There has been, I feel, considerable progress in amplifier design in the 57 months since Dr. Shanefield's forlorn letter. But, to judge by Ken Kessler's "Industry Update" in this issue, things haven't changed. Ken reports the publicly expressed views on reviewers of Ross Walker, Chief Executive of Quad Electroacoustics in the UK, manufacturers of electronic products that, to an extent, are designed down in quality rather than up. By this I mean that, when it is pointed out that a Quad product could be better in any specific area, the response from the company is that their customers don't need preamps with lower noise floors or better separation, speakers that can play louder, or amplifiers that can source more current into low-impedance speaker loads. And that to make their products capable of doing these things means unnecessarily high prices. In other words, there is a limiting performance below which improvements are not audible; it would be immoral to ask customers to pay for performance they will not benefit from.
This is a free country, and no criticism should be attached to a company like Quad for so straightforwardly making known its philosophy (see also "Manufacturers' Comments" in this issue). Indeed, in my experience, Quad products have good aesthetics, are well-made, have an excellent reputation for reliability (following a recent series of equipment breakdowns and failures in Santa Fe, I wish a larger proportion of high-end companies paid a little more attention to quality control), and if, in this observer's opinion, their electronics could sound better, they do sound excellent at the asking price. They just don't sound as good as I would want in my own system. Which doesn't bother me: I can buy something else.
What does bother me is when a commentator tries to enforce such a restrictive view as the law, as is the case with a well-written article in the July 1988 issue of Electronics & Wireless World magazine, entitled "Science & Subjectivism in Audio Engineering" (footnote 1). In it, reprinted in expanded form on-line here, Douglas Self, a British engineer working in the pro audio industry, attempts to lay waste to the whole field of high-end audio. He makes what on the face of it is a very effective case that we are all self-deluders or charlatans and that, if not quite crooks, engineers working in high-end audio have dubious moral standards: "the man who pays thousands of pounds [Sterling] for a preamp can only be described as a Hi-Fi victim...it is surely a morally ambiguous position to persuade nontechnical people that to get a really decent sound they have to buy $l2000 preamps and so on, when all technical orthodoxy indicates that this is quite unnecessary."
Using the design of amplifiers as the field of combat, he argues that in a competently designed amplifier, capacitors, cables, the quality of the power supply and its connections, whether or not a circuit is hard-wired or laid on a printed circuit board, whether a circuit has low or high levels of overall negative feedback, whether an amplifier is of dual-mono construction or not, all will be irrelevant to sound quality as they produce no effects that will show in either traditional tests of, for example, total harmonic distortion, or in a nulling test such as the one proposed by David Hafler (footnote 2).
Dismissing the preference that many have for tube amplifiers as a taste for added second-harmonic distortion, Mr. Self concludes that subjective reviewing is therefore largely a matter of hearing what is expected to be heard, that "more expensive" will always, therefore, equate with "better," and that it persists for the same reason that parapsychology does: that "there will always be people who will believe what they want to believe rather than what the hard facts indicate."
Hard facts are, of course, what are produced by the likes of Mr. Self, who mentions in his report that "most competent engineers with a taste for objectivity probably have better things to do than dispute every tendentious report." But as a writer of tendentious reports, I ask where is the objective data that proves that people who hear differences between amplifiers—and we do, don't we?—are suffering from self-delusion? A look at Mr. Self's published writings in the last few years indicates that he relies heavily on steady-state tests, based on the assumption that the signal waveform is a pure sinewave, infinitely extended in time. But, other than the fact that they are easy to carry out, why should such measurements alone be the arbiter of quality? (footnote 3)
In 1982, HFN/RR presented its Audio Award to Peter Walker, designer of the Quad electrostatic loudspeakers, father of the Ross mentioned above, and a man who possesses the rare ability to reduce a problem to a succinct expression of the essentials (footnote 4). During the course of an interview John Crabbe and I carried out with Peter, published in the July '82 issue of HFN/RR, he suggested to me that it was perfectly possible to design an electronic component that would pass every steady-state test of goodness—it would have suitably low levels of harmonic and intermodulation distortion, its frequency response would be flat, it would pass a squarewave intact (and on pure tones, it would even pass the Hafler nulling test with flying colors)—yet its effect on a music signal would be immediately noticeable, even objectionable.
This mystery component would be an amplifier whose gain varied with signal level; in other words, a compressor or expander. A steady-state measurement using a repetitive waveform allows the unit to stabilize its gain and it acts as any other "perfect" amplifier. On music, however, you hear the aberration in its response immediately. It was Naim Audio's Julian Vereker who suggested to me a decade ago that, in his opinion, many of the subjective shortcomings of amplifiers which nevertheless "measure well" are related to this kind of behavior. He suggested that an amplifier which gave 0.1% THD at 1W output and at 100W output would sound very much better than one which featured 0.0001% THD at 1W and 0.01% at 100W. The first amplifier's transfer function remains constant with level; the second, despite the low levels of distortion in absolute terms, is actually 100 times worse at the high level, indicating that the circuit is inherently behaving more like a compressor than an amplifier, with only the corrective nature of negative feedback allowing it to appear to work with continuous tones. But with music, the ear knows!
In a series of "Comment" articles some years back in HFN/RR, I argued that as the reproduction of music was a multi-dimensioned process, to reduce it to two for ease of testing, as in a traditional steady-state test, was perhaps to throw the baby out with the bathwater. (In a steady-state test, the way in which the signal changes with time is removed; yet music consists of that very information.) Of course, I am sure that Mr. Self would argue that such methodological reductionism is the very basis of Scientific Method: this analytical engine—where an observer puts forward a hypothesis concerning the nature of reality, then organizes appropriate experiments set up to test the validity of the hypothesis, the results of which provide the framework for further speculation—has served as the trustworthy bedrock of all progress since the time of Descartes.
Unfortunately, it can happen that scientific objectivity nevertheless leads to erroneous conclusions. First, although measurements will always have some connection with sound quality, the connection can often be tenuous; when those measurements are interpreted, the subjective nature of interpretation can render an "objective" test rather less so. Second, you will never be sure that, by minimizing the variables under test and by making the test procedures practicable, you have not changed the situation so much that the results will not be applicable to real life. It is on this small rock, for example, that "scientific" listening tests of the kind demanded by Stereo Review readers, the followers of Julian Hirsch and David Clark, often founder. To make the test performable, it can become too far removed from the real experience.
Third, consider an imaginary scientist who decided that the subjective impression of "color" correlated to spectral purity. He would therefore publish work along the lines of Mr. Self's, proving to his satisfaction that there was no such color as brown, there being no single frequency of light that corresponded to what people mean by "brown." You can imagine his anger as people obstinately continued to refer to things being colored brown, and even made value judgments as though such a color had a real existence. He had "proved" its nonexistence—doubtless, he wrote articles for all the right journals—yet why wasn't anyone taking any notice?
The answer, of course, as in the case of Mr. Self and all such naysayers, is that their initial world view is too limited. The hypotheses that they set up fail to contain the spectrum of human experience. Even if traditional measurements did imply that all amplifiers sounded the same, but most people report that the opposite is in fact the case, then—assuming that a conspiracy theory along the lines of "manufacturers and press consciously or unconsciously are hoodwinking the public" is as valueless as all other conspiracy theories—the only conclusion to draw is that the men behind the instruments may be looking in the wrong direction. It reminds me of the old story of the drunk looking for his keys under a street lamp. A passerby joins in the search, and after a fruitless few minutes, asks where the drunk has dropped them. "Over in the bushes," answers the drunk, "but it's too dark to look there."
Footnote 1: A similar but less thorough article by the same author appeared in the August 1988 issue of Hi-Fi News & Record Review.
Footnote 2: See Stereophile Vol.10 No.1, pp.85, 100, & 189; Vol.10 No.3, p.19; and Vol.10 No.5, p.25. While David Hafler correctly points out that his test will reveal every distortion and change effected on a music signal by an amplifier, interpretation of the results is fraught with problems, as linear errors due to phase shift at the frequency extremes with all amplifiers—except Hafler's own XL-280—conceal those due to nonlinear mechanisms. If you add circuitry to remove the masking linear effects, as with the Baxandall nulling test mentioned by Douglas Self, then how can you be sure that this is also not obscuring the very errors you seek?
Footnote 3: To be fair to Douglas Self, he does suggest that the intermodulation test suggested by R.A. Belcher of the BBC, first described in a BBC monograph and then in Wireless World, May 1978, pp.36–41, which tests the audio path at all frequencies at once, may be more effective at revealing differences between amplifiers. (My thanks to James Boyk for first pointing me toward this test.)
Footnote 4: To give you an example, when talking about amplifier design, Peter expressed the opinion that it was all "Ohm's Law and common sense," something that both has stuck in my mind ever since and has proved to be true.