James Boyk: All-Tube Analog Page 2

Then here's the game. The student takes the one that's less good and he gets to do anything to it to improve its performance. He wants to put in better transistors, fine. He wants to put in feedback, fine. He wants to take feedback away, fine. He wants to change the capacitors, the resistors, he has absolute carte blanche. Then he iterates. The intention is to leapfrog—only the problem is, there's never been any leapfrog; in three or four iterations, the tube unit has always come out on top. The students have tried various things on the transistor unit, but it simply doesn't begin to come close. (I want to put a caveat on this: these undergraduates are very bright and able, but, with the rare exception, completely inexperienced in audio. We do not purport to be telling something about what's happening at the state-of-the-art. The point is to give students a hands-on feel for the differences between tubes and transistors.)

Atkinson: Does the tube unit sound like the straight wire?

Boyk: No. But it's more like the straight wire than any of the transistor line amps. The thing about tubes generally is not that there's no difference in the sound, but the differences that the tubes introduce don't interfere with the semantics of the music. They don't get in the way of the musical content; the transistor unit disturbs the beauty of the music much much more, it just becomes uglier.

Atkinson: Is the way the harmonics of piano tone change with time the kind of thing you feel to be less well preserved through a transistor amplifier?

Boyk: Absolutely. Let me show you how it matters. The thing which characterizes the Steinway is that it's got a rich harmonic structure that is very alive; which harmonics are prominent changes from moment to moment. Most commercial piano recordings don't have much of this information, it gets lost somewhere. At the end of the first page of Chopin's Etude Op.10, No.3, there's a phrase that ends on a long E; the next phrase begins with an E an octave higher. Now the way I like to play that, I want the new phrase to appear out of the ashes of the old; I want the first note of the new phrase to appear to steal in, to appear as though it's the second harmonic of the previous note merely being emphasized. I don't want to hear the attack on the note, therefore I listen for the second harmonic development in the old note and try to begin the new note when the second harmonic is strongest (within the time window we can still speak of as being "in time").

Atkinson: And this kind of inner detail is passed best by the tube line amplifier in your test.

Boyk: Yes, I think so. Tubes generally are better at that kind of thing. And ribbon microphones are better than condensers. I am told that the fundamental resonances of condenser diaphragms—which are almost always disc-shaped, like little timpani, typically between half-an-inch and an inch in diameter—are right in the middle of the audio band. Basically you have a microphone operating in two different regimes: in the low end it operates below its resonance, and in the high end it operates above it.

A ribbon, however, has a fundamental resonance at the bottom of the audio band, and therefore operates in one regime over the whole band. I believe also that most condensers have all kinds of high-end resonances intentionally put in to goose up their frequency response curves so that they will look flat to 20k. Of course, then when they do go, they go fast! Whereas a ribbon microphone is much freer from such intentional playing around. It certainly sounds it: you get a natural detail that I never hear from any condenser mike.

Atkinson: I feel the same about the classic Coles/BBC 4038 ribbon, which I believe was used for the Sheffield Firebird recording. It is astonishingly neutral, its only problem seeming to be that, like the B&O, it doesn't have an extended enough high frequency range.

Boyk: Are my records rolled off in the highs? To me they sound musically natural: they may not the brightest things around, but I don't have any sense of missing information. There's all kinds of artificial brightness in the extreme high end in condensers—what Steve Haselton of Sheffield calls "condenser cackle"—which is simply absent in ribbons. It can take a lot of getting used to for people not directly familiar with the sound of musical instruments. I think the reason some people find ribbons dull, when compared with the condensers, is the lack of cackle rather than their actual gentle HF rolloff.

Atkinson: It has been said that, with the advent of digital recording, these fundamental problems with microphone behaviour will be much more important, because the digital process itself is much more transparent.

Boyk: The digital process as it now stands is a big disaster. I don't want to lean on it too heavily, but our double-blind test at Caltech [in which the digitally-processed sound was compared with the live microphone feed, to its detriment] was really a very careful test; as careful as we knew how to make it. We never formally published a paper on the test and its results, but I did give two talks at Caltech, and one out at Bolt, Beranek & Newman, the acoustical consultants, which were complete in every detail. I told exactly what we had done, and nobody has yet pointed out any methodological flaws.

I've also done a lot of listening over the years at Sheffield Lab recording sessions where there have been various digital machines on line. The polarities were correct and the levels matched, so we really had a kosher playback. I've also co-engineered three records with Doug Sax for Sheffield Lab in the last year and a half, including the Kodo drum CD and the Los Angeles Philharmonic Firebird recording with Erich Leinsdorf, all of which were also recorded digitally. I've had a lot of listening experience under what I would call "valid" listening conditions, where the comparison you are listening to is a comparison of what you think it is. Any scientific test must have valid listening conditions as a prerequisite, otherwise what you're doing is meaningless. Unfortunately, in my opinion most tests of digital—and most tests of audio equipment—are done under meaningless conditions.

The comparison of Sheffield Lab LPs with CDs is almost a scientifically valid comparison—there's one little ringer with the LPs, but if you do the CD versus the Nakamichi cassette then it's an absolutely fair comparison, the CD and cassette are symmetrical. I've played the LP/CD comparison for roughly 30 people over the last two or three years, and without exception—I don't think I'm a mesmerizer—everyone has said "Oh! The difference is day or night."

Atkinson: But a digital advocate, faced with the negative results from your blind insertion test, would say, "Yes, it's not surprising that you can hear the insertion of a PCM processor into the direct feed; nothing made by a human being is perfect. But surely what is important is the degree of that difference heard when compared with, say, that introduced by a professional-quality analog machine when it is inserted into the feed?"