Art Dudley Listening

Purely electronic distortions are a different matter altogether, and here I'll make an unscientific observation: The simpler the device, the more benign the distortion. The electronic distortion products added to a music signal by a simple tube amplifier tend overwhelmingly to be the second harmonics of the affected frequencies. Add global feedback to the circuit and, despite a reduction in the amount of second-harmonic products, the far more objectionable odd-order harmonic products are not only left behind like audible driftwood but can actually increase in amplitude. Change the active devices from tubes to transistors and the problem gets even worse. Nice.

Reorganized on the noise floor
Why do harmonic products become more objectionable as they rise in order? (footnote 3) To begin, consider that the first harmonic (1xF, where F stands for frequency) is the fundamental itself. The second harmonic (2xF) is the octave of the fundamental, and the third harmonic (3xF) is the octave-plus-a-fifth of the fundamental. In other words, if the pure tone being amplified is A=440Hz, the wonky amplifier would superimpose over that note an E.

But I just told a pretty lie: That third harmonic is measured as 1320Hz, yet the musical fifth of A=440Hz is actually 1318.5Hz. That's because, in well-tempered Western music, an octave is divided into a 12-tone chromatic scale (footnote 4), the 12 notes of which are not precisely in tune with the harmonic series. The third harmonic added by an amplifier to a fundamental will thus be very slightly off—a problem that worsens the higher one goes (again, except for the octaves). If you think third-harmonic distortion is a bitch, just get a load of seventh-harmonic distortion—which, for A=440Hz, is a 3080Hz tone. Your perfect A is now overlaid with a tone partway between F# and G. Ouch.

At the opposite end of the spectrum from ouch is second-harmonic distortion: perhaps not a desirable commodity per se, yet one that is far more agreeable—something you can confirm for yourself with a copy of Stereophile's Test CD 2 (STPH004-2), which this magazine released in 1992. In addition to music performed by Robert Silverman, Corey Greenberg, Peter J. Walker, and John Atkinson, it includes a series of test tones, offered both in undistorted form and overlaid with varying types and amounts of harmonic byproducts.

When I play my copy of Test CD 2, I begin with the worst: a 500Hz tone—slightly higher in pitch than the B above middle C—heard without distortion for 10 seconds, and with 10% seventh-harmonic distortion for another 10 seconds. It borders on the intolerable, the combined note and harmonic product sounding sour, keening, just plain wrong. Remarkably, in listening to the same tone combined with 3%, 1%, and even 0.3% of seventh-harmonic distortion (fig.1), the experience is nearly as unpleasant, even though a reduction in the relative level of distortion is apparent. Only at the lowest level on the disc—0.1%—am I untroubled by seventh-harmonic distortion.

My experiences with the CD's demonstrations of third-harmonic distortion are less overtly unpleasant but still not good. I'm distressed by the mildly sour sound of the samples of 10% and 3% third-harmonic distortion. At 1%, I can identify its presence but don't find it too terribly distracting. At 0.3%, I'm not sure if I can actually hear the added third harmonic, and at 0.1% I'm certain I can't.

May God have mercy on my soul: At 10% of the signal—a full –20dB with respect to the 500Hz test tone (fig.2)—second-harmonic distortion is clearly audible, but not really objectionable: not sour, not keening, not distressing, not unmusical. The character of the tone has been altered, like a woman's hairstyle after too much teasing: The sound is thicker, and the tone seems spatially larger, though not more substantial. At a level of 3%, second-harmonic distortion is thoroughly fine with me—feel free to quote me on that (footnote 5)—and I find 1% and smaller amounts of second-harmonic distortion difficult to detect. Second-harmonic distortion is like sugar on berries: The additive is clearly consonant with the host, though one wouldn't want to go overboard. I suppose.

Recently, I listened to those distortion tracks over and over again. As I did, I wondered: When we listen to live music, to what extent do our surroundings resonate and add their own musically consonant overtones? Do the best halls add second harmonics to some or all of the tones played therein? Do we ever hear live music without hearing with it a certain amount of naturally occurring distortion?

Perhaps blind testing could tell us. (Yes, that's a joke.)

Bonding with my Pepys
I believe that the essence of music is always satisfying. I believe that the human heart and mind can always be counted on to be soothed, or excited, or delighted, or challenged, or mystified, or enraptured by combinations of melody, harmony, rhythm, and the sounds and meanings of the words that sometimes accompany them.

But in real life, the sound of music isn't always pleasing. Sometimes it's decidedly coarse. I'm not just speaking of the distress caused by proximity to a too-loud electric band (although, as John Atkinson and I were reminded at a recent King Crimson reunion concert, that is one aspect of it)—disagreeable qualities are sometimes associated with instruments and voices we otherwise consider better mannered. I've attended countless chamber-music events, and I'm always reminded of what many of you already know: just how aggressive a violin can sound if one sits too close. And in that regard, the violin has nothing on its friend the flute, the influence of which has surely been documented in any number of violent crimes. Trumpets can blare and glare, trombones can shout, cymbals can pierce—and don't get me started on the subject of the human voice (footnote 6). I believe that J. Gordon Holt, may God exalt his name, has said pretty much the same thing. At least about flutes.

Is it the job of a good hi-fi to sound bad? Yes, but only in the same manner, to the same degree, and at the same times as do real instruments and voices themselves.

One seldom hears a domestic playback system that sounds thoroughly, unambiguously convincing in that regard. Then again, who among us really trusts the combination of his or her system, recordings, and senses when it comes to this sort of thing? Who isn't dismayed at the thought of getting it wrong—of finding that they've taken pleasure in something they thought was natural, yet proved to be . . . well, something else? [cue theme from The Crying Game]

The hardness that sometimes accompanies the sound of a flute in real life, and the hardness with which poor or overtaxed gear endows recordings of same, are not entirely alike. So we listen carefully and learn the differences, and we do our best to report those distinctions when we hear them. Other things are more ambiguous, such as knowing if a harsh vocal peak is the fault of an amp or loudspeaker or phono cartridge or worn LP—or if, in the case of a recording with which we're not thoroughly familiar, it sounds that way because the recording itself is too hot.

So we measure—a perfectly reasonable thing to do. And the results are perfectly reasonable things to trust, up to a point.

In the February 2012 Stereophile I wrote about the then-most-recent version of Shindo Laboratory's Haut-Brion power amplifier ($11,000): a low-power, push-pull tube amp built by hand in small production runs from a mixture of new and vintage parts. I described the review sample, which I purchased several months later, as "the pizzicato king," a reference to its uncanny ability to portray the component of force in recordings of plucked strings. I also noted the Haut-Brion's well-detailed sound, its enormous soundfield—and, on the other side of the coin, its somewhat overly big bass.

Recently, my more technically gifted neighbor Neal Newman invited me to come by with the Haut-Brion—named after the Bordeaux said to have been the favorite of the English diarist Samuel Pepys—to see how it performed on his test bench. The results held few real surprises: A small midbass boost was apparent in its 50Hz squarewave; the Haut's very good way with subtle detail might have been explained, in part, by its surprisingly low level of diode-related switching noise; and 60Hz hum, too, was very low.

Most interesting of all was a combination of two other measurements: The sensitivity of the three-stage Haut-Brion, which lacks global feedback and exhibits in use a very generous amount of gain, was 650mV for a 500Hz tone just before clipping. Also generous, at least by contemporary standards, was the amount of harmonic distortion present in the Haut's output: 0.8% at 1W, 2% at 5W, and 3% at 10W. In all of those measurements, the second harmonic predominated by a considerable margin, with fourth-harmonic distortion a very distant second; odd-order harmonics were all but unseeable on Neal's 'scope.

It's obvious that the amp's designer, the late Ken Shindo, made a conscious choice: As with so many other high-gain tube amps, the Haut-Brion delivers, in a companionable system, an excellent sense of musical drive—a quality I value highly, and one that, in my experience, is always missing from systems in which passive rather than active line stages are used. Yet, all other things being equal, additional gain stages bring to an amplifier the likelihood of additional harmonic distortion. Much of that distortion could be canceled out by the use of global feedback, but global feedback carries not only the taint of increased odd-order harmonic products: My experience suggests that it often makes for an amp that sounds small and constricted, lacking in both scale and the ability to let the music breathe.

The Haut-Brion also allows the recorded sounds of stringed instruments a generous helping of texture, more so than even most tube amps of my experience. Could 0.8% harmonic distortion at 1W—which is probably close to the amp's cruising altitude in my system, with my very efficient Altec Valencia speakers—be responsible for some of that texture? I think it's somewhat unlikely, though I can't deny the possibility.

A somewhat more interesting question: Considering that this amplifier's level of second-harmonic distortion rises with power output, and that the transient impulses that define the leading edges of notes tend to stress an amplifier's output section more severely than the sustain and decay components of those notes, is it a stretch to wonder if those second-harmonic products serve to highlight or even exaggerate the tactile qualities of plucked notes and other such sounds?

I admit that the question concerns me more as a reviewer than as a listener. The reviewer in me wants to make things as fair as possible for every product, and as clear and relevant as possible for every reader; the listener in me wishes for maximum pleasure and thus the highest fidelity in accordance with the performance characteristics that matter most to me. And it is my opinion that, as a listener, it is far easier to mentally compensate for colorations or added thickness than for lifelessness and lack of dynamic impact.

I still see the job of a perfectionist-quality playback system as getting the hell out of the way of the music. Yes, the style of gear I favor does add a little something to the mix, but it isn't much, and it is musically consonant. Gear that is my kind of good achieves its best results by not limiting the music's force or impact or sense of drive by even the slightest degree. Were I to change course and adopt a system with, say, powerful solid-state amplifiers, two or more pairs of speakers with complex crossover networks, active room-correction software, and a three-box digital front end, I would surely enjoy reductions in harmonic distortion and coloration, an improvement in dispersion, and better resolution of the spatial information in stereo and multichannel recordings. But I suspect that the qualities of recorded music that I most value would not get past all that stuff.

Though they haven't always been, my gear-buying decisions are now governed by the listener in me. Life is brief. If I'm to pay luxury prices for a playback system, I demand the sort of performance in which I can actually luxuriate. And if I wait till I retire, I won't be able to afford—and perhaps won't be able to even hear—the sort of system I want.

Which is why I have it now. The listener in me seeks the most enjoyment my money can buy: a policy I would force on all, regardless of their tastes in records or gear—and for that, the reviewer in me seeks the reader's forbearance.

---

Footnote 3: Except for the series of octaves of the fundamental, in which every tone is a perfect doubling of the one an octave below it.

Footnote 4: John Marks offers a primer on this subject here.—Ed.

Footnote 5: Sad experience tells me that, within a day of this issue's publication, I'll get the Abner Louima treatment from some loser on the Internet who has neither read this column nor even intends to, yet has been told, by a friend of a friend, that Art Dudley admits to loving distortion. God bless the simple.

Footnote 6: A noted audio expert recently observed in print that the human voice has "a limited dynamic range." For that individual, hell will be a front-row seat at a never-ending concert of amateur choral singers.