and ceiling surfaces at some frequencies th an at others. Of course, this is most important at frequencies in the range above about 300 Hertz. Recent comments being made by Gary Eickmeier, implying that I should investigate the importance of "radiation patterns" and that my loudspeake r designs have not considered their importance are, of course, completely unfounded. Indeed, I would like Gary to name any measurable or audible performance property that we did not fully take into account when designi ng every one of our loudspeaker models. Likewise, some of Gary's postings have implied that I (or my engineering staff) have not made sufficient measurements to fully quantify the performance of our loudspeakers and that they are merely "ordinary" loudspeakers with "ordinary performance". (In this regard, I am reminded of a World-War II boast by Adolph Hitler, who compared England to a "chicken " whose neck he was going to wring. Winston Churchill's reply was sweet an d simple: "Some chicken, some neck!") Several postings ago, I challenged Gary to provide measurements (competently made) of any other loudspeaker that exhibited either comparable or more accurate performance than those being made by DAL. A few days later (5/1/97) he accepted the challenge in a posting to me. I have forwarded a set of measurements (selected at random) made on a production pair of our SC-IV (mid-price) loudspeakers. I also included a copy of a "lengthy, peer-reviewed technical paper" I authored, entitled "The Importance of Phase Linearity on the Perceived Quality of Loudspeaker Reproduction". (I await Gary's comments.) I presented this paper before a joint Engineering Seminar/AES meeting, he ld in the Electrical and Electronic Engineering Department at the Universit y of Adelaide (in Australia) on the 19th of July, 1983. The paper was accompanied by a demonstration of a small-size loudspeaker with two identical drivers, specifically designed to demonstrate audible, on-axis differences between 1st and 2nd-order crossover networks, where the on-ax is "anechoic frequency-responses" were virtually identical. Test signals, such as impulses, square-waves, tone-bursts, pink & white noise, etc., we re used to demonstrate the existence of audible differences between the 1st & 2nd-order crossover properties. Virtually all of the 30-plus persons present (mostly engineers, audiophiles, etc.) could easily perceive the differences. On 5/9/97, Gary posted a denial of ever having accepted the above mention ed challenge and has yet to supply a "complete set of competently made measurements that demonstrate the existance of a loudspeaker with accurac y comparable to that of any of our standard models. (The offer is still open!) Meanwhile, I have undertaken a project to repeat much of the 1983 effort, but using a more sophisticated system and means for comparison - both within an anechoic environment and a typical listening room. When complete, I will attempt to find the time to author a paper for submissio n to the AES (or other suitable journal) for publication. I will also post relevant results of the project on the NET. With regard to radiation patterns, our considerable research into the subject has revealed that very wide beamwidths (vertically or horizontall y) or bi-directional radiation patterns can create a "perception" of pleasan t ambiance that appeals to many listeners - but which does not represent an accurate reconstruction of the original sound-stage, with respect to accurate imaging and true spectral balance (considering reflections withi n the average listening room). Of course, there are exceptions to every ru le and I doubt there will ever be a loudspeaker that will fulfill the listening desires of every audiophile. What we have attempted to accomplish is the design of loudspeakers which, when properly positioned within a typical audiophile's listening room, can reproduce music and voices with uncanny accuracy - given accurate recordings and good equipme nt (amps, etc.). One aspect of our human listening "process" that many, including Gary, ma y have missed is the ability of the average listener to "tune out" most reflections within a listening room of modest size after listening for a period of time - typically exceeding 20-30 minutes. During this initial listening period, most of us sub-consciously process the sound we hear an d reject the audible effects of reflected energy, if it arrives beyond the "fusion time" of our hearing and is attenuated at least 6-10 dB relative to the direct-path sound from the loudspeakers. Thus, if the loudspeakers w e are listening to exhibit well-behaved radiation patterns, with side-lobes attenuated more than 6-10 dB relative to the maximums of the main-lobe (V s frequency), and the listening room possesses reasonably good acoustics, o ur listening experience will generally be perceived as "good" and "faithful" to the original live performance. A significant amount of experience her e in DAL's listening room has revealed this to be true for countless visitors. Meanwhile, rest assured that neither I nor DAL take our claims for accuracy "lightly". Indeed, I repeat my challenge for anyone (including any company) to submit a full set of competent measurements (made at an on-axis distance of at least 8-10 feet) that prove another loudspeaker to be as accurate as those we manufacture. (We would also like to know of any other company that verifies the accuracy of every pair of their loudspeakers, including pair-matching within better than about 0.25 dB, by a complete set of anechoic measurements before being accepted for shipment). And, although many posters have repeatedly voiced the opinion that many loudspeakers, including the AR-3's, are capable of emulating a live group of musicians, with an accuracy that prevents most people from discerning the "live from the reproduced" music", we have not found this to be true. During properly controlled A-B comparisons within an acoustically good listening environment, most audiophiles have experienced very little difficulty distinguishing between live music and its reproduction by most loudspeakers, regardless of their make, price or size - except for our own models SC-IV thru SC-VI. (This includes some of the most expensive, large electrostatic types.) Perhaps doubters might want to visit our plant, as many have, and verify that this claim has substance. Having said all of the above, the bottom line must remain the individual's choice - based upon their individual perception of what sounds best (or most accurate) to them. However, for those who thirst for "truly accurate loudspeakers", begin by perusing a complete set of accurate anechoic measurements and see how well the loudspeakers reproduce an impulse and a step. If they can do so with near perfection, it is a reasonably good sign that they will also be able to accurately reproduce complex musical waveforms. Of course, assessing a full set of measurements (for those who know how) is a yet better means for determining the "potential accuracy" of the speakers. Measuring each loudspeakers insures that the consumer will receive a product that will sound identical to the product that they auditioned or read about in a subjective review. If a manufacturer does not measure each loudspeaker, how can they guarantee the audible performance of the individual units that are produced in respects to equalling the audible performance of the original units or units in a particular showroom, consumers home, etc.? But listening is still the final arbiter, and is the most important aspect of an individuals choice for a high end audio product. Must go for now! Best of listening - and keep asking the right questions! John Dunlavy ======== Newsgroups: rec.audio.high-end Subject: Audio truths (was "Time alignment") From: John Dunlavy <102365.2026@compuserve.com> Date: 6 Jun 1997 17:49:49 -0400 Thanks to Mary and Allan Moulton (of Aussie Land, I believe) for the interesting comments they posted on the 8th of May. Sorry it has taken so long to reply. In an attempt to prove their points, much of their discourse appears to be based upon a modernist/relativist misinterpretation of the teachings of logic and criteriology (uni courses I took back in 1948 before I switched my major to Physics). This may be seen from Allen's comment that, "... the true audible accuracy of any device, be it an amp, pre-amp, C.D. player, loudspeaker, etc. can exceed its measured accuracy; because expectations, which are (in part at least) determinative for what we count as true, cannot themselves be measured." Here, we are told that audible accuracy can exceed measured accuracy, based upon an individual's expectations being satisfied. It seems to me that this is merely another way of defining "self deception"! For, if truth exists only in the perception of the beholder, we eventually degenerate into a state where "truth" ceases to have any meaning and is replaced by the whims and self-deceptions of each person. (Gulp! - what a bonanza for sales-persons purveying snake oil, buzzard salve and flooby-dust.) And if, as Allen believes, "expectation" becomes a significant determinate of what we accept to be "true", I cannot help but believe that we leave ourself open to self-deception - to say nothing of becoming victims of clever marketing/advertising jargon that preconditions us with "expectations" that have no basis in reality. This is where true science enters the scene, for it involves a search for and discovery of new knowledge whose truth and relevance can be confirmed by proven, objective means. Without the knowledge and wisdom provided by "true science", I rather doubt that mankind would have ever freed itself from the bondage precipitated by false beliefs held to be "true" during the "Dark Ages"! However, in today's more enlightened era, an "expectation" of achieving a higher level of "real truth", confirmable by repeatable objective means, can (and should) motivate us to invest in basic research, the development of new technology and the design of products with levels of performance previously unavailable - performance that can be verified by the proper use of tools provided by "true science and true engineering". In this regard, however, the old Latin dictum "caveat emptor" (let the buyer beware) still deserves attention, especially in the audiophile market - where performance specs seldom have much real meaning because they probably originated in the fertile imagination of an advertising "genius" eager to raise profit margins rather than risk money on costly research and engineering. Hopefully, what I have said above will help readers to understand that expectations which have no basis in "objective truth and/or reality", and cannot be verified by a "complete set of competent/accurate measurement's", can eventually lead an audiophile to lose faith in their pursuit of ever more perfect reproduction of sound and music. And, for those who pursue expectations that do not match reality, "accuracy of reproduction" becomes relative to each individuals expectations - and real/verifiable truth eventually loses all objective meaning - while it empties our wallet. Best of listening! John Dunlavy ======== From 102365.2026@CompuServe.COM Mon Aug 25 18:51:11 1997 Newsgroups: rec.audio.high-end Subject: Determining Loudspeaker Accuracy From: Dunlavy Audio Labs <102365.2026@CompuServe.COM> Date: Mon, 25 Aug 1997 19:51:11 -0400 Recently, there has been quite a number of discussions and comments regarding loudspeaker accuracy and how this accuracy is achieved or determined. Most audiophiles are familiar with the use of "frequency response" (a plot of relative amplitude versus frequency) as a means of assessing the spectral-balance of a loudspeaker. Taken alone, however, frequency response only provides information regarding a loudspeaker’s performance in what is called the “frequency domain”. Relatively few audiophiles appear to be fully aware of the equal importance of assessing the performance of a loudspeaker in what is known as the "time-domain", which encompasses such measurements as impulse-response, step-response (the "integral" of the impulse-response), energy vs. time and the cumulative-spectral-decay (waterfall). Another important measurement that relates to both frequency and time domains is phase-response. A flat curve of phase versus frequency generally implies a flat frequency response as well as good impulse and step responses (and vice versa). Also, a good step-response signifies both a flat frequency and phase response. While frequency response (measured at a distance of about 10 feet to be meaningful) may permit an assessment of spectral-balance, it cannot reveal whether a loudspeaker has the potential to accurately reproduce complex musical transients, inner-detail, transparency and openness. Only time-domain measurements, such as "impulse-response" and "step-response", can disclose this important information. Informed professional engineers recognize that, all other factors being equal, a loudspeaker which exhibits good impulse and step responses will reproduce musical transients with far less blurring and or smearing than a loudspeaker with poor impulse and step responses. Technically speaking, frequency-response, phase-response, impulse-response and step-response are, in many respects, inter-related. Indeed, plots of both amplitude and phase versus frequency can be derived from an impulse-response measurement by computation using Fast Fourier Transforms (FFT). However, the reverse is not possible. For example, imagine a hypothetical loudspeaker with a tweeter located at a distance of 8 feet from the listening position, a mid-range at 9 feet and a woofer at 10 feet. Despite such a poor alignment of the drivers in the "time-domain", with about one millisecond separating the arrival of each signal, it is relatively easy to design a crossover network that will yield a very "flat" curve of "on-axis frequency-response". But the impulse and step responses exhibited by this configuration of drivers would look pretty awful, with multiple peaks and ringing - which no crossover design can correct (excepting an active digital type). However, an excellent impulse-response and a proper step-response imply “flat” curves of both amplitude and phase versus frequency. But, contrary to the false beliefs of some designers, obtaining such performance requires a near-perfect, pulse-coherent alignment of all drivers along the intended listening axis, combined with a properly designed, 1st-order, minimum-phase crossover network. Loudspeakers that do not exhibit a good impulse or step response may sound good, pleasing, sweet, musical and so forth, but can never sound truly accurate any more than a poorly corrected camera lens can focus all colors in a sharp and coherent manner. For the same reasons that impulse-response, step-response and square-wave reproduction are considered important considerations for determining the audible accuracy of an amplifier. C.D. player, phono cartridge or a DAC, so also are they important for assessing a loudspeaker's ability to recreate the full excitement of the original live musical experience. Best Regards, John Dunlavy CEO Dunlavy Audio Labs ======== Subject: Re: von schweikert vr4 or definitive tech bp2000 From: awrigby@aol.com (AWRigby) Date: 1997/08/28 Message-Id: <19970828231201.TAA15113@ladder02.news.aol.com> Newsgroups: rec.audio.opinion [More Headers] Here is John's response: Normally, I believe it is inappropriate for a manufacturer to engage in NET dialog that might be interpreted as being negative with respect to products designed and manufactured by a competitor. However, recent postings on the NET (and on RAHE) that have attempted to compare certain performance attributes of loudspeakers designed and manufactured by Von Schweikert Research with those of Dunlavy Audio Labs seem to justify informed comment. This seems especially true where comments and comparisons have been made, largely based upon claims and specifications that appear not to be consonance with well-known theory and practice - as they are known by competent members of the engineering and scientific community. Sadly, lacking appropriate technical underpinnings, a lot of sincere audiophiles fall prey to pseudo-scientific advertising jargon that simply has no basis in reality. The sole purpose of this post is to attempt to "set the record straight" regarding the design and performance of DAL products by merely reciting the teachings of well-known and well-understood theory and practice related to acoustics, network theory and loudspeaker design. A good example is the 26 Aug. 1997 posting to RAHE (Subject: Von Schweikert vs Definitive) by Jim Wald, who stated (in part): "You fail to mention that Von Schweikert speakers, like Dunlavy speakers, are also time aligned and phase coherent." And, "I disagree with the Dunlavy design philosophy mostly in two areas: First, the use of very flexible cones (as opposed to Von Schweikert’s use of carbon fiber and other stiffening designs) results in cones that have way too much flex, and thus, distortion. Secondly, the first order (6 db) crossovers in the Dunlavy’s results in TWO bad tendencies: (1) this results in vocals (and other midrange coming through the woofers, upper bass and high frequency information coming through the mid-range, and mid-range coming through (and distorting at louder levels) the tweeter. The 4th order crossovers in the VSR eliminates this. (2) the first order crossovers are also a big part of the reason that the Dunlavy’s have such a narrow sweet spot and the 4th order crossovers (and the rear ambient tweeter) are a big part of the reason why the Von Schweikert’s have such a wide sweet spot." And, "A more thorough debate on this subject would be best left to John Dunlavy and Albert Von Schweikert, but those ARE the cogent points. Strictly from a design standpoint, the Dunlavy’s have more in common with White Van Speakers that VR-4’s have in common with 901’s. But, I don’t think that either is a fair comparison to two excellent speakers." First, those who concur with these opinions might want to examine the claims being published for certain manufacturers products to determine if they are in consonance with the teachings of both engineering and physics. It might also add credibility to a loudspeaker manufacturer’s claims if they openly published anechoic measurements (with guaranteed accuracy), made at the normal listening distance of 10 ft., of such important performance attributes as 1) frequency response (without smoothing), 2) impulse-response, 3) step-response, 4) waterfall, 5) energy-time curve, 6) impedance (resistance/reactance vs frequency), and 7) non-linear distortion vs frequency at reasonable volume levels. (DAL does and guarantees the accuracy of all published measurements/specifications for each of its loudspeaker models.) Let me begin by comparing some of Jim’s (and Von Schweikert’s) comments with what well-known theory and practice have to teach us. 1) "... Von Schweikert speakers, like Dunlavy speakers, are also time aligned and phase coherent." Not so! To begin, I believe Von Schweikert claims that his loudspeakers use 4th-order crossovers which he refers to as being phase-coherent. But a 4th-order network adds the separate mid-range and tweeter on-axis radiation components at a phase angle of 360 degrees (at the crossover frequency). And, a 360 degree difference in phase creates a "full cycle" of time-domain delay between the separate mid and tweeter radiation components, resulting in performance that is not phase-coherent. Accurate measurements of VS loudspeakers (including impulse, step, waterfall, energy-time, etc., made at an on-axis distance of 10 feet within one of DAL’s large anechoic chambers) fully confirm that they exhibit neither time-aligned nor minimum-phase performance, as evidenced by grossly-distorted impulse and step responses and a relatively slow initial roll-off in the waterfall response. (Our measurements are accurate and we stand behind them!) 2) "... the use of very flexible cones (as opposed to Von Schweikert’s use of carbon fiber and other stiffening designs) results in cones that have way too much flex, and thus, distortion." Not so! Here, it is assumed that carbon-fibers, Kevlar fibers, etc. are ideal materials for constructing loudspeaker cones. This is simply not true. Indeed, all of the bass and mid-range drivers using carbon and Kevlar fibers that we have measured over the past several years have exhibited very poor impulse and step responses - not to mention poor amplitude/phase vs frequency curves. Part of the reason for this is that the composition, shape, internal-damping (within the cone material), edge-damping, etc. play a far more important role with regard to achieving accurate amplitude/phase response vs frequency, low non-linear distortion, and more constant directivity than the incorporation of high-strength fibers such as carbon and Kevlar, which only affect "flexure", etc. in a very minimal way - despite advertising to the contrary. The woofer, bass, mid and tweeter drivers we use are chosen on the basis of whether they exhibit: (1) very low levels of non-linear distortion (harmonic and IM), (2) the best possible impulse and step responses, (3) very-flat frequency and phase response vs frequency, and (4) more-or-less constant directivity throughout their intended frequency range. Drivers using cones incorporating carbon and/or Kevlar fibers have not been found that meet our very stringent requirements for accuracy, especially with respect to impulse and step response. These are, perhaps the two most important performance attributes that affect audible accuracy - especially with respect to the accurate reproduction of complex musical transients, etc. I believe that Stewart Pinkerton’s post of 27 Aug. (patent3popmail.dircon.co.uk (Stewart Pinkerton)), also covered this point rather well! 3) "... the first-order (6 db) crossovers in the Dunlavy’s results in TWO bad tendencies: (1) this results in vocals (and other midrange coming through the woofers, upper bass and high frequency information coming through the mid-range, and mid-range coming through (and distorting at louder levels) the tweeter. The 4th order crossovers in the VSR eliminates this. (2) the first order crossovers are also a big part of the reason the Dunlavy’s have such a narrow sweet spot and the 4th-order crossovers (and the rear ambient tweeter) are a big part of the reason why the Von Schweikert’s have such a wide sweet spot." Not so simple! With respect to the first part (1) we again encounter an over-simplification of the complex choice a designer must make between minimizing "doppler distortion" and minimizing "time-domain/phase distortion". Unacceptable levels of doppler distortion can result from a mid-range driver being fed too much bass energy by an incorrectly-designed 1st-order crossover or by an inappropriate choice of the crossover frequency. However, while a 4th-order crossover might reduce the level of doppler distortion at some frequencies (compared to a poorly-designed 1st-order network) its use always results in significantly blurred reproduction of complex musical transients, created by the time/phase distortion inherent in all higher-order networks (as mentioned above), which exhibit considerable waveform distortion, ringing, etc. Only a properly designed first-order crossover network can accurately reproduce an impulse, step and complex musical transients without blurring and ringing. Also, doppler distortion can be significantly reduced with 1st-order networks by the proper selection of crossover frequencies, the use of multiple bass and mid drivers (as in our vertically-symmetrical array), and by the judicious use of series-shunt components that only affect frequencies well below and above the crossover frequency. With respect to (2) above, i.e., "the first order crossovers are also a big part of the reason that the Dunlavy’s have such a narrow sweet spot and the 4th order crossovers (and the rear ambient tweeter) are a big part of the reason why the Von Schweikert’s have such a wide sweet spot" - not so, again! Again, not so! Of course, it can depend upon the definition given the term "sweet spot". If one refers to the sweet spot as being the width of the listening area over which the audible spectral balance does not significantly change, perhaps the VS loudspeakers may qualify as having a "wide sweet spot"! However, all DAL loudspeaker models exhibit equally wide "sweet spots" with respect to unchanging spectral balance over a very-wide horizontal listening area. (This is easily proven by the perceived constancy of spectral-balance as one cruises around the room while listening to DAL loudspeakers.) But if the "sweet spot" is defined as "the width of the listening area over which precise stereo imaging can be discerned by a critical audiophile listening to an accurate recording", the important performance requirement is the precise pair-matching of the loudspeakers with respect to their amplitude vs frequency and phase vs frequency responses, both on and off the intended listening axis. This is because the human hearing process primarily determines the direction of a given sound by the brain separately comparing the amplitude and phase components of the first arrival sound components as they are heard each ear. Thus, a sound appears to be arriving from directly in front of a listener if it arrives simultaneously at both ears and is equal in both amplitude and phase. (One should not forget that phase is related to time by the distance a soundwave travels during each degree of phase at the frequency of the sound.) Surprisingly, our ability to discern small differences in distance and the angle-of-arrival of a sound is determined more by differences in phase than in amplitude - a process not yet fully understood by science and medicine! It seems that the vertical angle of arrival of a sound is largely sensed by a "pre-processing" of the signal within the pinna (outer ear), which somehow reshapes the spectrum of the sound in the time/phase domain. Perhaps, this explains why it is much easier to discern the angle-of-arrival of a sound if it is a pulse (tic), whose frequency components cover a wide spectrum. Thus, it may be seen that precise "pair-matching" of stereo loudspeakers is a prime requirement for loudspeakers intended to exhibit precise stereo imaging. In this regard, the VS loudspeakers we have measured and heard tend to exhibit a pleasant spectral-balance over a wide horizontal area. But over this wide area they do not seem to be capable of providing a "precise, focused center image or stable sound stage" over which a centrally-located listener (equidistant from each loudspeaker) can determine the exact location of a given instrument or voice. With a precisely-matched pair of stereo loudspeakers having properly controlled symmetrical radiation patterns (both horizontally and vertically), flat amplitude/phase vs frequency properties, excellent impulse/step responses, etc, a competent listener, seated exactly equidistant from both loudspeakers, can marvel at his/her ability to pin-point the source of each sound being heard. With respect to a loudspeaker being able to replicate "concert hall sound" by reflecting sound off of the side and back walls using rear and/or side-firing drivers, consider that our perception of distance and space is largely controlled by the time delay of the sound reflected from the distant surfaces of the room or concert hall - not merely by "direction". Within even a relatively small hall, the sounds reflected from distant reflective surfaces along the walls, ceiling, etc. of the room typically arrive at a listeners ears with delays exceeding several tens to hundreds of milliseconds. These relatively long delayed sounds combine with the direct sounds from the performers, etc. to create a frequency spectrum replete with closely spaced peaks and nulls (with large differences in phase separating them). However, in a typically small to medium-size audiophile listening room, the reflections from room surfaces seldom arrive with a delay exceeding 10 to 20 milliseconds, creating a frequency spectrum comprised of a much smaller number of amplitude peaks and nulls with far greater separations in frequency than those produced in a much larger room or hall. These two quite different spectrums are discerned by the ear as being the result of entirely different size rooms. Thus, unless digital delay circuitry is used to simulate long delayed reflections (using separate loudspeakers to radiate the delayed sounds in the direction of side-walls, etc.), conventional loudspeakers utilizing side or back-firing drivers are simply not capable of accurately simulating or "replicating" a concert hall experience. Jim Wald suggested in his 26 August posting that, "A more thorough debate on this subject would be best left to John Dunlavy and Albert Von Schweikert ..." I would welcome such a debate and would be happy to participate!. Let’s all keep searching for the real truths and technical solutions that will lead us to the more accurate reproduction of sound by electronic means. Best of listening! John Dunlavy From awrigby@aol.com Fri Sep 05 18:06:16 1997 Newsgroups: rec.audio.opinion Subject: Re: Albert Von Schweikert's response to John Dunlavy - Part I From: awrigby@aol.com (AWRigby) Date: 5 Sep 1997 23:06:16 GMT Here is John's reply to Albert Von Schweikert's post: I find Mr. Von Schweikert’s response to me of 4 Sept. 1997 (posted here by Jim Wald) most interesting! Frankly, after reading it I was reminded of a wonderful cartoon by Rodrigues that appeared in Audio Magazine sometime during the 1970’s or 80’s (which hangs in a frame on the wall behind my desk). The cartoon depicts St. Peter at the Golden Gate asking a very humble and contrite looking soul (an ex speaker designer): "So you’re Joseph Paul Carruthers and you were president of Audivex speakers. You made many claims for your speakers. Tell me Joseph, what is ‘flux impulse driver’? And ‘no-lag midrange crossover’? What about ‘vectored ubiquitous maxi-woofer’, what is that Joseph?". Hmmm! Many of the explanations provided by Mr. Von Schweikert to explain how his (and other designer’s) loudspeakers work and perform do not use words, terms and expressions universally used and understood by competent engineers and physicists. As a consequence, Albert might wish to polish his grasp of the fundamentals associated with network theory, array theory, radiation patterns, wave propagation phenomena, radiation properties of common types of musical instruments, etc. It is common practice among competent engineers and physicists to explain the properties and operation of devices (such as loudspeakers, crossover networks, etc.) by using universally understood and accepted words, terms and expressions - thereby minimizing any possible misunderstanding of what meanings are intended during discussions. To meet this need, there is no shortage of words, terms, expressions, etc. within the present vocabulary used by members of the engineering and scientific communities. Indeed, the properties of sound, radiation, arrays of radiating elements, wave propagation, musical instruments, etc. are well-known and have been exhaustively treated within various textbooks, technical papers, treatises, etc.. Likewise, papers on the subject of loudspeakers, radiation, etc. are regularly published within scholarly, peer-reviewed journals such as the Journal of the Audio Engineering Society, the Journal of the Acoustical Society of America, the I.E.E.E. (transactions on Audio), etc. by well-understood theory. Being a regular reader of these journals, I do not ever recall encountering the language, expressions and explanations presently being used and promoted by Mr. Von Schweikert. Further, those who might be tempted to believe that everything known about loudspeaker theory and design can be traced to efforts made during the past decade or two should peruse the book LOUDSPEAKERS, by N.W. Mc Lachlan, D.Sc., published in 1924 by Oxford at the Clarendon Press. (Reading this book can be a humbling experience - for there has been precious little added to basic theory since that time!) It appeals to me that there is no reason (other than questionable marketing motives) for a loudspeaker designer to "invent" such amorphous terms as "Global Axis Integration Network" (GAIN), Acoustic Inverse Replication Theory (AIR), etc. As Albert defines them and applies them to loudspeaker design, they neither make sense (technically) nor do they accurately convey what actually occurs with respect to the manner in which sound waves are radiated, propagated through the air, reflected from room boundaries, etc. And the many pseudo-scientific terms and flooby-dust explanations used by Albert to describe the supposed attributes of his loudspeaker designs are hardly worthy on one wanting to be known as an engineer or competent designer. Assertions that a piano radiates an "omni-directional" or a "spherical" soundwave is absolute nonsense. The sound radiated by a piano over the audio spectrum, be it an upright, grand or a baby grand, is hardly omni-directional or spherical. For example, for notes whose fundamental frequencies are above about 1 kHz, the radiation patterns of a grand piano (with the "lid" open), are quite complex. This is especially true for some of the harmonics (overtones) of the fundamentals. For notes with fundamentals below about 200 Hz, the patterns are broader - although the patterns exhibited at higher harmonics remain complex and multi-lobed. These properties may partially explain why it is so difficult to obtain an accurate recording of a piano. Much the same may be said about the radiation properties of many other musical instruments. Very few exhibit omni-directional or spherical radiation patterns, etc. The term Inverse Replication, I believe, is used by Albert to express the belief that his loudspeakers are capable of "replicating", in an inverse manner, the properties of a recording microphone - with respect to its directivity, frequency response and phase response. Nonsense! The better recording mics (like the ones we regularly use to record our local Colorado Springs Symphony (during the past two seasons) exhibit a nearly perfect omni-directional pickup pattern and are flat within plus/minus 0.2 dB from about 5 Hz to well-above 25 kHz. A loudspeaker capable of truly emulating such a radiation pattern and amplitude/phase response Vs frequency would have to simulate a "pulsating sphere" - hardly anything like Albert’s loudspeakers (or anyone else’s). Hmmm! And then, there is the problem one encounters with Albert’s claim that his loudspeakers, with 24 dB/octave crossovers, exhibit "phase-coherent" performance properties (with excellent impulse and step responses). Not so! Unless Albert has found a means for defying the laws of physics and the teachings of engineering, it is, simply put, not possible. Although a properly-designed 4th-order (24 dB/octave) crossover network provides "in-phase addition" of the high-pass and low-pass signals, at the crossover frequency, they are actually 360 degrees out-of-phase (one full cycle). The only exception to this is an "active digital type of crossover network", where the phase component is not preserved in the "digital-domain". The use of a 24 dB/octave, 4th-order crossover network (with time-aligned drivers) yields a loudspeaker with impulse and step responses that are not minimum-phase and which have very distorted waveshapes - contrary to what Albert says and advertises. (We have accurate MLSSA measurements to prove it!) Albert also drops the names of many well-known designers and engineers he claims to have worked with and learned from in the past - such as Dick Heyser. I knew Dick very well before he died and cannot imagine him ever condoning the technical gobbledygook being exploited by Albert. Dick, by the way, reviewed my first commercially-marketed loudspeaker, the Duntech Labs DL-15, in the August, 1976 issue of Audio Magazine - as the most accurate reproducer of a piano he had ever heard (his most demanding test). Subsequent reviews of DAL loudspeakers in several respected audiophile mags throught the world confirm our claims for accuracy. (See the reviews of DAL’s SC-I, SC-IV and SC-VI loudspeakers in various issues of Stereophile and compare their independent measurements with those we publish.) Last, but not least, DAL publishes a full set of anechoic chamber measurements, with guaranteed accuracy, for every loudspeaker model the company manufactures. And every loudspeaker is individually measured (and stored in files for future reference) in one of our two large anechoic chambers, accompanied by an array of the best and most accurate measurement equipment available. Any purchaser of our loudspeakers can obtain copies of these measurements, free of charge, by merely requesting them. (I wonder if many other manufacturers make such an offer? Does Albert? Hmmm!) I have written the above, not to belittle Albert, his designs or his loudspeakers. There is plenty of room in the marketplace to accommodate the tastes of a wide variety of purchasers. And I believe there is a "place" for loudspeakers that exhibit a "lush, sweet, nice, pretty, engaging, etc." sound character. But such qualities should not be referred to as accurate. I also believe that there are two kinds of accuracy: subjective and objective. Subjective accuracy is relative to the individual "preferences and tastes" of the listener. Objective accuracy, by contrast, must be verifiable by a combination of a complete set of accurate and meaningful measurements, coupled with listening comparisons (in real-time) with live music - as we do during the season with our local symphony and occasional comparisons with a string quartet, piano, etc., here at our plant. In the end, I sincerely believe that most serious music lovers will wish to listen to and purchase loudspeakers and other system components that provide the highest level of objectively verifiable accuracy. For their wish will be to replicate the "original live experience" in their own listening environment - without the acoustics of that environment being superimposed on the accoustics of the environment within which the recording was made! Yes, Albert, we take "true accuracy" very seriously. Lets all work together to make performance specs and advertising claims more honest, complete and meanigful. If we do not, I fear we will lose the long-term confidence and respect of the audiophile community. Best of listening! John Dunlavy From awrigby@aol.com Thu Sep 11 18:26:19 1997 Newsgroups: rec.audio.opinion Subject: Re: Albert Von Schweikert's response to John Dunlavy - Part I From: awrigby@aol.com (AWRigby) Date: 11 Sep 1997 23:26:19 GMT A message from John Dunlavy: On September 5th, I posted a brief initial response to Albert Von Schweikert’s post to me of September 4th (through Jim Wald). However, my schedule last week and early this week did not permit me the time to provide a proper response to all of the issues raised by Albert that I believe need to be addressed. In one of his posts, Albert begins by thanking me for promoting "... this interesting discussion of loudspeaker design" but then refers to the "... interesting but incorrect assertions" I have been making - Hmmm! So far, Albert’s posts have been disconcerting because the technobabble language and explanations he uses are, from a technical point of view, nothing short of a bad joke. Is this my own personal opinion? I think not, based upon the comments posted by several competent engineers and physicists (many of whom are professors at well-known universities). In his posts, Albert has often claimed to have worked with or for the late Dr. Richard Heyser and has implied that his theories are traceable to Dick . Before his untimely death, I knew Dick Heyser very well and spoke with him frequently about loudspeaker performance, design criteria, etc. Dick was always insistent that accurate words and terms be used to describe and explain how and why loudspeakers perform and sound the way they do. Thus, I find it difficult to believe that Dick would condone the gobbledygook and flooby-dust explanations and specifications being used by Von Schweikert. Inventing meaningless "hi-tech" sounding words and expressions to describe loudspeaker performance properties, etc. seems unprofessional. An ample supply of technically-correct words and expressions already exist for explaining the operation and performance of any known type of loudspeaker - old or new. There is simply no justification or reason, other than questionable advertising motives, to use words or expressions with a "pseudo-scientific" flavor to describe any aspect of loudspeaker performance, etc., if the description conveys a false or useless meaning. A statement is "false" if the meaning it conveys is not in conformance with reality and/or what can be proved to be correct. Although such statements are frequently used by some manufacturers merely as a marketing ploy, they nonetheless represent a deception that can eventually erode the trust that audiophiles vest in the industry they support. (I personally believe audiophiles deserve better treatment!) My concern, and perhaps that of many other posters on the NET, is that statements and claims are being made under the guise of representing significant and genuine breakthroughs or advances in the state-of-the-art when they are not. Such claims are particularly onerous when they are made in a manner that attacks, without justification, the integrity of the design approach, features and/or performance of products made by other manufacturers. (Consider that it often requires only a few words to make a false claim but many words to refute it!) For example, Albert has frequently and emphatically claimed that his loudspeakers "exhibit phase-coherent performance using a 4th-order network". Not so, as several posters with excellent technical/professional credentials have explained. Phase-coherent (pulse-coherent) performance, on-axis, cannot be achieved using any type of passive, 4th-order crossover network. Period! Nor can a passive 4th-order crossover yield accurate impulse, step and phase responses. It is a priori impossible! Further, as I have previously stated, many of the descriptive terms invented and touted by Von Schweikert, such as Acoustic Inverse Replication, Global-Axis Integration, Vortex Kevlar Reference Screen, etc. are merely technobabble gobbldeygook devoid of any real-world meaning, with respect to reality and the universal teachings of engineering and physics. Ask any unbiased, competent engineer or physicist. And what Albert attempts to articulate with respect to the desirable properties of cone and dome drivers is equally "off base". The physics related to how cone and dome drivers radiate sound is somewhat complex and, at times, seemingly in conflict with what an untrained person might visualize. It is easy to imagine that a perfectly rigid diaphragm might be the most perfect radiator of acoustical energy. Not so! Such a diaphragm would suffer several problems, not the least of which would be a very poor modulus of amplitude Vs frequency (frequency response), an on-axis beamwidth that becomes increasingly narrower beyond the frequency at which the diameter of the cone/dome exceeds a half-wavelength. And, as the diameter increases beyond a half-wavelength, an increasing number of high-level side-lobes form, each being 180 degrees out-of-phase with the adjoining lobe. A rigid cone would also yield very-poor impulse and step responses (with lots of ringing, etc.) due "energy storage" traceable to the lack of internal damping within the cone material, etc. This knowledge has been well known, as I have previously posted, since the early 1930’s, re: books such as LOUDSPEAKERS, by N.W. McLachlan, published by Oxford at the Clarendon Press in 1934 and "ELECTROACOUSTICS - THE ANALYSIS OF TRANSDUCTION, AND ITS HISTORICAL BACKGROUND", by Fred Hunt, Prof of Physics at Harvard University, published in 1954 and 1982 by the American Institute of Physics for the Acoustical Society of America. These and other similar books provide an excellent treatment of the design elements and properties of cone, dome and membrane drivers. I am not aware that our knowledge and understanding of the "basics" has changed much since that time! Indeed, the essential basics are unchanging - although our understanding of them deepens with time and experience. Albert’s claim that the sound waves emitted by humans and musical instruments are spherical and/or omnidirectional is utter nonsense and without technical merit. For example, the Acoustical Society of America, a professional group hardly given to disseminating false information, has published a number of books (through the aegis of the American Institute of Physics and the Massachusetts Institute of Technology) that treat virtually every aspect of acoustics, acoustical phenomena, measurements, standards, etc. Good examples are the books "ACOUSTICS" (by the well known Leo Beranek) and "HEARING - ITS PSYCHOLOGY AND PHYSIOLOGY" (by Stanley Stevens, Ph.D and Hallowell Davis, M.D., both of Harvard.) In the book ACOUSTICS, one finds in Fig. 11.12 on page 343, a complete set of measured directivity patterns for the human voice, in both horizontal and vertical planes, at frequencies ranging from 175 Hz to 10 kHz. Measurements in the horizontal plane of relative SPL in dB versus horizontal angle, with 0 degrees being the direction of speaker’s mouth, revealed a pattern resembling a cardioid. Relative SPL’s at 180 degrees were -18 dB at 5-10 kHz, -13 dB at 2.5 kHz, -8 dB at 1.2 kHz and -4 dB at 175 Hz. Relative SPL’s at 90 degrees were -8 dB at 5-10 kHz, -6 dB at 2.5 kHz, -3 dB at 1.2 kHz and -1 dB at 175 Hz. Measurements of average voice SPLs directly above the human speaker’s head, i.e., at a vertical angle of 90 degrees, revealed SPL’s of -8 dB at 5-10 kHz, -6 dB at 2.5 kHz, -3 dB at 1.2 kHz and -1 dB at 175 Hz, relative to levels measured at 0 degrees. These values are hardly typical of those claimed by Von Schweikert for the "omni-directional, spherical waves he believes are radiated by most instruments and voices". It is also hard to believe that anyone trained in the disciplines of engineering and physics could believe that human voices and most musical instruments exhibit radiation patterns that might be identified, even roughly, as being spherical or omnidirectional. This especially applies to various horn-type instruments (which are known to exhibit directivity), pianos, most string instruments and various drums (the latter being essentially bi-directional with a pattern resembling a "figure-eight"). The size (area with respect to the wavelength) and shape of the radiating surface, etc. determine the directivity properties of all instruments. This includes, of course, loudspeakers. From these observations, it may be seen that Albert’s theory regarding the need for a loudspeaker to radiate a "spherical wave" (perhaps an omni-directional or spherical radiation pattern) has no basis in reality. His contention that a person can walk around a grand piano and hear no difference in amplitude level or alteration in frequency components only calls into question the acuity of his hearing and or his understanding of what factors determine the radiation patterns of various sound sources, etc. Further, claims that a loudspeaker with a near spherical radiation pattern (that results in reflections from many room surfaces behind and to the rear of the loudspeaker) represents an "ideal design", capable of acoustically emulating the live listening experience within a large concert hall is utterly untrue. The principal reason is that the differential propagation times between the direct and reflected paths, heard by a listener within a large concert hall are typically greater than 100 milliseconds (equal to a wavelength at approximately 110 Hz) while the same differential propagation times within a typical audiophile listening room of modest size would probably not exceed 10 to 20 millisecond (approximately equal to a wavelength at 1100 Hz and 550 Hz, respectively). Thus the spectrum of peaks and partial nulls produced by the vector addition of the direct and reflected sounds would be audibly very different between the concert hall and the listening room. Likewise, Albert’s assertions that DAL is not concerned with the off-axis radiation patterns of its loudspeakers and that we have not measured their patterns in the vertical and horizontal planes is simply not true. (How would he know?) Being a competent antenna engineer, with many fundamental patents and significant accomplishments to establish my credentials in a field where pattern measurements are a "must", why would Albert presume that I would not be equally concerned with the radiation patterns of loudspeakers. Hmmm! (Another of Albert’s non-professional assumptions.) And, although Albert seems not to agree, good reasons exist to believe that "phase coherent/minimum-phase" loudspeaker performance (proven by an accurate impulse and/or step response) is necessary and essential for achieving true, audibly-accurate reproduction of musical transients possessing a complex mixture of fundamental and harmonic frequencies. Evidence of this may be found in the book "Hearing, Its Psychology and Physiology" (mentioned earlier in this post), which states (on page 203), "Not only does the phase of a harmonic that is present in the stimulus have an effect upon the threshold for distortion, but it may also influence the subjective effects of a complex tone. This statement is contrary to the usual assertion that, under Ohm’s auditory law, the ear tends to analyze the components of a complex sound regardless of their phase-relations. Those experiments in which an auxiliary tone was made to beat with an aural harmonic prove definitely that the phase-relations among harmonic components of a stimulus are detectable, for otherwise these beats would not occur. A harmonic in the stimulus may reinforce or cancel an aural harmonic. Under the method of ‘best beats’ we perceive an alternate reinforcement and cancellation due to a constantly changing phase between the auxiliary tone and the aural harmonic. When, however, the auxiliary tone is identical in frequency with the aural harmonic, no beats are heard; but it can be shown that the auditory experience which occurs nevertheless depends upon the phase-relation between the auxiliary tone and the aural harmonic. "There is one phase-relation between these two tones which gives a definite increase in loudness, both of the harmonic and of the total experience; there is another which decreases the loudness. In other words, a given tone, plus another tone of exactly twice the frequency, may sound either louder or less loud than the fundamental alone. Furthermore, the phase yielding maximum loudness differs from that giving minimal loudness by 180 degrees". (Several graphs are presented in the reference to document the aural differences encountered in the presence of different phase shifts between the fundamental tone and its harmonics. Further confirmation of the importance of relative phase upon the audibility of different harmonic relationships is given on pages 171-172, 178,228, 229, 231, etc.) Albert’s apparent belief that a full set of measurements made at a meaningful distance of from 10-12 feet (both on and off-axis), including both amplitude and phase Vs frequency, impulse response, step response, waterfall, energy-time response, non-linear distortion, patterns in both horizontal and vertical planes, impedance Vs frequency, etc., mean very little with respect to how accurate a loudspeaker can reproduce music - especially when compared to the original live musical performance - is simply not true. While measurements are not a certain guarantee of how audibly accurate a loudspeaker can reproduce music - properly interpreted, they provide a flawless guide as to the "potential for a loudspeaker to accurately reproduce complex music". Here, the key operative word is potential. Thus, while a loudspeaker may exhibit near-perfect measured performance in all relevant categories - it still must be compared to live music by competent means - not always an easy or fool-proof task. In this regard, as I have mentioned many times, we have recorded our local 85-piece symphony (two performance per month during their long season), an internationally-known string quartet, individual musical instruments (a piano, violin, etc.), voices, etc. and directly compared the live music with that played back over our loudspeakers. (The recordings were made using a pair of perfectly-matched instrumentation-quality omni-directional mics and a 24-bit DAT recorder.) We used several competent local audiophiles (including several members of the Colorado Audio Society and members of the AES) as subjects for our listening evaluations and real-time comparisons. Believe me when I say that our loudspeaker possess the accuracy required to prevent most competent listeners from discerning any audible difference between the live sound and the recorded sound played back through our loudspeakers. And this includes all six models of our current production line. In this regard, we invite Mr. Von Schweikert to visit our facility during our next live Vs recorded session to ascertain the veracity of these claims. We have nothing to conceal. In a very recent post by David Kersh (apparently VS’s Dir. of Marketing), it appears that VSR is now backpedaling, alluding to their claims for accuracy having merely "... been formulated by our marketing department to create a mental picture of what customers can expect when they hear our products. Complicated math and acoustic theory is difficult for laymen to understand, so we have limited our discussion to general principles rather than mathematical formulae." Hmmm! Does this mean that VS’s recent claims for phase-coherent performance using a 4th-order crossover network are now being re-written - since they were never true? Hmmm! I could go on and on, but why? And by the way, I would put our measurement capability, including our two large 25 ft high X 30 ft long X 20 ft wide true anechoic chambers (each containing more than a full 60 ft trailer load of acoustical foam) up against anything VSR has. Our smaller 10' X 10’ X 8' anechoic chamber is used for testing every single driver before being accepted for use (with the response of each driver being recorded for future use should any ever need replacement in the field. I’ll also bet a king’s ransom that our list of audio test equipment is many times larger and more sophisticated (read accurate) than that of VSR. Any bets? I believe I posted an abbreviated CV on the net recently to provide readers with some information regarding a few of my professional accomplishments, patents, etc. since I began my professional career during January of 1950. If anyone would like a copy of my complete CV, I would be happy to mail one to them. In this regard, I wonder if Albert is willing to make the same offer? In closing, if Albert desires acceptance by accredited members of the audio industry (and informed audiophiles), perhaps he should seriously consider renouncing the use by his company of gobbledygook explanations, pseudo-technical descriptions, false advertising claims, alluding to professional credentials he has yet to confirm, etc., and stop berating the efforts of companies whose products and their performance are the result of competent engineering, application of recognized design approaches, a pursuit of meaningful performance properties and the use of accurate measurements combined with double-blind listening evaluations employing live musical instruments, voices, etc.. The performance of several models of DAL loudspeakers have been reviewed (including measurements, etc.) by Stereophile magazine, perhaps the most reputable audiophile magazine presently being published. I sincerely hope that Albert chooses to re-think his advertising and marketing approach and begins making supportable claims for the performance of his products. If he does so, he will have an opportunity to compete in a marketplace that is becoming ever more sophisticated in its search for truly accurate reproduction that can be verified by competent means. Good luck, Albert - we all want you to succeed for there is plenty of room in the marketplace for better and more accurate loudspeakers, selling at reasonable prices. The demand for such products will only increase while the sales of lesser products decline as audiophiles become more sophisticated, knowledgeable and better educated with respect to what constitutes true audible/measurable accuracy. Lets all work together to make performance specs and advertising claims more honest, complete and meaningful. If we do not, I fear we will collectively lose the long-term confidence and respect of the audiophile community. Best of listening! John Dunlavy From awrigby@aol.com Thu Sep 11 13:56:13 1997 Newsgroups: rec.audio.opinion Subject: Re: VR Marketing Names and the "Pinky" Attack From: awrigby@aol.com (AWRigby) Date: 11 Sep 1997 18:56:13 GMT >Please visit our web site for a picture of a VR-3 without a sock placed >next to one of our assembly line test computers: the picture on the screen >is a global axis polar response graph which takes more time to test than >the simple axial measurements done by Dunlavy. Dave, Since I can't ever remember giving you a tour of our facilities, I find it hard to understand how you have knowledge of the time that it takes to measure a single pair of our loudspeakers. As I have no idea regarding the methods by which VSR tests or measure their loudspeakers, I will not comment on the time that is devoted to such a task. However, so that there is some clarification regarding the matter as it pertains to DAL transducers, I would like to outline our testing procedures: Initially, we individually measure each drive units that we recieve (anechoically of course) in order to determine if the drive units meet our specifications. The initial, individual drive unit anechoic test is performed using the MLSSA software with a B&K 4133 instrumentation microphone. The indiviudal drivers receive a long and short window frequency response test (sampled a minimum of ten times for each window). A hard copy of each test is then printed out and kept with the drivers through the production process. The individual drivers are then matched as pairs and sets depending upon the different models of loudspeakers. The pair and set matching is held to a .2 dB maximum difference. The drivers are installed into the cabinets (which are also accompanined through the production process by a five page quality assurance checklist), and are then put into one of two large anechoic chambers (approximately 30'L x 25'H x 20'W). Both chambers are equipped with Pentium compters that have the MLSSA loudspeaker testing software. (As well, each chamber is equipped with a number of precision analog testing devices such as the HP 3580 Spectrum Analyzer, 239A Oscillator, Multi-Function Synthesizer, Trio Oscilliscope, etc.) Each chamber is equipped with B&K 4133 microphones. The loudspeakers are place upon a lift which has been acoustically deadened and raise to a height of approximately 6' so that any type of floor reflections are nullified/minimized (the entire chamber is deadened with high density, open celled foam). The microphone is set at a distance of approximately 10 feet from the loudspeaker (this gives a better assessment of what the loudspeaker is doing at a normal listening distance, rather than what a nearfield measurement would provide). Each loudspeaker is put through a battery of tests which include frequency response (long and short window), impulse response, step response, impedance plot, waterfall plot, energy time curve, etc. Each test is required to have a minimum of ten samples per individual test (all long window tests are required to have 40 samples per test). When the first loudspeaker has been tested and subsequently modified to the satisfaction of the bench engineer, the second loudspeaker of the pair is inserted into the chamber and the exact same battery of tests are undertaken. The final measurements and all overplots for the pair of loudspeakers are printed out and given to John Dunlavy for his personal evaluation and approval. The loudspeakers that are approved by John are then readied for shipment. The loudspeakers that are not approved by John are inserted back into the cahmber for further testing and modifications until they meet with his approval. These methods insure that every single DAL loudspeaker is guaranteed to not only meet, but exceed our listed performance specifications. After John approves and signs the anechoic test sheets, the loudspeakers are test by analog means in order to insure that nothing was overlooked. Just prior to packaging, each loudspeaker is "buzz tested" and polarity checked. All DAL loudspeakers, from the SC-I to the SC-VI, have their anechoic test data stored in permanant file so that if a customer ever requires a new drive unit, an identical replacement can be found and shipped. Just the anechoic test times (not taking into account the analog and "buzz tests") for the various units are: SC-I: 3 hours SC-II: 4 hours SC-III: 6 hours SC-IV: 6 hours SC-V: 10 hours SC-VI: 14 hours These figures are an average. Some units may take longer, some may take shorter to get them to within our standards. Sincerely, Andrew Rigby DAL From awrigby@aol.com Wed Sep 17 10:51:38 1997 Newsgroups: rec.audio.opinion Subject: Re: Albert Von Schweikert's response to John Dunlavy - Part I From: awrigby@aol.com (AWRigby) Date: 17 Sep 1997 15:51:38 GMT Here is John's latest post: In recent posts, several individuals have characterized DAL’s symmetrical, time/phase coherent array of drivers as being a "D’Appolito Array". This is certainly not true and calls into question whether these individuals have proper understanding of what a D’Appolito array is and how it performs. (It appears that many people are classifying any type of symmetrical driver array as a "D’Appolito" array, which can be an incorrect assumption.) Basically, a D’Appolito Array utilizes a symmetrical arrangement of drivers, i.e., mid-twt-mid, mounted in the same plane along a common vertical axis. As a consequence, the drivers are not time/path aligned along the listening axis. Further, the D’Appolito approach uses a 2nd-order crossover network, with the tweeter connected out-of-phase with respect to the mid drivers, etc.. Although this approach is viewed by some designers as having certain merits, the lack of time/phase-coherent operation yields very poor impulse and step responses in the time-domain, potentially blurring complex/fast musical transients. Because of these deficiencies, DAL believes the disadvantages of the D’Appolito scheme far outweigh its advantages (slightly higher power capability than that attainable with a 1st-order crossover, along with what some believe to be a slightly better vertical radiation pattern). All DAL loudspeakers use a vertically-symmetrical array of drivers, accurately aligned in the time-domain at a distance of 10 feet (on-axis). Along with a 1st order, minimum-phase crossover (compensated in amplitude and phase to correct driver irregularities), this design approach yields true pulse-coherent properties, as exemplified by near perfect impulse and step-responses. Our design methodology also yields very wide, symmetrical radiation patterns in both vertical and horizontal planes, with minimum occurrence of side lobes and nulls. Comments made here on RAO by certain individuals who commented that they were able to discern "big differences in spectral balance" when listening to DAL loudspeakers, between being seated on-axis, standing and moving from side-to-side is simply not true. Nor is the contention that DAL loudspeakers possess a very narrow "sweet-spot", with respect to spectral balance. However, as I have noted in prior posts, any pair of audiophile loudspeakers designed to yield both "pin-point stereo imaging" and a truly accurate reconstruction of the original soundstage, must exhibit a relatively narrow "sweet-spot" with respect to the necessity of a listener being equidistant from both loudspeakers in a stereo system. Other than designing a loudspeaker to exhibit a multi-lobed radiation pattern (perhaps like the "spherical radiation" claimed but not achieved by VSR), with sound energy reflected from many surfaces throughout the listening room, there is no way to cheat the rules ordained by "mother nature". But this kind of "solution", if one can call it a solution, merely yields a blurred soundstage with poorly focused imaging over a wide, so-called "sweet spot"! There is simply no way to change or amend or cheat the laws of physics and acoustics. There is simply no way to change or amend or cheat the laws of physics and acoustics. And anyone claiming they have discovered a way to do so should be able to explain how they managed the feat - without using flooby-dust and gobbledygook language. (After all, it is their design and they should know how it works [in accordance with known scientific principles]). Hmmm! As well, merely commenting that one once knew or worked for this or that person does not necessarily validate an individuals claims with regards to their own company’s products. I believe that identifying a proper background helps in establishing a reference point where consumers and those individuals following this particular thread can immediately identify and correlate comments made by a specific individual with their actual "in field" experience and meaningful, professional-level credentials that they possess, e.g., academic time, membership at an appropriate level in a recognized professional society (AES, IEEE, AIEE, IREE, AAAS, etc.), issued patents, peer-reviewed technical or scientific papers, lectures before professional societies and so forth. If anyone would like a copy of my own CV, a brief request by phone, mail or on the NET will ensure that one is promptly sent by mail. Since Albert (and other members or associates of VSR) began this dialog on the RAO by aggressively critiquing the design and performance properties of our products, while extolling the assumed superiority of his own designs, it seems to me that he should provide documentation in the form of a complete set of objective performance measurements (of guaranteed accuracy) for one or more of his loudspeaker models. Every pair of DAL loudspeakers undergo a complete set of measurements in one of our two large anechoic chambers. These measurements are printed out and are permanently stored for future reference. (The same applies to every driver used in every pair of DAL loudspeakers - so that if a driver ever fails, DAL can provide an exact replacement from stock, thus ensuring unchanged performance over time. This guarantee also extends to all components used in our crossover networks, although we have not been advised of a single failure during the past 5 years.) Copies of the final set of performance measurements for each pair of DAL loudspeakers are available to any current owner of Dunlavy Audio Labs Signature Collection loudspeakers. In summation, while DAL firmly believes that a complete set of competent measurements, properly interpreted, can be used to assess the "potential" a given loudspeaker possesses to achieve truly accurate reproduction of complex musical transients, inner detail, etc., the final test must always be a real-time comparison with live music. Being the company who has professionally recorded all sessions (but one during the last WCES) of the Colorado Springs Symphony during the past two years (for re-broadcast over our local classical and jazz FM station), we have repeatedly had the opportunity to perform live Vs recorded comparisons. Using instrumentation-quality omni mics (with a ruler-flat amplitude and perfect impulse/step responses) and a 24-bit DAT machine, our recordings are about as accurate as can be achieved with present state-of-the-art equipment. On many occasions, we have had skeptical magazine editors, engineers, and others present during the recording sessions. After witnessing the comparisons, their skepticism regarding our claims vanished. There simply were no consistently audible differences between the live orchestra and its reproduction over our loudspeakers. Indeed, the differences were so small that it typically required several minutes to discern any differences whatsoever. We anxiously await VSR’s promised second article so that we may continue this debate in hopes that audiophiles may become more knowledgeable with regards to stated performance claims and design concepts employed by a variety of high end audio manufacturers. As such, when and if they decide to make a product purchase, they may do so with full knowledge of a specific products design and related performance. Best Listening, John Dunlavy Subject: JOHN DUNLAVY RESPONDS TO ALBERT VON SCHWEIKERT From: awrigby@aol.com (AWRigby) Date: 1997/09/27 Message-Id: <19970927003301.UAA10836@ladder02.news.aol.com> Newsgroups: rec.audio.opinion As many readers will know, Albert Von Schweikert has finally appeared on R.A.O., via a posting by Jim Wald. It is a rather lengthy posting that appears intended to “set me straight” with respect to a number of matters dealing with engineering, design, performance, measurements, accuracy, specifications and so forth. Hmmm! It is tempting to simply dismiss Albert’s quasi-technical rhetoric as the writings of someone who lacks a full and proper grasp of well-known principles taught by engineering and physics that are relevant to the present discussion. So - - - where to begin? Hmmm! Lets start at the beginning. Albert states: “John, before I get to answering your technical challenges, I want to discuss the general tone and attitude you seem to have toward everything about me and what I have accomplished. This has been not only distracting from the real issues, but detracts from the value of anything either of us might have to say about our field of endeavor. Now, I think that everyone would agree that everything about your attitude about my work, as reflected in your posts, tell us in no uncertain terms that you feel that you are the leading authority on speakers and measurements, and you obviously believe that your approach is the only possible, not to mention acceptable, way of designing a “correct” speaker system.” OK, lets analyze what Albert has said! To begin, I paid little attention to what Albert and his crew posted on the NET until he and/or his marketing personnel (and some of his devotees) began deriding the design and performance features of DAL loudspeakers, claiming that they did not perform as we advertised and that VSR loudspeakers were far more accurate reproducers of complex musical waveforms. Specifically, VSR representatives (presumably under the aegis of AVS) claimed that: 1) DAL did not measure the radiation patterns of its speakers. Not true! (detailed refutation follows) 2) The radiation patterns of DAL loudspeakers are narrow and cannot re-create an accurate soundstage like VSR loudspeakers, which exhibit spherical radiation patterns that emulate live musical instruments.. Not true! (detailed refutation follows) 3) The “paper” cone drivers used by DAL create large amounts of distortion not present in the carbon-fiber cones used by VSR. Not true! (detailed refutation follows) 4) DAL’s minimum-phase crossover networks create high levels of distortion that VSR’s higher-order crossovers do not. Not true! (detailed refutation follows) 5) VSR loudspeakers are time-aligned and phase-coherent just like DAL’s. Not true! (detailed refutation follows) 6) While VSR loudspeakers are internally-wired with silver wire, DAL loudspeakers use inferior copper. 7) DAL’s insistence that a loudspeaker should exhibit good impulse response and step response to be considered accurate is not correct. Not true! (detailed refutation follows) 8) VSR loudspeakers exhibit “Acoustic Inverse Replication” properties while DAL loudspeakers do not, making VSR’s products more accurate. Not true! (detailed refutation follows) 9) Dunlavy’s postings exhibit a “holier than thou” attitude. 10) What Dunlavy won’t tell you is that the human ear CAN NOT distinguish the 0.5 mS delay introduced by VSR’s 4th-order crossover, making the delay irrelevant. 11) A paper by Erik Baekgaard in the May 1977 issue of the Journal of the AES provides evidence to prove that higher-order networks can provide phase-coherent performance. 12) Albert’s latest post to me stated, “It should amuse you then to know that one of the greatest influences (Dick) Heyser had on my thinking was that he distrusted anyone who believed that measurements could totally portray a product’s sound quality as you would have us believe.” Lets now take each of the above numbered assertions by VSR and/or AVS and attempt to show where they are incorrect. 1) DAL has two excellent, large anechoic chambers (24’ X 20’ X 16’) for performing loudspeaker measurements, including pattern measurements in both horizontal and vertical planes. A large rotating platform, with ball-bearing support, capable of supporting several hundred pounds, is used to measure patterns in both planes with a high degree of accuracy. (Albert, do you have a similar setup for measuring patterns and would you supply them upon request? DAL does!) 2) DAL performs complete pattern measurements of each of its loudspeaker models during their design phase to ensure that they provide optimum angular dispersion with respect to room reflections and their interaction with the direct sound heard at the listening location. 3) DAL regularly receives samples of the latest drivers from virtually all of the best-known producers of audiophile loudspeaker drivers, crossover components, etc. Our selection of drivers and components is always based upon measurements and audible performance - never on the basis of “high-tech appearance” and or price. In this regard, we have consistently found that cones fabricated of Kevlar and carbon fibers did not yield the most accurate impulse response, step response, modulus of amplitude Vs frequency, low levels of non-linear distortion, wide angular dispersion, etc. The drivers chosen by DAL for its loudspeakers are the best and most accurate available. And no, Albert, they are not made of “paper”. Rather, their cone material is a proprietary composite of fibers, fillers and various plastic materials chosen to provide optimum damping Vs frequency, most accurate impulse/step response, minimum modeing and the widest angular dispersion (radiation). 4) DAL pays considerable attention to the production of non-linear distortion by its loudspeakers. We have detailed, accurate measurements to prove that DAL’s loudspeakers exhibit levels of non-linear distortion that are inaudible at any reasonable listening level at a distance of 10 feet. (We supply copies of these measurements to purchasers who request them.) 5) VSR loudspeakers are neither accurately time-aligned nor phase-coherent as claimed by Albert (or perhaps his marketing department). We have accurate measurements of his loudspeakers to prove they are not phase-coherent. Indeed, it is not possible to achieve phase-coherent reproduction, on axis, using a high-order crossover network. (See further comments regarding this assertion later in this posting.) AVS states that the time-delay thru a 4th-order network is less than about 0.5 milliseconds. The actual delay may be less or considerably more, depending upon the frequency of each section of the crossover. 6) OFC Copper (99.9999% pure) is certainly not inferior to silver for the internal wiring of a loudspeaker. In fact, copper is in many ways superior to silver for this application. To state that silver is audibly or measurably superior for the short runs within a loudspeaker is pure nonsense - ask any competent engineer, etc. 7) DAL insists that all of its loudspeakers exhibit excellent impulse and step responses because it makes good sense. Would anyone consider purchasing an amplifier that exhibited a lousy impulse or step response? I doubt it! Since linear systems theory teaches that the “end result” (measurable and audible) is the same regardless of where a performance aberration is produced within a series linear system, why permit it to occur in a loudspeaker but not in an amplifier. Hmmm! 8) Acoustic Inverse Replication (AIR) is an invention of Albert and/or his marketing staff. It is a marketing ploy that has no basis in engineering or the sciences. It is, simply put, “total nonsense” devoid of credibility. It is hardly worth the time to explain why - although some qualified engineers and or physicists might wish to take the time to do so if interest persists. 9) Hmmm! I thought they merely attempted to state provable, well-known “facts” taught by competent engineering and physics.(detailed refutation follows) 10) I believe that DAL’s research, soon to be published in a reputable technical journal will prove otherwise. 11) The paper by Erik Baekaard in the May 1977 issue of the Journal of the Audio Engineering Society begins by mentioning that, “Hansen and Madsen have shown that one form of distortion that is audible is phase distortion, and that it affects transient performance. Thus it is not sufficient, as previously believed, to be content with a flat amplitude-frequency characteristic and low harmonic and intermodulation distortion ...” Hmmm! Baekaard then proceeds to show examples of crossover networks feeding a three-way system (bass, mid and tweeter drivers), wherein the bass and tweeter are fed signals through 2nd, 3rd or 4th -order low-pass and high-pass networks. BUT, the mid-range driver (referred to as a “filler driver” by the author) is fed by a 1st-order network. The sum of the on-axis radiation from this combination of drivers, if they are time-aligned, results in a phase-coherent system capable of accurately reproducing impulses, steps and square waves. But this is apparently not what Albert’s VSR loudspeakers and their crossover networks do - based upon measurements of their on-axis reproduction of complex waveforms (which is very poor, to say the least). But the system suggested by Baekaard does not really solve the system power-handling problem as suggested by AVS because it still requires the use of a 1st-order network feeding the mid-range “filler-driver”. Thus, the “system” power-handling capability of such a loudspeaker is limited by the power-handling of the filler driver, fed by a 1st-order crossover network. Can you explain this otherwise, Albert? Of course, it appears that Albert has avoided the use of a “filler-driver” and feeds each of the drivers in his loudspeakers via higher-order crossover networks to maximize power-handling capability - at the expense of very poor reproduction of musical transients, squarewaves, etc., which he seems to feel is inaudible and inconsequential. Hmmm!. 12) Wow! Another mis-quote from Albert. What I have repeatedly said is that “a full set of competent measurements can be used to assess the potential a loudspeaker possesses to accurately reproduce complex musical transients, etc.” It has always been my belief, as I have clearly stated numerous times, that “neither measurements nor listening, taken alone can determine the accuracy of a loudspeaker - both are needed, interpreted interactively.” Please, Albert, read my posts more carefully and do not attempt to further misquote what I have said. Further, I knew Dick Heyser very well and communicated with him frequently before his untimely death. Believe me, Albert, Dick Heyser was a firm believer in the use of proper measurements for “determining the potential of any loudspeaker to yield audibly accurate reproduction”. And I completely affirm Dick’s belief in the necessity of using a full set of accurate measurements as the first step in evaluating loudspeaker performance. (I suggest that Albert re-read some of Dick’s loudspeaker reviews in Audio Magazine during the 1970’s and some of his papers in the Journal of the AES. Dick, by the way, reviewed my first commercially-marketed loudspeaker, the Duntech Labs DL-15, in the August 1976 issue of Audio Magazine - as the most accurate reproducer of a piano he had ever heard. (The reproduction of a piano was his most demanding test for determining the audible accuracy of loudspeakers.) I have written the above, not to belittle Albert, his designs or his loudspeakers. There is plenty of room in the marketplace to accommodate the tastes of a wide variety of purchasers. And I believe there is a “place” for loudspeakers that exhibit a “lush, sweet, nice, pretty, engaging, etc.” sound character. But such qualities should not be referred to as accurate. I also believe that there are two kinds of accuracy: subjective and objective. Subjective accuracy is relative to the individual “preferences and tastes” of the listener. Objective accuracy, by contrast, must be verifiable by a combination of a complete set of accurate and meaningful measurements, coupled with listening comparisons (in real-time) with live music - as we do during the season with our local symphony and occasional comparisons with a string quartet, piano, etc., here at our plant. In the end, I sincerely believe that most serious music lovers will vote for and purchase loudspeakers and other system components that provide the highest level of objectively verifiable accuracy. For their wish will be to replicate the “original live experience” in their own listening environment - without the acoustics of that environment being superimposed on the acoustics of the environment within which the recording was made! Yes, Albert, we at DAL take “true accuracy” very seriously. Lets all work together to make performance specs and advertising claims more honest, complete and meaningful. If we do not, I fear we will lose the long-term confidence and respect of the audiophile community. Best of listening! John Dunlavy --from the bass list-- Recently, a few individuals have raised the subject of “planar loudspeakers” and why I don’t design and manufacture them instead of loudspeakers using cone and dome drivers. Their belief has been that planar speakers are more accurate because of their very low-mass diaphragms and large radiating area (which some believe produces wider angular dispersion). Some also expressed the opinion that the bi-polar or bi-directional radiation of planar loudspeakers provides a more accurate “coupling” of sound energy into the listening room. Because so many misconceptions appear to exist with respect to relevant performance features of both membrane and cone/dome drivers, I would like to share some of my thoughts on the subject with those of the BASS group who might be interested. Lets begin by examining some of the basic physics involved in the radiation of sound waves from all types of diaphragms: cones, domes and planar types. A driver with a circular-shaped cone, suspended at its outer edges by a compliant “surround” and driven at or near its apex by a low-mass voice coil symmetrically immersed in a strong magnetic field, is probably the most accurate type of transducer presently available. The important caveat here, though, is whether it was properly designed, e.g., with: 1) a cone made of materials possessing the right acoustical properties (rigidity, internal-damping, etc.), 2) an outer “surround” with the correct compliance, damping, etc., 3) a voice-coil and magnet structure having an optimum “BL product” (the product of the magnetic flux and length of wire), 4) a spider assembly which properly centers the voice-coil and exhibits correct compliance, damping, etc. and, 5) a frame or basket that is rigid and anti-resonant. Of course, the driver must also exhibit the desired frequency response, impulse response, efficiency, impedance (resistance and reactance Vs frequency), resonant frequency, Qt, Qe, Qm, etc. It is interesting to note that elliptical-shaped cones, popular until the late 1970’s, provided an opportunity to achieve a “flatter” frequency response and more useful angular dispersion (wider horizontally and narrower vertically) than that available from a cone having a circular cross-section. Contrary to popular opinion, a dome shaped radiating surface does not provide as wide an operating bandwidth or as wide a beamwidth as a well-designed cone driver having the same diameter. This is partly because the dome is driven from its outer edge, thereby defining the diameter of the radiating area at all frequencies within its operating range. Also, a well-damped dome with a diameter larger than about 2-3 wavelengths at its intended high-frequency limit, e.g. a 3” dome at 10 kHz, usually radiates relatively little energy from its center region - radiation mainly being confined to an annular region surrounding and adjacent to the voice coil. Thus, a dome tends to exhibit the beamwidth properties of a annular (ring-shaped) radiating surface possessing a “constant diameter”, with a center region that radiates less energy with increasing frequency. Since the half-power (-3dB) beamwidth (in degrees) of a radiating source is approximately equal to 58/D (where D is the “effective diameter” of the radiating surface, expressed in terms of a wavelength at the frequency being evaluated, e.g., approx. 13,600/freq. in Hertz), a 3 inch dome would exhibit a beamwidth of less than about 26 degrees at 10 kHz. It is interesting to note that an annular-shaped radiating area also tends to exhibit high-level “side lobes” that alternate in phase relative to the main lobe, compared to a more uniformly-illuminated radiating surface. By contrast, a properly damped, cone-shaped diaphragm, driven by a voice coil located near its apex, experiences a more constant beamwidth over a much wider bandwidth than a dome. This is because properly engineered cone material exhibits “internal damping properties”, intended to progressively absorb more energy at higher frequencies. This confines much of it to regions closer to the voice coil. Thus, while a 3 inch diameter cone type mid-range driver may have an effective radiating diameter of about 2.5 inches at 1 kHz, this probably reduces to less than 1 inch at 10 kHz, yielding a half-power beamwidth of about 80 degrees, nearly 3 times that of a 3 inch dome. These relative beamwidths, between a 3 inch dome and a 3 inch cone, are only approximate and do not take into account the reduced “velocity-of-propagation” which occurs along or through a “lossy medium”, such as a damped diaphragm. A well-designed cone driver, with optimum damping and flat frequency response, can exhibit an excellent impulse response, with the first “overshoot” approximately minus 15 dB (without crossover). Further “ringing” is typically down more than 20 dB and persists for no more than 100 to 200 microseconds. (A well-designed crossover network can often improve on these values of overshoot and ringing.) Although well-designed dome drivers can also achieve a good impulse-response, they usually exhibit moderately more overshoot and ringing than a well-designed cone driver of the same size and efficiency. As can be seen, cone and dome drivers typically perform quite differently than most audiophiles believe. While a dome might appear more “hi-tech” than a cone, it has many performance limitations and, for many applications, is inferior (overall) to a cone driver of equal “design quality”. As a consequence, domes tend to perform best for tweeter applications, where the diameter of 1 inch dome corresponds to about 1.5 wavelengths at 20 kHz (yielding a half-power beamwidth of about 38 degrees). An “inverted dome” is generally inferior in performance because the inverted shape forms a “cup-shaped” cavity that can exhibit a resonant property if the depth approaches 1/4 wavelength within the operating range (about 5/32 inch at 20 kHz). “Jazzy-looking” cone and dome materials (often made from yellow-colored, “bullet-proof” materials, such as Kevlar) usually exhibit poor internal damping properties, resulting in an impulse response characterized by considerable overshoot and subsequent ringing. Kevlar, and similar materials, while providing rigidity, lack proper internal-damping properties required for good impulse response and truly flat frequency/phase response. As a consequence, Kevlar cones (and domes) are generally confined to applications in loudspeaker designs using higher-order crossover responses, where impulse, step and phase responses are not considered important by the designer. Cones and domes made of metal (such as titanium) or ceramic are even worse because these materials provide virtually no internal damping properties, resulting in poor impulse response, poor step response, etc. While the cones with the best measurable and audible performance may appear to be made of a dark-gray colored “paper” material, they are most likely a complex formulation of felt, cellulose/carbon fibers, a binder, and a coating that provides optimum “damping”, minimum formation of undesirable modes at higher frequencies, excellent impulse/step response and very flat frequency/phase response, etc. Most audiophile-quality cone and dome drivers exhibit a Sound Pressure Level (SPL) of about 88-90 dB, at an on-axis distance of 1 meter, for an input of 1 watt (2.83 volts RMS across 8 ohms). This is many times the efficiency of typical planar diaphragm radiators, having the same radiating area. However, cone and dome drivers with SPL’s higher than about 91 dB (re: 1 watt) usually possess a Qts (total Q factor) that yields poor damping and less than good impulse and step responses. Lets now switch to loudspeakers with “planar diaphragms” (flat plastic membranes) and examine how they work and perform, compared to drivers with cones or domes. To begin, there are typically two classes of loudspeakers or loudspeaker drivers based upon planar diaphragms or membranes: 1) electrostatic types, and 2) magnetic types. An electrostatic type usually employs a thin, light weight, plastic film diaphragm (possessing low-loss dielectric properties), stretched mid-way between two parallel “wire grids or metallic mesh”. A D.C. polarizing voltage, typically several thousand volts, is connected between the diaphragm and both grids. The audio input signal is fed to the two grids, out-of-phase, typically through a step-up transformer. The magnetic type of planar loudspeaker usually makes use of a thin plastic diaphragm with an electrically-conductive coating of parallel wires (other configurations are possible and may be used). This diaphragm is tightly stretched between an array of small magnets that are configured adjacent to and on either side of the parallel wires. When a signal current passes through the wire grid, it is attracted to one array of magnets and repelled by the magnets on the opposite side, creating sound. A signal current flowing in the opposite direction causes the diaphragm to move in the other direction. (Early models made by one manufacturer used magnets on one side only, resulting in high levels of even-order harmonic distortion.) Electrostatic and magnetic types of planar loudspeakers have been designed with several different configurations. Some use a single planar diaphragm with a large area (several square feet) that radiates all frequencies. This, of course, results in undesirably narrow beamwidths, at high frequencies, in one or more planes. An improved variation of the single diaphragm uses a narrow width with a large height, which increases the horizontal beamwidth at the expense of decreasing the vertical beamwidth (which might reduce some undesirable multipath effects attributable to floor and ceiling reflections. A second version uses a wide, tall panel to radiate bass and lower mid frequencies, while a tall but very slender strip is used to radiate higher frequencies. Other variations, such as arrays of small, individual panels, have been designed to satisfy certain performance criteria deemed important by the designer. So! What about the popularly held belief that most membrane type loudspeakers exhibit properties that are measurably and audibly superior to loudspeakers using cone and dome type drivers? The most universally held convictions are 1) that a super-light radiating surface or membrane can accelerate much faster than a cone or dome, thereby providing superior impulse response and more accurate reproduction of complex musical sounds, 2) that the much larger surface area of a planar membrane radiates a much broader beam of acoustical energy, 3) planar membrane type loudspeakers exhibit a very flat frequency response, 4) the bi-directional radiation pattern of planar loudspeakers provides a more natural and realistic sound in most listening rooms. Lets examine each of these assumptions to determine if any are valid. 1) The ability to rapidly accelerate any type of diaphragm, planar, cone or dome. is dependent mainly upon two parameters: the moving mass (or weight) of the radiating element and the total “force” (electric or magnetic) acting upon the full surface of the diaphragm. The simple expression from basic physics, F=MA (or A=F/M), tells it all. From this expression, it can be seen that, although the mass M of a planar diaphragm is very small, the forces acting on the mass are even smaller - compared to an ordinary cone or dome type of driver with a large magnet and a voice coil with a large number of turns (a large “BL Product”). The larger BL Product of cone and dome drivers provides a 6-10 dB advantage and is the reason why membrane type loudspeakers require a large “radiating area” to obtain reasonable efficiency and satisfactory sound levels. The lower “efficiency versus radiating area” of membrane loudspeakers frequently equates to a relatively poor “damping factor”, due mainly to the lack of any “restoration force” needed to “restore” the membrane to its original position, after the input signal drops to near-zero at the end of a transient. Further, all membranes must somehow be mechanically attached to a rigid frame. Thus, when the membrane is set in forward motion by a signal, it’s surface becomes curved because the edges adjacent to the frame are not free to move. As a consequence, sound energy is reflected from the frame back toward the center of the membrane. This reflected energy combines with the original “incident” energy to form “standing-waves”, with maximums and minimums spread across the membrane at intervals of one-half wavelength at each frequency of excitation. This is, perhaps, one of the reasons that many manufacturers have chosen to build their membrane loudspeakers as an array of separate modules of smaller size, with resonances and standing-waves that are easier to control. It is also possible that some manufacturers have discovered unique means for acoustically damping the edges of the diaphragm where it is attached to the frame so as to absorb sound energy and reduce reflections that might create standing-waves. 2) Because of the much larger radiating area required by membrane type loudspeakers to reach acceptable levels of efficiency, a design using a single, large membrane (to radiate all frequencies) exhibits a beamwidth that approximately drops in half for each octave above the frequency where the membrane measures approximately one-wavelength in the plane being evaluated. This assumes, of course, that the listener is located at a distance that is greater than about three times that of the largest dimension of the membrane (commonly referred to as the “far-field” distance). For a membrane 4 to 5 feet high, this would equate to a listening distance of about 12 to 15 feet. Peter Walker, a highly competent engineer and designer of the QUAD ESL-63 Electrostatic loudspeaker, employed an ingenious series of concentric, annular diaphragm sections surrounding a circular center section. The center section radiates all frequencies, while successively larger diameter annular sections are fed signals thorough a series of low-pass networks that roll-off higher frequencies as a function of the radius outward from the center. This configuration yields “on-axis beamwidths” that decrease only slowly with increasing frequency. However, at frequencies above about 5 kHz, its beamwidth is still much narrower than that of a well-designed loudspeaker using conventional cone and dome drivers. The relatively narrow beamwidths exhibited by membrane loudspeakers can significantly alter the spectral balance for a listener seated off-axis by more than a few feet. Such narrow beamwidths can also alter the ratio of the on-axis response to the integrated room-response, causing the speaker to sound “heavy” in the upper-bass and lower mid-range regions. The half-power (-3dB) beamwidth (in degrees) of a uniformly-illuminated planar radiating surface is, as mentioned in par. 5 (above), approximately equal to 58 divided by the dimension of the surface (in the plane being evaluated), expressed in wavelengths. One might argue that membrane loudspeakers using a narrow (but tall) ribbon for the tweeter range solves this problem of limited angular dispersion in the horizontal axis. And this is certainly true, but does not take into account that, while it significantly improves horizontal dispersion, radiation in the vertical plane remains very narrow - literally only a few degrees above about 7 kHz. 3) While Peter Walker’s ESL-63 exhibited a very flat, on-axis, amplitude versus frequency curve (nearly plus/minus 2 dB), most membrane loudspeakers barely meet a plus/minus 4 dB spec, on-axis, at a distance of 10 to 12 feet. In reality, however, this is representative of most hi-end audiophile loudspeakers, for very very few even approach their advertised specifications. (Sadly, as many wise audiophiles have concluded, “measurements don’t lie but measurers often do!”) 4) Although more psycho-acoustic research is needed (and the jury is still awaiting additional evidence), there are good reasons to conclude that bi-directional (or dipole) radiation, while it may produce great sounding music and a wide sound-stage that pleases many devotees - is not truly accurate when compared to the original live performance. Nor does it provide a true “pin-point” center image (for well-recorded center vocals) in most rooms. The inability of bi-directional radiation to accurately emulate the original performance may partially be explained by the fact that many (if not most) musical instruments radiate their sound with a directivity pattern that is essentially unidirectional, at least at mid and high frequencies, where the relevant dimensions of the instrument become large with respect to a wavelength. Obvious exceptions, of course, are the organ, drums and string bass. Another deficit of a bi-directional radiation pattern is that low frequency sound refracts (bends) around the edges of the outer frame where it combines, out-of-phase, with the sound radiated from the opposite side of the membrane. This tends to seriously limit the low-frequency response, usually calling for use of a sub-woofer. But this, in turn, creates a time/phase alignment problem between most sub-woofers and membrane loudspeakers, resulting in sharp “drop-outs” in the frequency response within about one-octave of the crossover frequency. I won’t mention brands, but several manufacturers have tried to widen the relatively narrow horizontal dispersion angle by curving the surface of the radiating panel so that it appears convex, as seen from the listening position. Sounds like a simple means for achieving a worthwhile end! But Mother Nature is often a hard-headed “Task-Mistress” that seldom provides any free lunches when it comes to trade-offs. And this certainly appears to be the case with respect to curving the surface of a membrane loudspeaker. Intuitively, one would think that curving the surface would simply widen the radiation pattern in the plane of curvature, without incurring any undesirable consequences. Unfortunately, the price to be paid is destructive wave interference at frequencies where the depth of the curvature is roughly equal to a half-wavelength (and its multiples). This shows up very clearly in the impulse response and step-response of a curved membrane radiating panel. It is also revealed in it’s frequency-response curve, derived from the impulse-response by FFT. This applies even to curved membrane loudspeakers costing big bucks! For example, one very expensive curved membrane model that we measured (at 10 feet) exhibited an on-axis frequency response with plus/minus 5 dB undulations, beginning with a deep drop-out at about 3 kHz, followed by additional drop-outs at 6, 9 and 12 kHz, with “healthy 3-4 dB peaks” in between the nulls. These nulls and peaks persisted off-axis, with little change in the overall response curve up to about 45 degrees, except for a steeper high-end roll-off above about 10 kHz. If there is a solution, without suffering serious degradation of other important performance attributes, I have neither discovered it nor measured any loudspeaker that has successfully achieved the goal. If anyone has found or knows of a means for curving a membrane without “paying a price”, please speak out and set me straight. Best wishes. John Dunlavy From 102365.2026@CompuServe.COM Wed Oct 29 12:26:10 1997 Newsgroups: rec.audio.opinion Subject: Re: Question for Ken Kantor & John Dunlavy From: Andrew Rigby <102365.2026@CompuServe.COM> Date: Wed, 29 Oct 1997 13:26:10 -0500 With respect to the questions and answers concerning the break-in time for loudspeakers and their drive units, I generally agree with the opinions expressed by Ken Kantor. Most loudspeakers and their drivers require little or no break-in time to achieve their level of long term performance. Indeed, 10 minutes of break-in time using music containing a wide range of tones, especially in the low bass region of the audio spectrum should be quite satisfactory. However, not all woofers, mids, and tweeters are created by the same designer using the same materials, etc. Some woofers with a relatively stiff suspension may require more break-in time than others. Tweeters and mids seldom (if ever) exhibit any measurable or audible change in properties with use, despite contrary claims by the "golden ear" set. As well, adequate break-in usually occurs during the time required for the rigorous testing and measurement procedures given loudspeakers by manufacturers concerned with insuring the accuracy and consistency of their products. Best of listening! John Dunlavy Dunlavy Audio Labs From 102365.2026@compuserve.com Thu Nov 06 09:59:38 1997 Newsgroups: rec.audio.high-end Subject: Dunlavy responds to Waveform From: John Dunlavy <102365.2026@compuserve.com> Date: 6 Nov 1997 10:59:38 -0500 During the past few weeks, a number of postings have contained comments regarding the best crossover slope, alignment of drivers, driver array configuration, etc. for obtaining the most accurate/realistic reproduction from loudspeakers. One of the postings on Sept. 4th,, a rather long and impassioned one from John Otvos (J.O.) of Waveform, somehow missed my attention and has gone unanswered - until now! Many of his comments are interesting and deserve an informed reply. In addition, several others have expressed opinions that call into question, "what is true accuracy and how can it be determined?" One simple answer might be "the precise duplication of the original sound by the reproduced sound". But such a definition seems too general and does not specifically address such questions as under what listening conditions, what types of sound, etc. Thus, it seems to me that any reasonable definition of "accuracy" requires the use of "standards", both measurable and audible - intended to avoid, as much as possible, the slippery slopes of personal opinion. For example, Webster's Collegiate Dictionary defines the adjective "accurate" as: conforming exactly to truth or to a standard. And, Roget's International Thesaurus groups the noun "accuracy" with "exactness, preciseness, precision, faithfulness", etc. Indeed, without a proper definition and a standard by which to judge or measure it, accuracy becomes an amorphous term conveying little relevant or acceptable meaning. This seems especially true with respect to the many audiophile loudspeakers that claim to be "accurate", yet sound and measure quite different from each another. Although several papers on the subject have appeared in such peer-reviewed journals as that published by the AES, etc., standards for accuracy have yet to be defined and published for industry use. Until such standards are accepted and used, it is unlikely that the term "accurate", as applied to loudspeaker reproduction, will ever gain and sustain respectability among audiophiles and professional users. Sadly, many manufacturers seem to believe that the design of a new loudspeaker model should begin with the choice of an "enclosure" having an attractive "high tech" appearance, an appealing shape, a beautiful wood veneer and a furniture-quality finish. While most deem "appealing sound quality" important for maximizing sales, few address the subject of "true accuracy". Indeed, most loudspeaker "designers" are merely cabinet makers that possess no technical/academic credentials or professional experience. As a consequence, many audiophiles refer to loudspeaker manufacturers as "box stuffers" - a sad commentary on an industry tasked with designing what is probably the most critical link in the reproduction chain, with respect to defining the overall accuracy of an audiophile system. I would suggest that "true accuracy", from the perspective of a serious audiophile, exists when the sound being reproduced within a typical home listening room is virtually indistinguishable from the original live sound/music - as it was heard and recorded by a competent recording engineer. Accurate reproduction in typical audiophile listening rooms implies the use of loudspeakers having well-controlled directivity/dispersion properties in both horizontal and vertical planes. Otherwise, loudspeakers with wide, omni or bi-directional radiation properties will tend to excite room resonances or reflect undesirable amounts of sound energy from the walls, floor and ceiling. If these reflected components arrive at a listener's ears within less than about 5 to 15 milliseconds after the direct sounds radiated by the loudspeaker, it is highly likely that an audible "comb-filter spectrum" will be created, with widely-spaced peaks and nulls, resulting in the blurring of complex musical transients and/or shifts in spectral balance, especially at bass and lower-mid frequencies. While such time-domain and/or spectral alterations of the original sound may be interpreted by some listeners as adding a "lushness, sweetness, fullness, musicality, etc.", the result can hardly be described as being accurate and true to the original sound. By contrast, the sound reflections within a concert hall or reasonably large room arrive at the ears of listeners with typical time delays exceeding tens to hundreds of times those experienced within a typical audiophile listening room. These much longer time-domain delays create a comb-filter spectrum with a very large number of closely-spaced maximums and minimums in amplitude across most of the audible frequency-domain. This type of spectrum is usually interpreted by our "ear/brain processing" as merely one of the artifacts indigenous to a large room or hall - frequently adding a desirable level of ambiance. But "time-domain" distortion, whether it results from obtrusive room reflections, improper time-alignment of drivers along the listening axis, the use of a high-order crossover network (e.g., 2nd or higher), enclosure edge diffraction, drivers that exhibit poor transient response/ringing properties, etc., should always be of concern when loudspeakers are intended to yield "accurate reproduction". Likewise, loudspeakers with broad radiation patterns used within what might be described as an "overdamped" listening room (having a large number of absorptive foam panels, etc.), often yields an objectionable imbalance between the high/mid frequencies and bass frequencies. This may be traced to the much lower absorption efficiency of most materials (rugs, open-cell acoustical foam panels, etc.) at frequencies below about 500 Hertz. So, how do these observations relate to the comments posted to me by J.O.? Well, J.O. seems to believe that loudspeakers using 4th-order crossover networks and an asymmetrical vertical array of drivers (w-m-t) achieve more accurate reproduction of music than loudspeakers employing a 1st-order crossover with a symmetrical array of drivers (w-m-t-m-w). Not so! As mentioned earlier, a wide radiation pattern in the vertical plane will considerably enhance the amplitude of reflections from the floor (and sometimes the ceiling) of the listening room, creating blurring of musical transients and poor low-frequency response at the listening location. For example, a typical floor reflection, with a delay of about 2 milliseconds (relative to the direct sound heard at a listening distance of 10 feet), will result in a quasi comb-filter spectrum, beginning with a broad maximum below about 125 Hz, a minimum at about 250 Hz, another maximum at about 500 Hz, etc. (probably extending out to frequencies above a few kHz). This condition is exacerbated when using an asymmetrical vertical array of drivers, consisting of a woofer near the floor, a mid-range driver above the woofer, and a tweeter above the mid-range driver. Although some less critical listeners might interpret such reproduction as providing "increased ambiance", a "fuller sound", etc., it certainly does not qualify to be called "more accurate". Well, where does all of this lead us? For those concerned with obtaining true documentable accuracy from loudspeakers, their pursuit should recognize that a complete set of competently made measurements, properly interpreted, can predict the "potential" of any loudspeaker to achieve accurate reproduction. Stated differently, a loudspeaker that does not measure accurately in several key areas of performance may sound good, pleasing, etc. - but never truly accurate. Thus, one can state that the audible accuracy of a loudspeaker can never exceed that predicted by a complete and accurate set of measurements, properly interpreted. But it is also true that a loudspeaker may not sound as good as predicted by such measurements - because of questionable measurement accuracy, faulty interpretation, etc. Therefore, the "true accuracy" of a loudspeaker must always be verified by both measurements and listening. This means a complete set of competently-made anechoic-chamber measurements, combined (interactively) with exhaustive listening comparisons using live musical instruments. But such live Vs recorded comparisons are rife with difficulties, as may be seen from old newspaper and magazine accounts of what were called "tone tests" during the early days of the present century, when public demonstrations were conducted to show that most listeners could not discern whether they were listening to a popular Met Soprano or her voice being reproduced by an Edison Phonograph. Hmmm! Of course, it can be effectively argued that today's audiophiles are far more sophisticated and discerning than their predecessors were 60-80 years ago. But, even today's most ardent audiophiles and musicians are often easily fooled into believing that only modestly accurate audio reproduction "sounds live". Thus, "subjective opinions" and manufacturer's claims (void of tangible proof) are of little value for determining "true accuracy". Much the same is true of measurements - taken alone! Although a complete set of accurate anechoic measurements, interpreted by an experienced and competent engineer, can establish the "potential" a loudspeaker possesses to accurately reproduce complex sounds and music, they are not infallible. Therefore, both measurements and competent comparisons with live musical instruments/voices are necessary to establish whether "truly accurate reproduction" can be achieved by a given loudspeaker within a suitable listening room. So, is there any relatively simple and reliable means available to the average audiophile for determining the accuracy of a loudspeaker or entire system. The answer is a qualified yes! It depends upon the expectations of the individual and the acoustical properties of the listening room. Expectations are important because "absolute accuracy" will likely never exist - except as a "potential" to strive for. Likewise, there will probably never be a perfect listening room. Thus, some level of compromise will always be necessary as we ascend the "accuracy staircase". So, how do audiophiles thirsting for accuracy improve their systems? To begin, the loudspeaker is almost always the weak-link in the reproduction chain. If it is inherently inaccurate, as shown by a complete set of reliable measurements made by a competent engineering staff at the normal listening distance of about 10 feet within a good anechoic chamber, there is little chance of achieving truly life-like reproduction of good recordings. These measurements must include frequency domain (plots of amplitude Vs frequency, impedance, non-linear distortion, etc.), time-domain (impulse response, step response, waterfall, energy-time curve, etc.), radiation patterns in both horizontal and vertical planes and, last but not least, how closely the speakers are "matched as a pair". Any audiophile or manufacturer that says that such measurements are meaningless for establishing the "true accuracy potential" of a loudspeaker is either ill-informed or simply not knowledgeable. If a loudspeaker does exhibit good measurements, how then can an audiophile further establish its audible accuracy by listening to music? Dunlavy Audio Labs uses the sound of live music (the Colorado Springs Symphony during rehearsals, a live grand piano, a live string-quartet, voices, etc.) in direct comparison with several of our loudspeaker models (including the SC-I). Again, there are very few, if any, audible difference to be heard. (We invite skeptics - even competitors - to visit our facility to determine the validity of these claims.) So, what is the bottom line of all these ramblings? Simple - it is possible to design and manufacture "truly accurate" loudspeakers whose reproduction cannot be reasonably discerned from live music when reproduced within a listening environment possessing good acoustical properties. It is also possible to design and manufacture loudspeakers that are not truly accurate but which satisfy the needs of listeners wishing to hear reproduction that might best be described by such adjectives as "musical, pretty, sweet, nice, full, etc.". And, as long as the customer is happy with their choice, freedom wins again! But to call such reproduction "accurate" is misleading and inappropriate. So how does all of this rhetoric lead to answers that address the comments made in the September 4 posting on the INTERNET by John Otvos? Perhaps, the best way to proceed is to take each of his opinions in the order they appear in his posting. However, because many of his comments beg the question, "what is accuracy?", it might be Q-1: J.O. states that while he respects our efforts at DAL, he believes the science that underpins our engineering and design efforts is "wrong science" while his loudspeaker designs apparently represent the application of "good or right science". A-1: Hmmm! I had always believed that the teachings of science were both universal and well understood by those possessing appropriate academic and professional credentials. To me "wrong science" is not true science and should not be labeled "science"! One of the best definitions of Science may be found in Webster's New Collegiate Dictionary as: "knowledge covering general truths or the operation of general laws esp. as obtained and tested through scientific method." And as: "a system or method reconciling practical ends with scientific laws." Thus, science may be seen as an on-going, objective study of things and a rational attempt to explain their properties in as concise, accurate and complete manner as is possible. Indeed, true science always has as its goal the discovery of truth. Q-2: "Based upon the time measurements that John Atkinson has published on the (Waveform) Mach 17 in the June 1997 issue of Stereophile, our design can 'by no means ... considered a time-coherent design.' But reviewers and owner testimonials have consistently written about its 'speed and dynamics'. However, in Mr. Dunlavy's book, the 17 would not be considered accurate. A-2: The subjective views of reviewers and owners are only an indication that a product satisfies their individual perceptions of "accuracy". This is abundantly clear to anyone who regularly reads the reviews of audiophile products (and particularly loudspeakers) in the many magazines presently being published. Q-3: "If your claim is true, then all speakers designed solely in the frequency domain, utilizing tuned ports are compressed slugs." A-3: Designing loudspeakers to exhibit a flat curve of amplitude Vs frequency covers only one of the several performance parameters that are important for achieving truly accurate performance. Indeed, performance properties related to time-domain parameters and radiation patterns can also significantly affect the audible accuracy of a loudspeaker. Q-4: "I (J.O. speaking) will drag race the transient response of the Mach 17 against your DAL VI or any other speaker known, in public, within a properly structured listening test, anytime and anywhere that I can afford to travel there. I submit, that DAL has not done the psychoacoustic evaluations necessary to postulate such a claim. Both of you (speaking of Andrew, our Marketing Director) have made other claims that cannot be substantiated in the literature of the AES or for that matter on your own web site." A-4: Too bad, but you are very wrong. We at DAL have exhaustively done our homework with respect to psychoacoustic evaluations. In addition, as it has been pointed out, the web site that you have referred to is not an authorized DAL site, and any information contained on this site should be disregarded as erroneous. (If anyone wishes to obtain copies of DAL's white papers, feel free to contact our facility and copies will be made immediately available.) Q-5: "If the DAL methodology were superior, and created better sounding speakers, then more designers would employ the techniques you champion." A-5: Many do! But designing and manufacturing loudspeakers which exhibit truly accurate reproduction, substantiated by both measurements and competent listening evaluations, is a costly endeavor - requiring a competent engineering/technical staff possessing suitable academic/professional credentials, a world-class measuring facility (with large anechoic chambers and a full range of highly accurate test equipment), many years of research & development experience, a sincere and abiding devotion to the truly accurate reproduction of music, a listening room possessing known acoustical properties, and last, but not least, a production facility and Q.C. capability for manufacturing large numbers of loudspeakers that consistently exceed all advertised performance claims and specifications. I wonder how many of the above capabilities, etc. are possessed by Waveform? Q-6: "If the DAL, 'design within the time domain' was indeed better than 'design within the frequency domain', then products from other manufacturers who employ time domain theory would sound reasonably similar. I submit that speakers employing NRC's school of thought such as Paradigm, PSB, Energy, Mirage, Axion, Snell, Waveform and others, sound far more similar with some important differences, than speakers developed using any other means of development." A-6: J.O. attempts to make a "cogent point". However, his expression "...with some important differences' may say it all! Anyway, all of DAL's loudspeaker models, although varying greatly in size, number of drivers, etc., sound virtually identical - except, of course, in the bass region, where the size the cabinet and the woofer largely determine low-frequency limits. Q-7: "Just because a few magazines publish time domain measurements is no assurance that there is any correlation between what is measured and that what is heard in a neutral listening environment. Indeed, the reviewer at Audio magazine stipulated at the end of the first speaker test where the magazine published time domain measurements; that speakers could not be both accurate in the time and frequency domain at the same time. Thanks Ed! So then why publish them Michael?" A-7: Any editor who states that "speakers could not be both accurate in the time and frequency domains at the same time", obviously needs a course in Fourier Analysis, etc. Indeed, frequency response, phase response, waterfall, etc. can all be derived from a measurement of a loudspeaker's impulse response (a time-domain parameter) by use of FFT analysis. We do this every time we measure loudspeakers using Doug Rife's MLSSA system. Hmmm! Q-8: "Transient response is a function of frequency. By definition, lower sounds are 'slower' and higher sounds are 'faster'. The speaker that has the flatter response in the 3-dimensional soundfield (read 0, 15, 30 and 60 degree family of curves) will have the superior transient response and sound more neutral and 'faster' to the cognoscenti as well as non-technical listener, because that's what happens in real time, once again by definition. The Mach 17 is designed to mimic real life sound sources and behaves very much like a point source, but is still many generations removed from the perfectly theoretical model of the point source, a pulsating sphere. Equal radiation in the frequency domain, in all hemispheres, with identical amplitude response." A-8: Here, J.O.'s comments are so wrong that it would take several pages of technical discourse to fully reveal their technical inaccuracies. First, transient response cannot be derived from the frequency response of any component, including a loudspeaker. It is the opposite: frequency response can be derived from the impulse (transient) response of a component using FFT analysis. Further, "the speaker that has the flatter response in the 3-dimensional soundfield will have the superior transient response and sound more neutral and 'faster' ... , because that's what happens in real-time, by definition." Wrong! Very wrong!! Once again, J.O. is providing evidence that he simply does not have a grasp of the connection between the time and frequency domains. Ask any competent engineer or math major who is correct! Q-9: "Waveform doesn't claim to have invented the first all active system or be the first users of the egg as a baffle or first implementers of corrective contours to better interface with rooms of differing reverberant decay or to offer a wide dispersion design etc, etc. We only claim to be the first company to have put all the best features known to applied, acoustic science into one package, and to claim that it works very well indeed and to offer it at a fair but unfortunately still high price. That's all. That's all we've ever claimed." A-9: My, my! Really, claiming Waveform to, "... be the first company to have put all the best features known to applied, acoustic science into one package ..." is quite preposterous! Substantiating such a claim, would make a very interesting technical paper for a peer-reviewed journal, like that published by the AES. Hmmm! Why not? Q-10: "DAL can claim to make the most accurate transducers forever, but claiming that doesn't make it so. I personally find that continued statement intellectually arrogant, especially in light of the fact that your site has been perpetually "under construction." A-10: As specified before, the web site that you have referred to in not an official DAL website. If you would like to receive a complete set of measurements from any DAL loudspeaker (we measure every single unit, and keep hard records of this data on permanent file), I would be more than pleased to send you a copy. Q-11: "Please flaunt your goods. Publish the family of curves for the DAL VI at your web site. Don't just claim that: 'off-axis response of all the SC loudspeakers is extremely good". Which leads me to ask, why couldn't you have just turned the speaker for JA in your chamber so he could measure a 30 degree curve of the DAL VI? Display them for your latest products as Waveform has since Nov. of '95, for the world to critique. We are all in the public domain here and must take our lumps. Mistakes have & will be made." A-11: DAL does, of course, publish a complete set of performance specifications for each of its loudspeaker models - with a guarantee of their accuracy. We have also measured their radiation patterns in both horizontal and vertical planes within our large anechoic chambers, using a heavy-duty rotating mount that can support up to about 1,000 pounds. However, measuring the radiation patterns of dimensionally-large loudspeakers, even within an anechoic the size of those at DAL, is a daunting task. Such measurements tend not to yield the levels of accuracy, particularly at low frequencies, that we feel comfortable publishing alongside the very accurate measurements we publish for all other measurable performance parameters. We do, of course, rely on the interpretation of such pattern measurements in the design of all of our loudspeakers. Q-12: "Your claim that a speaker should behave like a competently designed amplifier is really stretching credulity. Yes, an amp should reproduce a pristine squarewave, its destiny after all is to process electricity and part of that function is passing a decent electrical waveform with flat frequency, low IM and THD as well as having an inherently non-audible S/N ratio. Yet a transducers function is to change that electrical waveform and *transform* it into an acoustical waveform within the 3 dimensional soundfield. Prove that a speaker's job is to pass a squarewave. Back up your assertions with qualified and quantified listening tests that have been peer reviewed". A-12: A course in "Linear Network Theory" taught to E.E. students at most universities, clearly teaches that, "within any linear system comprised of several different electronic components connected in series, if one of the components exhibits linear distortion (such as an inability to reproduce a complex waveform), the result remains the same regardless of the location of the component within the chain of components comprising the system." Therefore, if you would not consider the purchase of an amplifier, pre-amp, etc. that exhibited poor reproduction of complex waveforms, why would you purchase a loudspeaker with such a shortcoming? Hmmm! I would also call J.O.'s attention to the many, many "peer reviewed" magazine articles that have covered the performance of the many different models of loudspeakers that I have designed since 1974. Every one of these reviews have provided rave comments regarding their audible and measured performance. In fact, my first "production loudspeaker", the DL-15, was reviewed by the highly respected Dr. Richard Heyser in a 1976 issue of Audio - as the most accurate reproducer of a piano he had yet heard (his standard for assessing the audible accuracy of a loudspeaker). In fact, several of the Duntech loudspeaker (which I designed) and our present DAL line, have consistently received "best-sound-at-the show", "product -of-the-year", loudspeaker-of-the-year", etc, in Stereophile and many other magazines - including highly respected overseas journals. In addition, DAL loudspeakers are currently the "reference monitors" used by numerous, well-known recording and mastering studios, both in the U.S.A. and overseas. Although J.O. made a similar claim for his loudspeakers within a recent magazine review, one of the well-known recording companies he mentioned denies that they are using his speakers as their monitors. Hmmm! Q-13: "Those who employ first order filters and time alignment are confusing technique with theory. They are letting their nifty measurement techniques for phase and time, dictate their theory rather than vice versa. The ear is sensitive to level not phase. Level is amplitude and that equates with frequency response. This has been stated time and time again. Time align a speaker ... no problem. Now go back and time align the room, and every room that your loudspeaker may see service in. I won't hold my breath. In MHO, this condition of placing the cart before the proverbial horse, arose within the high-end, when the judgment of a decent hi-fi system was reduced to its ability to portray a good soundstage first and foremost as opposed to its ability to play in tune with "pace and rhythm". Gee Martin, what ever happened to the terms frequency & transient response? Should we wonder further why music listening is sliding?" A-13: "... confusing technique with theory." Really? My, my - how judgmental! I really believe that DAL and its staff of qualified engineers and technicians deserve a better assessment. But, maybe J.O. does not possess the professional and academic qualifications to make such a judgment. Perhaps he would like to make copies of his C.V. available to those of us who might like to determine from whence he speaks? (My own C.V. and that of other engineers at DAL are always available, upon request, for examination.) Q-14 : "The problem for speaker designers as well as for magazines who really wish to understand the correlation between what is heard and what is measured is that they both need to take a longer pulse from the patient . Now, this means they aren't taking *enough* measurements in the frequency domain. Many do. More still do not. There are the off-axis measurements and the sound power calculations to arrive at, that are *equally* as important. Then there is the listening window computation & and distortion measurements too. Spend more time in the chamber and rotate that turnstile please. Or just move the damn mic for crying out loud!. Recording engineers do." A-14: Wow! How judgmental - again! We at DAL probably spend more time taking accurate measurements of loudspeakers, and comparing them with live music under carefully-controlled conditions, than any other single competitor, including Waveform. And this includes copious measurements of frequency domain parameters, time domain parameters, non-linear distortion, radiation patterns, etc. As mentioned earlier, DAL also spends a significant amount of time and effort recording a symphony orchestra, a string quartet, a Steinway grand-piano, voices, etc. and comparing the sound in real-time with that reproduced by several different models of our loudspeakers. (Perhaps J.O. missed the TV show "Beyond 2000" which aired a lengthy, real-time comparison of our 1980's Sovereign 2000 loudspeaker with that of a live singer/guitar-player, seated directly between the pair of loudspeakers. Not even a professor of music at the local university was able to discern any audible difference between the live and recorded music. Hmmm! We have repeated this with our SC-IV and SC-V loudspeakers, in conjunction with a nationally-recognized string quartet (previously recorded in one of our large anechoic chambers). Q-15: "You choose 10' in a chamber for a "meaningful" measurement. John, the world standard is 2M. The world standard for MLSSA testing is 1M in a normal room. Not less and not more. Should anyone wonder that we as an audio community have separated when the recording industry could not agree on a few basic minimum standards such as: 1) minimum CD length, 2) spars code labels on exterior for all releases whether new or old, 3) a set playback level for all discs *** for pity sake!, 4) elimination of an idiotic jewel box cover with hinge tabs that habitually break off. A-15: As far as I know, there is presently no "world standard of 2 M" for loudspeaker measurements nor a 1 M standard for taking MLSSA measurements. We use a distance of 10 feet because we believe it represents the average listening distance for most audiophiles. (By the way, Doug Rife, the inventor and producer of the MLSSA system uses our SC-V loudspeakers in the reference system at his home in Florida.) Hmmm! Q-16: (Begins by quoting my net posting of 19 Dec. 1996, re: "... a symmetrical, time-coherent arrangement of drivers ... can yield an effective 'point-source' of radiation with symmetrical radiation patterns in both vertical and horizontal planes that are much wider than the radiation patterns of any large diaphragm electrostatic loudspeaker.') J.O. then states: "The comparison, I'll give you that. There is little dispute since electrostats are no longer much of an issue for the consumer even within the high-end, but not wider than a design that employs fewer drivers and takes much more care in physically aligning the acoustic centers of a single midrange and tweeter. Your designs are a far removal from the concept of a point source and actually come closer to being described as a quasi-line source, although they really aren't tall enough for that either." A-16: Wrong! At the normal listening distance of about 10 feet, our loudspeakers effectively simulate a "point source of radiation", usually defined in "wave theory" as exhibiting a maximum error of about 18 degrees across the frequency range (in this case the mid and tweeter ranges, where wavefront phase might be audible). Antenna measurements employ the same criteria with respect to the distance a measurement must be taken to achieve "far-field accuracy", i.e., an error of less than about 18 electrical degrees across the radiating aperture at all frequencies being measured. Since human hearing is less responsive to differences in phase than amplitude at low frequencies, the 10 ft. distance we use seems quite appropriate, especially since we are measuring time-domain parameters such as impulse, step, waterfall, energy-time, etc., all of which would reveal any departure from an effective "point source of radiation" if such existed. Our measurements show no such departure, and reveal impulse, step and other complex performance parameters that are better than those exhibited by most expensive, top-of-the-line tube amplifiers. Can J.O.'s loudspeakers match this kind of performance? If he believes they can, let him submit his objective proof. Q-17: "When too many drivers are employed, the response must be normalized by using compression techniques such as 1st order filters and many corrective circuits in the pass band which also cause dynamic compression. Got to make those consumers think they're getting high technology; just sell 'em more drivers. I once saw the "platter layout" for your 2001 Sovereign. Tell me that all those cazillions of components don't affect the dynamic compression of the basic design? Better yet, tell us why after a decade, the pros seem to prefer the Sovereign to your latest offerings? A-17: "Compression techniques such as 1st order filters and many corrective circuits in the pass band"! Really, J.O., I believe you are demonstrating an apparent lack of understanding circuit theory, etc., by making such a statement. Passive network components, such as high-quality polypropylene capacitors, air-core inductors and linear resistors cannot, and do not, create any non-linear signal compression in the amplitude domain. All such components exhibit linear properties and, as such, do not exhibit "dynamic compression". Ask any competent engineer. With regards to professional use of our current line of products, such elite recording, mastering, and editing facilities such as Dorian Recordings, Reference Recordings, The Hit Factory, Absolute Audio, Sony Music New York, Sony Music Los Angeles, Sony Music Canada, Sterling Sound Studios, Synchrosound Studios, among numerous others are currently using Dunlavy Audio Labs Signature Collection loudspeakers as their studio reference monitors. Q-18: "A geometric symmetrical driver array laid out on a baffle, does *not* produce a symmetrical wave launch. This is evidenced by listeners having to find the sweet seat and then not leaving it, even for a lean. There have been other posts as I'm sure you are all too aware of. No speaker with that many midranges and woofers can have a smooth vertical response. That sir, defies the laws of physics. The vertical and horizontal response combine to produce the total radiated power response that *will* become compressed in such a design. The listening height with your methodology is far more critical than with a design that has effectively no baffle for the shorter and more directional frequencies and that: A0 uses no felt to absorb the off-axis radiation of its drivers, B) employs the most pure physical shape known to acousticians such as an egg, to virtually reduce diffraction to negligible levels." A-18: Wrong, again, J.O. A geometrically symmetrical, time-aligned array of loudspeaker drivers (w-m-t-m-w) does produce a symmetrical wave launch, if that means the radiation of a wavefront that is symmetrical about the center of the tweeter's effective point of radiation. It also means that the wavefront produces a coherent image, independent of frequency, at an on-axis distance of 10 feet. This coherent wavefront persists at off-axis horizontal angles of 30 degrees, or more, at 10 feet. Much the same is true for vertical angles within a window of about plus/minus 10 degrees. Q-19: "Where is diffraction in your model?" A-19: By using efficient acoustical absorbing material, covered by Pat # , diffraction of sound waves from the edge (and other baffle boundaries) of our loudspeakers is reduced to inaudible and virtually immeasurable levels. Without the use of such efficient absorptive materials, DAL could not achieve the accurate performance it does from our loudspeakers. Q-20: "You finish by stating in your article ... that; . J.O. then goes on to comment that, "The word accurate is loaded within this context. Our speaker would fail your test methodology as the time measurements in Stereophile have clearly shown. The Mach 17 would fail your 'standard' in your test lab but then your IV, V and VI would fail our standard in our test lab. So who wins? This is where most of the print media has failed the consumer so miserably in not demonstrating more leadership. This is where the intellectual wrestling ends and the real work begins." A-20: What standards of performance is J.O. referring to in his statement that DAL's loudspeakers "would fail our standard in our test lab"? Hmmm! It would be interesting to discover just what Waveform's lab and its equipment consists of - Hmmm! Perhaps, J.O. would like to compare a list of his lab facilities and measurement equipment with a list of those here at DAL? J.O. then continues his diatribe by describing the rules for a hypothetical "racing contest" between different loudspeaker models. Gosh! I always thought cars and bikes were used for racing contests, not loudspeakers! How does one go about racing loudspeakers - or have I missed something? I originally entered the field of loudspeaker design because I believed that loudspeakers, with few exceptions, were the most misunderstood of all audio components and represented the weakest link in the chain of audiophile components. I still hold that belief. Loudspeaker design may appear simple - but it is one of the most complex fields that exists - if we are referring to loudspeakers capable of "truly accurate reproduction", based upon both a full set of competently-made measurements and carefully conducted blind A-B listening evaluations. Do I believe that truly accurate loudspeakers can be designed almost entirely by means of controlled "listening" sessions in an acoustically appropriate environment by someone possessing few, if any, relevant professional/technical credentials? Hmmm! Do I believe that horses have wings and can soar with eagles? My considered answer to both is an unqualified NO! Best regards, John Dunlavy From 102365.2026@compuserve.com Wed Nov 12 12:36:01 1997 Newsgroups: rec.audio.high-end Subject: Radiation Patterns of Dynamic Loudspeakers From: John Dunlavy <102365.2026@compuserve.com> Date: 12 Nov 1997 13:36:01 -0500 Several recent postings here on the net have provided blurred and or incorrect opinions regarding the radiation patterns of loudspeakers using cone drivers. An example is the belief that the shape of the enclosure or housing surrounding a driver controls the directivity of the loudspeaker over its frequency range. As a consequence, we find designers using egg-shaped, spherical and cylindrical enclosures, claiming that their chosen shape provides a near spherical radiation pattern etc., representing a more accurate reproduction of the original sound. Hmmm! Lets examine the facts taught by competent engineering and physics. To begin, at frequencies above which the diameter of a loudspeaker driver exceeds about one-half wavelength (answer in inches equals 6,500 divided by freq.), its radiation pattern becomes directional (less than about 90 degrees between -3dB points, i.e. plus/minus 45 deg.). Assuming more-or-less "pistonic" motion of a cone or dome driver over its entire surface, the width of its radiation pattern (between minus 3 dB points) is approximately equal to 51 divided by the diameter of the cone/dome, expressed in wavelengths at the frequency being evaluated. Likewise, the width of the radiation pattern between first nulls of the same driver can be found by dividing 114 by the same cone/dome diameter. (For example, the beamwidth between minus 3 dB points for a 5 inch diameter midrange driver would be approximately 53 degrees and 2.5 kHz and about 26 degrees at 5 kHz.) However, at frequencies below that at which the diameter of a driver's radiating surface becomes less than about one-third wavelength, the width, height and shape of the surrounding enclosure begins to affect the shape of the radiation pattern, the width of the main lobe, etc. In this regard, the math and computer programming needed for determining the radiation properties of almost any shape of enclosure, incorporating one or more drivers and different crossover slopes, is well-known and understood. Further, when a loudspeaker consists of a vertical stack of tweeter, mid and woofer, the spacings between driver centers also affects the vertical dispersion patterns at frequencies near the crossover points. So, hopefully this information will help clear the air with respect to the radiation pattern properties exhibited by individual drivers, drivers-plus-enclosures and arrays of drivers. But, sadly, there will always be "self-appointed" experts - lacking appropriate engineering credentials and relevant professional-level experience - who will proclaim otherwise. And, personal "taste" will always play a role in what each listener perceives to be "accurate reproduction". But those searching for truly accurate reproduction, that closely matches the original live performance, will almost always prefer listening to loudspeakers whose accuracy is verifiable by a complete set of accurate measurements and critical comparisons with live music in real-time! Caveat emptor - nuf sed! 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.