Box 392: Letters to the Editor (Vol.1, No.5: Winter 1977/78)

For the first time, in addition to subscriber correspondence from both the professional and the private sector, we're including reactions and rebuttals from manufacturers and equipment designers whose products we have reviewed in past issues. Our criterion for printing such material is the correspondent's attempt at reasoned and factual argument, whether the alleged facts are true or not. Purely vituperative and uninformative who-the-hell-do-you-think-you-are letters have been, and will continue to be, ignored. The letters we publish in this column may or may not be excerpted, at the discretion of the Editor.

Ellipsis (...) indicates omission. Address all editorial correspondence to The Editor, The Audio Critic, Box 392, Bronxville, New York 10708.

First, the letters of general editorial interest, starting with a real mind stretcher from the world's smartest audio designer below the legal age of adulthood.

The Audio Critic:

I can sympathize with your frustrations over the search for conclusive amplifier testing. Trying to quantify those nebulous characteristics which we call "musical accuracy" can often seem impossible. I think, though, that partly responsible for the repeated failures en countered by critics and designers alike is a desire {or one test which will tell everything one needs to know about a system's performance without requiring any other test data, knowledge of circuit design, or basic understanding of the nature of a musical signal.

Take, for example, your "pet" test, CCIF IMD. This test had been yielding some reasonably correlatable results until suddenly you encountered the condition where two amplifiers measured drastically differently, but sounded al most equally good. The assumption here should not be that the test is invalid (after all, as Chris Russell of Bryston stated in your issue Number 4, "distortion is distortion'), but rather that Amplifier One, which is not at all fazed by this test, makes some other error, which Amplifier Two carefully avoids.

The net audible effect, call it the total distortion of each, is approximately the same.

Now you are faced with the situation where you know that there are at least two conditions which must be tested for, but you don't even know what the second one is. When I reached this juncture some time ago, I decided to forget all of the preconceived notions and attitudes which led me to this predicament, and begin from the beginning.

I felt that a better knowledge of the kinds of errors occurring in amplifiers, as well as an understanding of how and why they exist, would lead to a more conclusive and correlatable body of testing procedures.

Any sound can be characterized by a change in amplitude over some period of time. This can be expressed as AA /at or, in the limit as At approaches zero, as dA /dt, where A is the amplitude and t is time. This is all Edison needed to know in order to invent the phono graph, and it's all we need to know to examine the performance of an amplifier.

Any attempts to transform this basic truth into the frequency domain, or any other set of coordinates, assumes conditions (i.e., periodicity, etc.) which serve only to complicate matters. Remember that while harmonic structure can be represented as amplitude versus time, the reverse is not necessarily true. Frequency domain test methods have be come our standard as their results are quite easy to observe. It's too bad that generally they don't correlate with what we hear.

Expressing sound as dA/dt, we can characterize the complete function of any amplifier as "output equals gain times input," or, for input equal to dA/dt, outputs equals G(dA/dt), where G equals the gain of the amplifier. This leads to two, and only two, possible families of distortions:

The first type of distortion is a result of variability of gain. For an amplifier to pass a signal undistorted, its instantaneous gain (that is, the gain at any point in time) must be absolutely constant. Harmonic distortion is simply the case where gain varies with amplitude, usually due to nonlinearities in device transfer functions. In a completely distortion-free amplifier it would be possible to divide any output signal into a series of vanishingly small time integrals and find that the gain of the amplifier for each of these periods is the same.

The second type of distortion is caused by a change in time base characteristics. From the above expression of any amplifier's function, it is clear that there should be no effect whatsoever on the time component of the input signal. That is, for any interval 'dt' in the input signal its value must be unchanged by the amplifier. For example, if the input were composed of two pulses separated by, say, one millisecond, they should appear at the output exactly one millisecond apart. If their width at input is, say, one hundred microseconds then their width at output must also be one hundred microseconds. Unfortunately, in real amplifiers things are not this simple.

Take this case of a signal composed of two pulses. We know that real amplifiers will not respond to this signal instantaneously, but will exhibit some propagation delay. If this delay is the same for both pulses, then the distance between them will be unchanged from input to output. However, if the instantaneous forward delay of the amplifier is not absolutely constant, and one pulse is delayed longer than the other, then the distance between them will not be accurately preserved. Thus, in a distortion-free amplifier, time delay must be totally invariant.

Now that we have established the two basic performance criteria for all amplifiers, we can examine what happens to signals in real circuits. Take, for instance, the "emitter follower", a very common transistor power amplification configuration. Because of the effects of junction capacitance there will be some delay inherent to this circuit. This delay is a function of emitter current which is, in turn, partly a function of base voltage. As this current is increased the capability to overcome the junction capacitance is greater, thus reducing the delay. The problem is that increased signal amplitude results in higher base voltages, and the result is obvious: this is one where propagation delay is not constant, but varies with amplitude.

This leads to an interesting discussion of the Class A phenomenon. In general, Class A stages tend to sound better than Class AB stages. This difference has been commonly ascribed to the absence of any form of crossover notch in Class A configurations, yet none of the better Class AB amplifiers exhibit any such aberration. (This distortion is easily measured as increased THD or IMD with decreased signal amplitude.) The reason that Class A is capable of sounding better, in some cases, than Class AB is simply that the output stages are constantly handling such high currents that they are easily able to overcome delay producing capacitances. Because the values of the delays are shorter, any resulting changes in delays are shorter, and thus less audible.

Given that increased current always results in increased amplitude-related distortions, as well as decreased time distortions, the fact that Class A can sound better than AB indicates that maybe time is more important than amplitude. Of course, understanding these ideas should make it possible to design an AB con figuration with minimum time and amplitude distortions.

Heat is another factor which can result in variable propagation delays.

The laws of physics tell us that conductivity is directly proportional to heat; as the junction temperature of a transistor increases it becomes a better conductor, reducing its effective capacitance. This means that as an amplifier heats up, its delay time is reduced.

Most amplifiers, especially those de signed to operate "cool to the touch", exhibit rapid changes in junction temperature and thus rapid changes in delay times. Consider the case of two high amplitude pulses appearing at the input of a cool amplifier. The first pulse is delayed in reaching its peak value, but once it has, the resultant increased cur rent demands on the circuit cause a sudden increase in operating temperature.

By the time the second pulse reaches the amplifier, the devices are no longer cool, and the delay time has changed. Here again Class A designs excel because not only are the values of delay changes smaller, but the enormous thermal capacity afforded by their necessarily mammoth heat sinks slows down temperature shifts.

All of this leads to another interesting point. It has been generally accepted that amplifiers with capabilities for very wide bandwidth sound better than those whose bandwidth is limited, even though we can't hear above twenty kilohertz or below twenty hertz.

The fact is that every amplifier can be considered a bandpass filter, and depending on the exact characteristics of the equivalent network, this filter can have quite variable delays. Those amplifiers incorporating a 12 dB per octave high pass filter, whose rolloff begins just below twenty hertz, exhibit changes in delay times a decade or more above the cutoff point. Thus delay in these amplifiers will change with frequency. Extending the passband of a circuit above and below audibility will push these time nonlinearities out of the audible range.

In addition, one cannot overlook the effects which reactive loads have on amplifiers. Since present circuits are far from ideal voltage sources (they all have some source impedance and reactance), their interaction with loudspeaker networks results in some filter circuits having definite time characteristics. As the source impedance of the amplifier is decreased, the effects of this phenomenon will become minimized. Also, for reactive loads, the current demands placed on the amplifier in order to maintain a given output voltage can change, thus possibly changing the delay characteristics of the circuit.

It is tempting, at this point, to assume that if the delay of an amplifier is constant it can otherwise be ignored.

Unfortunately, with most amplifiers this is not the case. As long as feedback is involved, delay of any kind is an important matter. During its forward propagation delay period an amplifier is operating open loop, resulting in a short burst of increased "instantaneous gain'.

As defined by our original representation of an amplifier's function, this is distortion. Fortunately, most amplifiers are fast enough so that this simple aberration is below the threshold of audibility.

Unfortunately, in most amplifiers the condition is prolonged by overload.

When the open loop amplifier clips, it can often take many times its propagation delay to once again become linear.

This phenomenon can be quite audible.

Susceptibility of an amplifier to this condition is a function of its delay time, amount of feedback used, overload recovery characteristics, and signal rise time.

Simply put, to prevent the amplifier from overloading, the product of propagation delay times signal slew rate should never exceed the maximum output of the amplifier divided by its open loop gain. If this condition is met, then there will be some feedback returned to the amplifier input before the signal reaches a level which will clip the open loop amplifier. Ensuring that this is the case involves minimizing propagation delay, minimizing open loop gain (and thus feedback), increasing maximum out put capability and/or filtering (slowing down) the input signal. Therefore, in order to evaluate the effects of a given propagation delay for a specific amplifier one must know something about its circuit design. That is, if two amplifiers have the same constant propagation delay, the one with less feedback, greater headroom, better overload recovery, and a filtered input will probably sound better. However, if one has a lot of feedback, but good overload recovery, and the other has no headroom, but a filtered input . . .

These are the kinds of problems which cause audible distortions. Until our testing procedures are designed to discover such inaccuracies we will never be able to separate better from best on the laboratory bench.

Very truly yours, Andrew S. Rappaport A. S. Rappaport Company, Inc.

Armonk, NY

Cf. our comments introducing Part III of the power amplifier survey in this issue. (Thank you, Andy, for some exceedingly nourishing food for thought.)

-Ed.

The Audio Critic:

...Concerning the use of separate woofers/subwoofers: If the time domain is so important, and pulse response measurement becomes likewise important in loudspeaker measurement, why doesn't the use of a physically displaced woofer cause poor time domain response and poor pulse response measurement? Unless I misunderstand what I have read on the subject, a pulse, because it is of no specific frequency, acts as all frequencies at once; therefore, part of the pulse would be reproduced by the woofer.

If this were so, if the woofer were displaced, as with the DQ-1W behind the DQ-10's or in conjunction with the Rogers, wouldn't the pulse be distorted? (Or even with the Beveridge 2SW using the subwoofers.) Evidently this is not the case, as borne out by measurement and listening tests (or else you would have not failed to mention it), but I am un able to figure out why. Could you please elucidate? . . .

Sincerely, Michael Steiner; APO San Francisco

You're absolutely right; a separate subwoofer can be time-aligned with respect to the main system only with one specific placement, if at all, which is unlikely to be the placement used in an actual listening situation. Luckily the ear is increasingly less sensitive to wave-front coherence as we go lower in frequency, at typical subwoofer cross over frequencies (say, 70 to 120 Hz) it's quite insensitive. In our pulse tests, we use pulse widths of I msec to 0.1 msec, corresponding to half-wavelengths of 500 Hz to 5 kHz. That's where the action is when it comes to time coherence.

-Ed.

The Audio Critic:

I would like to have an accurate speaker system. But I have one big problem (along with most other audiophiles, I imagine): I cannot afford any thing that fits in that category.

You and your ''consulting engineers and other technical advisers" seem to know what it takes to make a reason ably accurate speaker system. Would it be possible for you and/or your fellow worker(s) to design a system that would pass the scrutiny of your panel of golden ears and would not be too expensive; then write an article on where to get the drivers and how to build the system? If this is at possible, I and a lot of other stereo nuts would be very grateful. If not, well I tried and thank you very much anyway for your time.

Sincerely, J. Dennis Smith Belle, WV

A reference loudspeaker made of conventional (meaning nonproprietary) drivers is definitely one of our future projects. Whether the typical audiophile will be able to duplicate it without the aid of a calibrated microphone, signal generators, oscilloscope, spectrum analyzer, etc., remains to be seen. Don't hold your breath, though, until this project comes to fruition.

-Ed.

The Audio Critic:

Having plowed through the Mark Davis correspondence, I think that Mr. Davis has a point. This is not to say that I completely agree with him, but he has done the homework and run controlled tests, and you have not. The point, in other words, is that he has established the minimum conditions which must be met in order to test the significance of any parameters other than frequency response. Therefore, if you run comparative tests without satisfying those conditions first, your conclusions are as questionable as if you twisted your cartridge around at random between A-B tests.

If, as you say, you are serious about trying to correlate measurements with sound, and if you find Mr. Davis' conclusions unacceptable, then the only way to refute him is to duplicate his tests using your equipment and see if he is right. Informal tests and *'the painstakingly accumulated wisdom of an entire generation of audio perfectionists" just won't do it-after all, for hundreds of years the accumulated wisdom was that the earth was flat. And, Mark Davis notes in his letter that in several informal tests he was able to hear differences which disappeared upon formal testing. To simply state that his system had insufficient resolving power without showing that a better system would demonstrate differences is giving in to the "golden ear mysticism" that you deplore. And trying to find more subtle objective-subjective correlations without first eliminating the known effects of frequency response and level variations is simply sloppy.

Having said all this, I do want to say that I think you have presented a lot of good, solid information in your publication and I hope your efforts to find "good" measurements succeed. It is simply common sense, as well as good science, that the more sensitive the measurement, the more careful one has to be to eliminate any possible extraneous factors.

Sincerely, James Lin; St. Paul, MN

You're quite right on a general, philosophical basis, but you're ignoring the specifics of the situation. When a renowned practitioner announces a start ling and controversial discovery, it does indeed behoove other practitioners to spend some time and effort on duplicating his experiments. You forget, how ever, that Mark Davis isn't exactly a Maxwell or a Helmholtz. He's the guy who blew his credibility when he denied the existence of TIM and impugned the research of Matti Otala (and distinguished predecessors). So if Mark Davis should suddenly announce that smearing chicken fat on the transistors improves transient response, we wouldn't necessarily drop everything then and there and run to the deli to initiate a verification. On the other hand, we might publish his letter on the subject because it would probably amuse and stimulate our readers.

Quite independently of the above, however, the new preamplifier survey reported in this issue did include listening comparisons between units with exactly matched frequency response and signal level-not because of Mark Davis but because it was easy and convenient to do so. And, what do you know, we heard distinct differences. Read it and weep.

-Ed.

The Audio Critic:

I object to the philosophy in Mark Davis's letter that the pursuit of excellence in high fidelity is a waste of time if it goes beyond the known psychoacoustic limits of the ear. The goal of high fidelity is to reconstruct a time-varying sound pressure field that spatially and temporarily duplicates some original field. The ear is irrelevant to the statement of the goal. The goal is objective, not subjective. It is a matter of physics and not of psychoacoustics.

Psycho-acousticians can provide guidance whenever a design trade-off must be made, to make the best choice.

But when no trade-off is required, when the SOTA permits 0.0008% distortion equally well as 0.2% distortion, who is Mark Davis to say I'm foolish to prefer 0.0008%? Likewise with TIM. If TIM exists and we can eliminate it, then let's do so. Who cares whether it's audible? I am glad to see that someone be sides Richard Heyser understands the importance of a loudspeaker's impulse response and will defend it in print. It is a fact of mathematics that if a loud speaker operating in its linear range has a good impulse response, then it must have a good frequency response (flat magnitude response and linear phase response). The converse is also true.

Good frequency response (magnitude and phase) implies good impulse response. However, a good magnitude response alone does not imply a good impulse response.

There is no excuse for any loud speaker designer today not to use impulse response testing . . .

Sincerely, Stephen D. Stearns Research & Development Engineer GTE Sylvania

Mountain View, CA

The Audio Critic:

In your most recent "Admonitor" you censure Epicure and Bose for sophistry in the design of their speakers. While I agree with your conclusions regarding the generation of time delay in the two speaker systems, the argument you presented was simplistic. The reverberation recorded during a performance is inadequate for completely portraying the acoustics of the concert hall. In fact, recording engineers deliberately attempt to exclude much of the natural reverberations of the hall by using close-mike and directional-mike re cording techniques. The reverberation is added subsequently using artificial means.

Excluding the natural reverberation is actually desirable. If, as you state, the full character of the reverberant sound were retained in a recording, then ideal recordings could be obtained by placing two microphones at your favorite listening position spaced in correspondence to the distance between your ears. Unfortunately, the result of recording using this attractively simple technique is that the sound becomes garbled. The admixing of the echoes results from the loss of vital information in the process of re cording-information about the directionality of the individual echoes making up the reverberant sound. Without the directional information all the echoes are homogenized, producing an aesthetically displeasing result. Theoretically two channels of recorded sound should be sufficient to convey the required information. After all, our two ears detect only two signals in the original experience. The problem is recording the proper signals. Stereophonic recording is fundamentally incapable of accurately conveying directionality, particularly when the virtual source is not located between the two speakers. Binaural re cording must account for directionality in the reverberant sound field, information which is not conveyed in the present art.

A fundamentally different approach is to synthesize the required information. Instead of attempting to capture the echoes of the hall in the original recording, the delayed replicas are generated during the playback process according to our knowledge of the situation in which the recording was made. Unfortunately, this technique has practical problems as well. A typical concert hall has thousands of significant echoes. The early echoes are especially important in determining the acoustical character of the hall. To effectively re create the sensation of the original hall, the timing, amplitude, and direction of each echo need to be carefully controlled.

The problem with the two speakers which you justifiably criticized is not that they attempt to generate the echoes, but simply that the temporal and directional pattern of the echoes which they do generate is wrong. The echo response of the speakers is dependent upon the geometry of the home listening room. Accordingly, the directionality of the echoes will correspond to the listening room and not the concert hall. Furthermore, the delays will be about an order of magnitude smaller than in a concert hall-milliseconds in stead of tens of milliseconds. Your ear will not resolve two signals spaced by only a few milliseconds. The result will be what you described as blurring or smearing of the sound. Nevertheless, this distortion is an implementation problem which is not inherent in the synthesization approach. Synthesizing the reverberation during playback is basically a valid and powerful approach.

One of the most important problems still remaining with high-fidelity reproduction is accurately portraying the ambience of the original hall. Traditional recording techniques do not realistically convey reverberation be cause they discard directional information. Synthesizing the reverberation during playback is a valid approach but the echoes must be generated properly or, as in so many things, the cure will be worse than the disease.

Very truly yours, Jeffrey Borish Sound Technology, Inc.

Campbell, CA

We find this very sad. Here's a maker of sophisticated instrumentation for measuring audio accuracy (we have a Sound Technology distortion measurement system in our own laboratory), and what is he doing? He is arguing on behalf of inaccurate audio! The best way to dispose of all arguments in favor of 'synthesized' reverberation or any other tampering with the signal is to listen to Volume One of the Mark Levinson Acoustic Recordings (an organ and choral program) on a properly designed and aligned play back system. The record was made with two microphones straight into the tape recorder, tape recorder straight into the cutter amplifier, no signal processing, no added reverb, no black boxes, no nothing. And the acoustics of Dwight Memorial Chapel at Yale are there beyond the wall of your listening room, big as life. End of discussion-next! (But see also Max Wilcox's article on the same subject in the back of this issue.)

-Ed.

The Audio Critic:

A simple misreading of the definition of 3, and no "careless trig,"" caused me to erroneously measure L sin 8 of the Grace G-707 and SME tone arms, and then to contradict your claim that "tonearm designers (have) forgot ten their high school geometry." (We agree. Your misreading was indeed simple.-Ed.) I don't quite understand the intensity of the editorial blast that resulted (Have you forgotten your remarks about our "irresponsible sloppiness," causing our "credibility to deteriorate rapidly," etc., etc.? -Ed.), but it is quite clear that, having correctly found one error in one of my arguments (I apologize profusely; consider yourself lucky that it didn't cost you $28 to read it) (huh? -Ed.), you then implied that all my remarks are "just plain wrong" and proceeded to "make an example of them." Unfortunately, you sidestepped the larger technical questions I was attempting to bring to your attention:

1. What good is correct tonearm geometry if you can't exactly align your cartridge with that arm? This question is mainly directed to tonearm and cartridge manufacturers, who could quite easily produce self-aligning cartridges and headshells, along with the correct geometry (I hope) they are now designing as a result of your article. Until such a design is available, we are left with the tedious and inaccurate alignment method described in your issue #4.

There is a very interesting sentence contained in the description of that method (p. 55): "Then you twist the cartridge in the headshell so that you can see absolutely no tracking error at those two points." Please explain to us how, without benefit of a transparent headshell, this can be accomplished with the accuracy of 0.1 degrees which you imply is necessary? Up until that point, the alignment procedures are quite sound, but it is at the crucial point of angular alignment of the cartridge that all of this hard-won accuracy is thrown away.

(Not thrown away, but-you're right-in jeopardy. This is by far the most troublesome, least predictable and therefore least formalizable part of the alignment. With cartridge bodies having all right angles, as in the Denon DL-103 series, it's relatively easy. With more complicated shapes, such as the GAS 'Sleeping Beauty' series, it can be a real pain. A small mirror, accurately scored with the appropriate guidelines, is one way around the problem. Our next issue-Number 6-will contain additional information and suggestions on this very subject.

Meanwhile, we can assure you that doing the alignment as best you can, minor inaccuracies and all, will still get you a whole order of magnitude better performance than not doing it at all. -Ed.)

2. At no place in my letters did I suggest or intend to suggest that a Bessel (Q = .58) response was either optimum or practical for a woofer. As anyone who reads my letters can see, I was merely correcting your obviously erroneous contention that Butterworth tuning provides the best possible transient response.

(Butterworth tuning does provide the best possible transient response while maintaining flat amplitude response to the lowest possible frequency. -Ed.)

In a larger context, it needs to be made clear that Butterworth and Bessel responses are merely mathematical abstractions: two particular points on an infinite continuum (a multi-dimensional continuum, by the way) of filter tunings; that Butterworth is a calculated set of conditions which (it can be proved) provides the maxi mum bandwidth of a system with completely monotonic (i.e., no ripples) frequency-amplitude response; and that Bessel provides a roughly analogous frequency-delay response. The whole point of my remark was that neither you nor I nor anyone else has ever shown that any particular tuning point on this continuum of possibilities is either "correct" or "the best tradeoff between amplitude and time response." (This is sheer sophistry. Do you mean, for example, that in a sealed-box woofer the '"'correctness" of Q = 2.5 is equally defensible-or its "incorrect ness" unprovable? Come on, Alan.

You've proved to the fans that you, too, know something about filter theory, now be practical. -Ed.) To make matters worse, in the reviews of loudspeakers which follow, you repeatedly chant the litany of the importance of time domain over frequency domain, and go so far as to review very highly a speaker (Cizek Model #1) which provides a switch position for either Q = 1 or Q = .6. Q = .6 is awfully close to Q = .58 (Bessel). You go on to describe the bass as “close to state of the art . . . solid, well-defined, musical," after having re plied to me that this response has "pretty grim amplitude characteristics." In addition, you maintain that for this speaker "the damping is absolutely correct," and then immediately describe the two dampings that are available at the flick of a switch. Are they both "absolutely correct"? (Yup. In a world full of sealed box woofers with a Q of 1.7 and 2 and even 2.5, either 0.6 or 1 is damn near correct. "Absolutely" may not have been the best-chosen word, but you know very well that you're just hassling us for having caught you with your trig down. -Ed.)

3. The RCA Philadelphia/Ormandy recordings (by and large) sound bad.

Max Wilcox was the producer of those recordings. The producer of a recording is responsible for its sound quality.

Therefore, Max Wilcox is responsible for a very large number of bad sounding recordings of a very fine orchestra. It doesn't particularly matter to me if he recorded them in the seventh floor men's room. They sound bad. Why? Didn't he have any influence over the choice of recording site? Is the recording site all that is wrong with the RCA Philadelphia/Ormandy recordings? (Well, you see Alan, Max really wanted to fly the orchestra to the Concertgebouw in Amsterdam, but those crass skinflints at RCA said no.-Ed.) Audio Critic #4 raises some similar questions: on page 35 you make the statement that the Q of the Ohm F is roughly 1.4, unless it is placed in the middle of the room, "in which case the Q is approximately correct." (* "Correct"? Here we go again . . .) The Q of the loudspeaker changes with room placement? I think it is about time you described to your readers (for your benefit as well as theirs) just exactly what "Q" means.

(You're a bit confused here. The Ohm F is one of the very few speakers that can be realistically analyzed as looking into a 4 pi (full-sphere) space; the "correctness" of Q = 0.707 is based on a 2 pi (half-sphere) space. The "correctness" of Q = 1.414 for a 4 pi space is therefore at least defensible. -Ed.) Regarding your use of near-field measurement, if you will refer to Don B. Keele's paper on the subject, you will see that a fundamental premise of that technique is that the speaker is a direct radiating rigid flat circular piston, etc., etc., and that an Ohm F most certainly is not! (You see?'-Ed.) The main idea of nearfield loudspeaker measurement is to allow one to measure the loud speaker's ""anechoic" response in a non anechoic environment. If you can change the measured response of a loudspeaker by relocating it in your room, your measurements are clearly not independent of the room, and therefore not successfully emulating anechoic conditions.

Impulse testing, in the full sense of the word (which means quite a bit more than just looking at pulses on a scope) reveals both time and frequency domain characteristics (which are merely different views of the same thing) rather completely. Impulse testing is in no way opposed to amplitude response measurement, as you seem to imply (is sue 3, p. 8, issue 4, p. 28). In particular, the "near perfect response" you refer to on p. 8 of issue 3 implies not only mini mum time smear, but it also indicates near perfect amplitude response. The fact that the DCM speaker shows "ragged" frequency response indicates that its impulse response isn't really "near perfect"; only that it looks that way on a scope. Don't forget that the ear is capable of detecting probably one part per million errors, while the eye at best can see one part per hundred errors on an oscilloscope display. The real power of impulse testing is that the resulting device output can be analyzed to provide extensive and precise time and frequency domain information. Journalistic imprecision with the concept of impulse testing could provoke a frequency response vs. impulse response civil war that would be just as confused as its tube vs. transistor, TIM vs. feedback, and belt drive vs. direct drive predecessors; but with a lot less technical basis.

I think your readers would enjoy a careful discussion of the time vs. frequency "problem," as well as the related measurements that are now available (3-D plots, FFT, etc.) and that this could well prevent the kind of confusion referred to above.

(Here you're absolutely right-it's the second Tuesday of the month, so it's your turn-and we're indeed planning all sorts of goodies on the subject. -Ed.) When I wrote out my check for $28 I had very high hopes for your magazine. Since that time those hopes have largely been fulfilled, and I have been quite gratified by your accurate and concise reviews, refreshingly broad and profound insight into audio, and your relatively high technical competence (who else talks about Thiele's and Small's work, for instance?). Per haps I am being too demanding in asking you to refrain from participating in the name-drop and technical mysticism cult, the new-buzz-word-a-month club, which seems to be the core of the audio scene. It seems that everyone, before he can be spoken to or about, must be labeled as a mathematical type or academician, expert or amateur, golden-ears or techno-freak. Complicating this phenomenon are the members of the Harry Pearson/James Bongiorno school of thought, who seem to believe that technical ignorance can be disguised with so much invective and name-calling.

The published critic puts himself in a very dangerous position, vis-a-vis the clearly productive artist or engineer, when he is intemperate in his (however valid) criticism of the other person's creative labors (for example, your often petty remarks inserted in Mark Davis' letters). I humbly suggest that you refrain from this kind of sniping; it is really quite irrelevant to your avowed purpose, as well as (I believe) the interests of most of your readers.

Furthermore, it casts a pall of small mindedness and egotism over what is otherwise quite high quality work. I, and most other audiophiles I know, have the greatest respect for your efforts and the resulting magazine; please don't endanger the credibility of those efforts by participating in the petty paranoia and oversimplified technical horn-blowing which seems so rampant among audiophiles. I think most of your readers are more interested in the qualities of the equipment you review than offhand remarks about the perceived competence of other audiophiles, equipment designers, and in some cases, your own sub scribers. It seems particularly a shame to dilute reader interest and confidence in this way, in light of the high quality of your equipment reviews and technical articles.

Sincerely yours, Alan S. Watkins

Burroughs Corporation; Pasadena, CA

Well, well, look who's preaching temperance all of a sudden . . . We hap pen to disagree with you, even though your basic perceptions are quite in line with ours. We feel, however, that when a charlatan launches a new product, for example, it isn't enough to tell our sub scribers that the product doesn't sound (and measure) as it should. That's much too abstract. It's extremely useful for the consumer to understand that the man is a charlatan. High-end audio is by and large a one-man-one-design kind of field, and the credibility of the designer (or maker) is actually a more important issue than the performance of a single product. We want our sub scribers to understand that from certain practitioners they can expect clear thinking and sophisticated product de sign, and from others just the opposite.

In other words, we want our subscribers to know exactly what we know, i.e., what we would tell our personal friends if we didn't have a publication.

-Ed.

And now a few words from our reviewees.

The Audio Critic:

Considering the entire report on the Dual CS 721, we would probably be well advised to forego any comment.

However, our appreciation of your customary precision and evidence of fair play prompts us to risk whatever editorial comment may in turn result. Thus emboldened, we have one quibble and two somewhat more serious observations to make.

The quibble is with your use of the word '"'except" between the statement that "We don't find anything seriously wrong with it, except that (a certain manual combination) is even better and costs $60 less." Perhaps a new sentence beginning with "However" might have been a bit more felicitous. After all, no matter how many other choices may exist, they don't make anything seriously wrong with the CS 721.

(That's a matter of perspective. It could be argued that there's something seriously wrong with a given choice when a better choice is easily available. But why quibble with a quibble? -Ed.)

Later, you express concern about the mechanism that locks the cartridge holder to the headshell because of the "possibility" (italics ours) under "'worst case conditions" (italics also ours) that there may be "play (and) lack of positional repeatability." Certainly, those end results, if they existed, would disturb a tracking-error fanatic, as you state.

Since there is a locking three point fixed position, with no play possible, we don't quite understand where you found any lack of positive positioning. (Incidentally, this entire design was awarded patent number 3,247,032, a copy of which is attached.) (Well, we found it was possible to lock the cartridge holder into several positions that differed by a hair either horizontally or vertically. It isn't a precision device-nor is a patent a guarantee of quality. Of course, we're talking about very small variances, but then it takes a shift of less than 0.2 degrees in some cases to change the sound of a record grove.-Ed.)

Our other point has to do with the anti-resonant mechanical filters, which you do acknowledge with heady praise as not a 'totally wrong-headed solution," given certain designs and price considerations. Since this device resulted from considerable Research & Development and produces demonstrable and repeatable results, we'd be interested in knowing what tests might have been made with balsa wood or silicone gunk that make them "definitely more desirable." Certainly they would be innovative, even if not commercially practical.

(Come on, Murray, you want additional R& D for $28?-Ed.) However, all things considered, we have no quarrel with your final conclusion that the CS 721 might be "best for the money if you must have an automatic." Sincerely yours, Murray I. Rosenberg General Manager United Audio Mount Vernon, NY The following response to our review of the RAM 512 power amplifier, by Mr. RAM himself, was preceded 11 days earlier by a shorter letter signed ''Peter Ledermann [now, at Soundsmith] , Electronic Engineer, RAM Audio Systems, Inc." which in our opinion fell into the mindlessly-hostile-and-uninformative category and is therefore not allowed a free soapbox here.

The Audio Critic:

In reference to your article, I would like to discuss the facts omitted, side tracked, distorted and over-biased in your review, Vol. 1, No. 4.

1. I never told you or suggested that I was the M of C/M Laboratories, on the contrary, I told you that I was not the M, but was chief design engineer of their high-fidelity products. I was, however, a founding partner of Audio International, Inc./C/M Laboratories in Connecticut.

(You never told us anything at all on the subject, but who cares? -Ed.)

2. The RAM 512 is indeed an all out design in the 180/180 watt (8 ohms) power class. We at RAM Audio Systems are dedicated to one thing and one thing only, and that is the limited production of the highest accuracy audio reproduction electronics. We follow the basic industry standards for pricing as based on raw materials, parts cost, assembly labor cost, warranty cost, sales and distribution cost, and retail dealer profit; if that is "exorbitant" then the entire electronics industry pricing structure is wrong. Remember Arrow Electronics' "we will be here when you need us"?

3. The RAM 512 is probably the strongest mechanically and best-finished power amplifier in the world. ( The world? -Ed.) The interior subchassis is formed from three (3) closed 'C' channel 18 gauge (0.049 in. thick) cold-rolled steel structures, and the rear panel that is 11-gauge (0.091 in. thick) 6061-T4 anodized aluminum. The subchassis sup ports the power transformer, the two (2) power modules, and power supply filtering and electronics. The closed box shaped structure is inherently rigid.

Extra strength is added by the two (2) 'C' channel shaped covers that are manufactured from 11-gauge (0.091 in. thick) 6061-T4 anodized aluminum. The flat, machined front panel is manufactured from 4-gauge (0.204 in. thick) 6061-T4 anodized aluminum. Hardly a "basically flimsy box." (Any 180/180-watt power amplifier we're able to lift without ruining our back is a basically flimsy box. -Ed.)

4. The power supply is a dual voltage DC power source utilizing a power transformer manufactured to our design with 4% silicon transformer steel (type M-6). The better steel allows us to have a 30% weight reduction and over 100% improvement in load voltage regulation at a 100% increase in cost over the use of ordinary power transformer steel. The dual filter electrolytics are 24,000 microfarads each and are low ESR, high-ripple-current, long-life computer grade (2,000 hr. specified life at 85° C) aluminum capacitors.

(Well, we happen to know that Dick Majestic is a good enough engineer to understand why the RAM 512 doesn't have a good enough power supply for a 81150 amplifier-not as good as the one in the Bryston 4B, for example -- and we're willing to make that statement in 9-point Times Roman italic type with I-point leading on a 13-pica measure, just in case irrelevant numerical specifications are the name of the game here. -Ed.)

5 .The RAM 512 is kept cool by four (4) anodized extruded aluminum heat radiators. Each heat radiator has 244 square inches of surface area for a total radiation surface for both channels of 976 square inches. The four heat radiators, two (2) on each side of the amplifier, are placed in a vertically enclosed rectangular box. Radiant energy from the heat radiators is also absorbed by these vertical walls, there by increasing the air flow and cooling.

These top and bottom open boxes utilize the chimney effect to increase the air velocity across the surface of the heat radiators. This cooling system is quiet and efficient, and will allow the ampli fier to be F.T.C. preconditioned at 1/3 power at 8 ohms without either of the two over-temperature protection thermo stats shutting down the amplifier. This design allows the power amplifier to run cooler, therefore longer useful life will be attained from the amplifier with out the use of a noisy fan.

(We consider this kind of cooling system, especially without a fan, to be inferior to large external heat sinks- and so do a lot of amplifier designers. But, you're right, the RAM 512 could still be a great amplifier if it were only for this.-Ed.)

6. As to your weight comparisons: the RAM 512 is mostly aluminum in construction. Aluminum is 34% lighter than its equivalent in steel. The use of better transformer steel allows further weight reduction, all at the sacrifice of increased cost; but then, filet mignon costs more per pound than ground round steak!

(Filet mignon construction but Gainesburger sound? -Ed.)

7. The RAM 512 will, with ease, produce 38 volts rms across 8 ohms at both channels and will also, with ease, produce 34.6 volts rms across 4 ohms at both channels, just as shown on the specification sheet. Typically, all 512(s) will also produce 38 volts across 4 ohms at 360 watts.

8. The square wave as reproduced by the RAM 512 is a replica of the input square wave, only deviating in slewing rate if the input square wave is slewing greater than 0.9 volts per microsecond. The RAM 512 exhibits a square wave with little ringing and no "peculiar kinks" at any power level from 10 mW to clipping, 20 Hz to 20 kHz and into reactive loads. Please note enclosed oscilloscope pictures taken on a new Tektronix 475 using a square wave source with 5 nanoseconds rise time. Note the smooth and linear rise and fall of the output signal slewing at 20 volts per microsecond, little to no ringing and without any 'peculiar kinks'.

(Most power amp designers would agree that a slew rate of 20 volts per microsecond is on the low side and may be partly responsible for the not quite-first-rate sound of the RAM 512. On the other hand, we're willing to con cede the possibility that only our particular sample put kinks in the square waves and that your pictures are of a more typical unit. -Ed.)

9. The phase shift at all frequencies between 20 Hz and 20 kHz should be no greater than 10° as stated on the spec sheet. Production 512(s) measure 7° at 20 Hz, 2° at 2 kHz, and 3° at 20 kHz. (We measured more. -Ed.)

Input AC coupling creates the low frequency phase shift while the higher frequency phase shift is caused by the input RF attenuation network. Other minor amounts of phase shift result from the slew rate limiting and the output high-frequency phase compensation net work.

I cannot change your opinion, arrived at after listening, but will say I'm sorry I didn't engineer in your preference in electronic coloration.

(What about that review in Stere-Opus? They, too, prefer electronic colorations? Everybody except you?-Ed.) 1 can only suggest to you that you try the difference signal check. All that's necessary is to sum the inverted output signal with the input signal after readjusting for the amplitude difference and look at the resulting difference signal. (Ah! But you have to look at both voltage and current differences! -Ed.)

It might prove educational to someone interested in audio electronics that most power amplifiers do not just increase the power level between their input and their output terminals. Loudspeaker loads cause some power amplifiers to do very undesirable and audible distortions when operated in the real world. Whereas the RAM 512, when operated in the same real world situation, will exhibit little or no distortions of the input signal.

Our audio industry is basically very small. We are quite friendly and co operative people, and the propagation of hate does nothing to improve our sonic goals. (Come on, Dick-you know we don't hate you. We just don't like the sound of your amplifier very much.

Your problem may be that you can't distinguish between the two.-Ed.) The consumer is just as dedicated to his purchase as we manufacturers are dedicated to the production of better audio equipment. We at RAM Audio Systems hope that what motivates you will soon be the same force that makes a manufacturer seek to improve this industry, improvement through a positive attitude.

Very truly yours, Richard A. Majestic, President RAM Audio Systems, Inc.

Danbury, CT

Where does all this leave us?

Just here: The RAM 512, at $1150, doesn't sound as good as the Bryston 4B, the Futterman H-3aa, the Electrocompaniet, the $399 Audionics CC-2 (!) and any number of other amplifiers.

We have yet to see a really favorable review of it in a noncommercial publication or meet a sophisticated audiophile who thinks highly of it. Thus the audible end result seems to bear out our own technical insights, not Dick Majestic's labored arguments.

-Ed.

The bitterest imbroglio so far, by a wide margin, concerns the Infinity QLS speaker system, as you can see from the following two letters.

The Audio Critic:

I did not answer the remarks concerning disparagement of the Watkins dual-drive woofer in earlier issues, as I do not believe anyone wins an argument with an editor in his own medium. How ever, after reading issue Number 4 of The Audio Critic, 1 feel the record should be set straight, so that owners of Infinity QLS speakers and those considering QLS will know the facts.

Concerning the uncomplimentary remarks of the American professor and the electro-acoustician: My original disclosure article appeared in the December, 1974, issue of Audio. Of over two hundred letters received, only two were unfavorable-the two referred to by you. They were interesting, as neither contained any sort of technical complaint against the principle involved in the dual-drive woofer; the writers were close friends of each other, and both letters contained insulting remarks. I assumed the writers had a good laugh with each other, and I dismissed the letters as a prank. They apparently have no technical complaint, and why do you keep their names secret? In issue 4 you spend two pages condemning the writer of an anonymous letter against you, yet you create one against me by printing the smut from their letters against me without telling who wrote them.

(For moral obtuseness, this last remark takes the prize in our entire range of editorial correspondence to date. We know who said these things about your woofer article and you know who said them. We know what was said and you know it. We know it was disparaging and you know it. We know the academic credentials behind the disparagement and you know them. There's no disagreement and no concealed information between us! We're both reluctant to name names because we both want to stay out of needless trouble. And you compare that to an anonymous poison-pen attempt to put The Audio Critic out of business by someone whose identity still remains unknown to all concerned! This from a man ostensibly dedicated to scientific logic and reason ... -Ed.)

Concerning the letter from Dr. Richard H. Small of the University of Sydney in Australia: Dr. Small was kind enough to do an analysis of the dual-drive woofer. He is one of the world's leading electro-acousticians, and it was gratifying during the early stages of development of the woofer that he found it mathematically sound, and also that he was complimentary of it from an application point of view. The "knuckle raps" as you call them, concern how to explain the principle of operation-not the performance of the dual-drive woofer. The fact that you take things out of context does not speak well for your credibility.

(Maybe you didn't fully under stand Dr. Small's letter. What about that little quip, for example, about "taking more power from the amplifier without getting more acoustic output-in fact getting less?" That wasn't about the explanation of the principle but about the principle itself. -Ed.) Concerning the 2 dB advantage you mention in issue Number 4: The 2 dB figure is from Dr. Small's analysis with a single drive network, and this is a broadband efficiency gain (for the same shape of curve). As explained in the Audio article, the gain is 5 to 6 dB at low frequency with the double drive network-that is with an additional circuit to decouple the main voice coil at and below resonance. This is the circuit used in Infinity QLS speakers.

(But Dr. Small had some misgivings about the two-network driving system, didn't he? -Ed.)

I agree that an impedance peak and an amplitude response peak are two different things, however, in the case of the dual-drive woofer, response and impedance are very closely related. You are evading the issue any way, and if you will measure an Infinity QLS, you will find that both frequency response and impedance are improved.

(Improved over what? -Ed.) You imply that the Watkins dual drive woofer has no advantage when it comes to linear throw. A more linear throw can indeed be achieved. One simply makes the voice coil winding length longer, increases magnet to bring efficiency back up, and takes care of any overdamping at resonance with the low impedance voice coil to obtain the desired Q. Certainly, as you say, the woofer in the QLS needs a lot of power to drive it-it was designed that way. However, smaller QLS speakers use a slightly different woofer and have more modest power requirements.

In any event, the proof of the worth of a device is in what it can do, and the Infinity QLS speakers using the dual-drive woofer have been marketed for over a year now. During this time there have no problems with the dual drive, and in fact I am very pleased with the acceptance these speakers have received. It offers a level of performance in a given size sealed enclosure that was previously unattainable.

With reference to conventional design, it gives more extended low frequency response, higher efficiency, or a smaller enclosure can be used. In all cases the impedance peak and the phase shift at resonance are drastically reduced, and this presents a less reactive and more resistive load to the amplifier.

You say you sell information and advice. For the price you charge you should at least offer correct information.

Sincerely, William H. Watkins Kingsport, TN

The Audio Critic never called the Watkins woofer worthless. It merely raised an eyebrow about the claims of a free lunch obtainable from such a de sign. And it reported that a distinguished professor of electroacoustics had dismissed the woofer with contempt.

What we do believe after careful consideration of the subject is that the Watkins woofer represents a rather trivially conceived design trade-off, namely higher efficiency in its upper range of frequencies than would be possible without the dual drive-all other things remaining equal-but at the cost of chewing up a great deal more amplifier power at the lower frequencies around the system resonant point. That's all the invention essentially is; there's nothing more to it, and that's what all the shouting is about.

It's possible to duplicate the frequency response, time response, efficiency, distortion characteristics, etc., of any Watkins woofer with a conventional single voice-coil woofer simply by accepting a somewhat different set of trade-offs. Big deal. How about a real invention, like a large-area electret woofer, for example? The Audio Critic:

We were naturally surprised at some of the comments about the Quantum Line Source loudspeaker made by The Audio Critic in its last issue. How ever, it would serve no purpose to debate Mr. Peter Aczel's subjective reactions, differing as they do from those of other reviewers.

(The subjective reactions published in The Audio Critic are those of at least three persons, more often of five or six, never just the Editor's. As for other reviewers, Sound Advice, for example, called the QLS "'dreadful'' and suggested that the whole idea be scrapped and started all over. We were much kinder. -Ed.)

We feel bound to write this letter because of the distressingly pejorative way in which Mr. Aczel refers to people of considerable ability, whose efforts have made real contributions toward advancing the state of the art of high fidelity. (Name one.-Ed.) For someone so willing to be pejorative, Mr. Aczel should at least be more consistently correct in the factual information he presents, and more careful in his use of technical terms. The latest issue of The Audio Critic shows carelessness and inaccuracy which does not bode well for the publication's future credibility.

For example, Mr. Aczel criticizes Infinity for naming one of its speakers the Quantum Line Source, on the grounds that there are, in fact, two parallel line sources. Mr. Aczel does not go on to explain exactly what the disadvantage of this configuration is- it is, of course, primarily that there will be some interference effects. The interesting thing is that elsewhere in the same issue (page 24) he himself calls the Rogers LS3/5A (an excellent speaker, we agree) a 'point source". It is, of course, not a point source-because there are two sources of sound which will interfere with equally deleterious effects. Else where in the same issue, the Beveridge (also an excellent speaker) is called "'a true line source" -when it cannot operate as a true line source by virtue of its dimensions. Of course, neither can the QLS. But for someone who is so frequently critical of other people's uses of language, Mr. Aczel is uncommonly careless with his own.

(We never said that the Infinity QLS is made up of two parallel line sources. You're the one who is saying it, and of course you'd have a better speaker if it were true. The fact Is, however, that the QLS is comprised of individual drivers radiating spherically in the nearfield. This generates a constructive/destructive interference pattern spread out over the sizable dimensions of the system. The end result is that in the farfield, where planar radiation exists, the sum of the energies generated by the system does not resemble that of a true-or quasi-line source. A line source radiates a uniform energy field--i.e., controlled interference-in the nearfield. Thus a line source already starts in the nearfield with controlled line radiation and not with a bunch of hemispherical radiators whose vector sum is simply not a line. The Beveridge is far ahead of the QLS in this respect. As for the Rogers, it's obvious that you're just trying to hassle us. You know as well as we do that the few inches between the two drivers of the LS3/5A4 shrink to virtually nothing--i.e., a point -from the acoustic perspective of a listener ten or more feet away. Of course, to a very short listener a very short distance from the speaker it might not appear as a point source.-Ed.)

In the review of the Rogers LS3/5A, for another example, the re viewer implies that this most excellent loudspeaker was designed by paying "special attention to the time and phase characteristics" which, in fact, it was not. Mr. H.D. Harwood, who was heading the BBC team that designed this loudspeaker, has stated several times that he totally ignores these parameters in the design of loudspeakers because he regards their effects, at best, secondary compared to other problems. The re viewer then goes on to imply that the fundamental resonance of the LS3/5A is at 130 Hz with a Q of 2. If he had measured the impedance curve, he would have found that the fundamental resonance is, in fact, around 80 Hz, corresponding closely to the -3 dB point and a Q of 0.7. This is probably why his visitors told him that the bass sounded just great-although limited in extension of course. (An indication of the disregard for time alignment in the design of this system is the position of the woofer, which is actually mounted on the back of the front baffle).

(Intentionally or not, the Rogers does reproduce pulses quite accurately, albeit somewhat off axis. That's why it sounds best from that same angle off axis. Many advancements in technology have been serendipitous. The point is, would the Rogers sound as good as it does if it couldn't reproduce a pulse at all? When it comes to the Q of 0.7 you're in deep mathematical trouble, fella. Q = 0.707 means a maximally flat--i.e., Butterworth-contour, with the amplitude response of the system 3 dB down at the resonant frequency with respect to the nominal bandpass. Maximally flat means just that: no ripples whatsoever in the response. Now, if you measure a second-order system using carefully calibrated equipment in the manner de scribed by D. B. Keele, Jr., and find that the amplitude response shows a peak of 6 dB near the system resonance, then drops to a steady level 6 dB below that peak as the higher frequencies are continuously swept-well, that just isn't a Q of 0.707, Arnie. It's a Q of 2 and that's that-period, end of story, not negotiable. Ask anyone who has studied the subject. It's perfectly true, of course, that the Q of a system can be derived by analyzing its impedance characteristics. In fact, with a low-Q system, it's the only way. But in the case of a relatively high-Q system like the Rogers, nearfield amplitude measurement will quite accurately indicate the value of Q. -Ed.)

The faith that Mr. Aczel is now placing in his pulse tests is extra ordinarily naive. The information that one can obtain from such tests is very limited and, doubtless because of other much greater imperfections in loud speakers, there is little relation to perceived quality. The Rogers speaker is one example of this-we ourselves have found a number of other examples.

The QLS is the case in point. The first few of these speakers that we made, and the original prototype, had the woofer mounted forward of the remainder of the drivers in order to produce a more ac curate pulse response. We discovered after a while, however, that whatever the pulse response benefits of this arrangement, they were masked by other coloration caused by practical mounting arrangements. We have encountered this same problem several times since--i.e., the effects of staggering drivers for time alignment can introduce worse colorations than they were designed to reduce.

(Well, let's see who some of the other "naive'' practitioners are who put their faith in pulse tests and related measurements in the time domain. Richard C. Heyser, for openers. Berman and Fincham of KEF, the 3-D response display people. Harold Beveridge, who de signed his speaker almost exclusively with the aid of time-domain tests. Are we in bad company? As Andy Rappaport points out in his letter elsewhere in this column, audio is nothing more than amplitude variations with respect to time. We all live, and hear, in the time domain. Of course, if you use time smeared program material-as your re marks on cartridge alignment below seem to indicate-then you won't hear the difference between speakers with good and bad time response.-Ed.)

We believe in scientific methods in the design of speaker systems just as much as Mr. Aczel--but at the present time there are a number of reasons why any individual measurement technique offers only a very limited guide to the overall perceived quality of the system.

The main one, of course, is that because all loudspeakers are so far from being truly accurate transducers, a consider able subjective element of design must occur in terms of making a judgment as to which particular set of imperfections is least objectionable.

Coming onto Mr. Aczel's comments on tone arms, we are surprised at his contention about the necessity for very accurate setting up of lateral tracking error. There are two separate issues, it seems to us-first, the audibility of small amounts of lateral track ing error. We have yet to hear this successfully demonstrated under controlled conditions but are willing to accept it may be audible. It is the second issue which interests us-if a cartridge is slightly misaligned, all that happens is that the position or positions on the disc at which there is zero error, change --i.e., there is now zero error on different parts of the disc than previously.

(Unless, of course, with unparalleled ingenuity the arm is arranged to move as the record is playing.) (Wow! You really blew it there. Alignment for optimum tracking geometry means, by definition, minimizing the value of a/R throughout the record--a being the lateral tracking error at any given point and R the radius at which it occurs. That the solution of this mathematical problem results in two nulls, where there's no tracking error at all, is purely incidental; the goal is to keep the peak values of «/R as low as possible. If a/R could be further minimized by not going through the nulls, that's the way it would have to be done. But the mathematical nature of the beast is such that there exist those two nulls and-here's the point-their position is fixed, as long as the maximum and minimum radii are specified. Move the nulls and you've increased the value of a/R somewhere between the maximum and minimum radii. We find it tragic that this should have to be explained to an engineer who sells tone arms for a living. -Ed.)

Then again (page 52) the writer shows his lack of knowledge in his comments about the importance of anti skating compensation. The Audio Critic should make itself aware of some work done in England on the wide variations encountered in the frictional force be tween the record and stylus (which, of course, causes the so-called skating force). It has been clearly demonstrated that the frictional force varies with a number of parameters including the stylus itself, the record material, the temperature, and the modulation of the record groove itself. These variations are not just small-they are the order of several times. It is therefore absolutely impossible to set bias force accurately for anything other than one particular and very limited set of conditions-because the force actually varies while any given record is being played.

(You're wrong. With a given stylus, a given tone arm and a given vertical tracking force, the antiskating compensation can be set permanently for all practical purposes. We shall go into the theory behind this in a future issue; this parenthesis is hardly the place for a treatise. In any event, the simple test we describe on page 55 of our Number 4 issue will tell you whether or not the compensation is correctly set.-Ed.)

Then again the magazine does not seem to be aware of the benefits of low mass, judging from its comparison of the Grace G-707 with the Infinity Black Widow. Of course, the Denon cartridge that you were using is not a very com pliant one. More modern cartridges need the very lowest-mass arm if they are to work correctly-some of these cartridges are even too heavy for their own compliance and theoretically, therefore, would need an arm of negative mass.

This aspect seems to have been completely ignored. We also need to comment about the review of the Black Widow arm per se--the bearings are not "prone to jitter" under actual conditions of use. Indeed, the geometry and the forces involved prohibit it.

(It just so happens that the best of today's cartridges are not the ultra compliant ones and therefore have no need for an ultralow-mass arm. See the cartridge reviews in this issue. As for the bearings of the 'Black Widow' arm, if they aren't tight this way, they aren't tight that way. No amount of sophistry will make them as positively located and as jitter-free as those of the Grace G-707. But then a little shift here and there in the stylus-to-groove relationship is of no importance, is it Arnie' -Ed.)

More facts. Mr. Aczel mentions in the QLS review that the midrange dome used in the QLS is a Peerless driver.

In fact, it is not, and originates in Ger many from another member of the group, MB Microphonbau.

(So it's a Dodge instead of a Plymouth. Sorry about that, but it's hard to tell those Chrysler products apart. -Ed.)

The holes in the dome do not act like a Helmholtz resonator-they are much too small to do that effectively. The holes improve the transient performance of this driver, at the expense of low-frequency extension. Mr. Aczel then goes on to say that Peerless also supplies the 5" midbass driver-it does not, as a matter of fact. These issues in themselves are not, of course, particularly important, but they do indicate the haste and eager ness of Mr. Aczel to speak out without checking his facts first.

(If it isn't the Helmholtz principle that makes the holes increase the output of the dome, pray tell what it is? The California sun shining through them? Obviously, energy is being derived from the presence of these holes punched in the diaphragm, suggesting rather strongly a harnessed resonance used to reinforce the output from the front surface of the dome. The dome is vented, so to speak, to increase output over a narrow band of frequencies, and this method most certainly relies on Helmholtz resonator behavior. The claimed improvement in transient response flies in the face of science, we're afraid.

Given a system defined by a transfer function, increasing the complexity of the system-all other things being equal--will degrade rather than improve transient behavior. In this case, the mid range dome is essentially a second-order device whose behavior can be mathematically specified by a second-order transfer function. This transfer function specifies both the magnitude and phase of the system. Increasing the complexity of the system by punching holes in the diaphragm raises the order of the system from second to third or fourth.

Even if the characteristics of the system are preserved-say, by maintaining maximally flat behavior-the phase response is substantially degraded, suggesting in the time domain a poorer, rather than improved, transient response. The Audio Critic's tests, however, indicate a severely peaked output where the holes be come effective, and that means a substantially worsened transient performance. As for the 5 midbass driver, whether Plymouth or Dodge, it rings like an alarm clock and should be pulled out of the system and stomped on.-Ed.) As to the reviewer's subjective conclusions on the sound quality of the QLS, we cannot usefully comment, except to say that insufficient attention may have been given to setting up the system, and to the associated components. It certainly is surprising that his findings are so widely at variance with those of other independent reviewers, and this would seem to indicate that something else was wrong. His comments about the stereo image, for ex ample, could result from the use of a tone arm of too high mass.

(Okay. Name one audio professional-just one-who has visited The Audio Critic's laboratory and listening room lately, and then felt that better procedures and cleaner program sources were available elsewhere.-Ed.)

We find Mr. Aczel's gyrations on the Watkins woofer to be quite amazing.

Having first dismissed it completely in an earlier issue, in his review of the QLS Mr. Aczel attempts to backtrack. Having originally condemned the Watkins de sign out of hand altogether, he now ad mits, as Dr. Small has said, that the concept does indeed allow an improve ment in low frequency response. But he is determined it appears not to lose face and so has attempted to denigrate the design. An improvement of 2 dB in output, other things being equal, is in deed far from negligible. It means that a closed box system, with its advantages over vented systems of high mechanical impedance at sub-audio frequencies, and a gentler cutoff slope, can nevertheless be directly competitive with such vented systems in terms of efficiency. More over, the extra output is obtained and the damping is increased at the same time. We should perhaps quote directly from Dr. Smalls letter to William Watkins, to which Mr. Aczel refers, ". . . my congratulations to you on devising a significant improvement for loudspeaker design." Hardly "condescending"!

(Yup. Dr. Small is a nice guy and didn't want William Watkins to feel that the weaknesses pointed out in his design made it totally worthless--which, incidentally, we never said, either. We refuse, however, to pursue the subject of the Watkins woofer any further than our reply to Watkins's own letter above. That's how we see the whole thing and that's all, folks. So, from here on down, we'll just let Arnie noodle with his favorite theme to his heart's content and with impunity.-Ed.)

We agree the Watkins design does not have any advantage in terms of power handling ability over an ordinary sealed box, other things being equal. We never claimed it did. Of course, reflex and 'transmission line' enclosures make use of an acoustic resonance to rein force the bass-thereby increasing the power ability, other things being equal--but these designs have another very much more serious disadvantage, which is that the mechanical impedance below the audio band is very low, allowing the cone to make large displacements. In practical circumstances, this causes far more distortion than a greater cone movement within the audio band. We are surprised that Mr. Aczel managed to overload the QLS woofer-because this is not a problem that we have come across. It is possible that he clipped his amplifier, because the acoustic effects are sometimes difficult to distinguish.

We agree with Mr. Aczel (and now come to that Dr. Small letter) that the Watkins principle is difficult to explain--what Mr. Aczel called "professorial knuckle-raps” in Dr. Smalls letter are, in fact, connected with Mr. Watkins' explanation of the principle-not the principle itself. Indeed, Dr. Small himself says, "unfortunately I cannot offer any simple cure-all suggestions for presenting a better explanation." If Mr. Aczel has any suggestions as to how we may explain the principle more simply, we would be delighted to hear of them. However, he should recognize that there is a difference between the validity of a design principle, and the ease with which it can be explained in simple technical terms. He is putting up a smoke screen to conceal his own lack of credibility by debating the method of explanation-the fact is that William Watkins indeed has a useful and valid contribution to make and this is at direct variance with Mr. Aczel's original comment, which, as usual, was as hastily prepared and inadequately researched as it was pejoratively stated.

Yours sincerely, Arnold Nudell. President Infinity Systems, Inc.

Canoga Park, CA

Our subscribers may be wondering why we reproduced this letter in its petulantly long-winded entirety and why we made an attempt to refute it point by point. Well, for one thing, we want the audio community to under stand once and for all that a critiqued manufacturer does have the opportunity to argue with us in this column, even though we don't solicit such an argument before publishing the review. At the same time, the letter has some consumerist value, since Infinity is a rather heavily promoted audiophile brand and the consumer has a right to know just what kind of technical thinking and preparedness is behind those products.

Since the chief technologist of the company is President Arnie Nudell himself, you can form your own conclusions from the above. -Ed.

---------

[adapted from TAC]

---------

Also see:

The Credibility Crisis in Equipment Reviewing

Various audio and high-fidelity magazines