Article as first appeared in Electronic Engineering Times magazine
Monday July, 8, 1996 Issue 909
As process technology marches inexorably forward, DRAMs and microprocessors are ramping in 0.35-micron CMOS. ASICs and FPGAs are moving into various 0.6- to 0.5-micron processes. Gallium arsenide is consolidating its hold on 2-GHz communications. And the latest thing in high-end audio design is----vacuum tubes? Paradoxical as it might seem, those mammoth, glowing artifacts of the days of Lee DeForest are once again a---ahem---hot item. Collectors scour garage sales, paying hundreds of dollars for the rusting hulks of consumer amps that have lain in attics for decades. Sleuths seek out unused inventories of tubes left over from shuttered repair shops, abandoned hobbies, even military-surplus stores. But the tube movement is far from just a historical retrospective. Designers are turning out tube amplifier designs in numbers not seen since the 1950s. Prices range from thousands of dollars to tens of thousands for high-powered, single-channel power amps. The designs are fueling a demand for new tubes-a critical problem in the United States, where the last tube manufacturers shut down decades ago. So distributors are courting tube makers in Eastern Europe, the former Soviet Union and the People's Republic of China to supply copies of classic vacuum-tube types. Rare, unused original tubes sell for more than the equipment for which they were built.
Bowing before the inevitable, Western Electric has restarted manufacture of one of the oldest and most beloved--or despised, if you had to design with it--tubes, the 300B triode.
Pursuing the Inevitable
What is going on here? A generation ago audio designers fled the power, heat, reliability, distortion and noise problems of tubes to embrace transistors. Perhaps that first hug for germanium power transistors wasn't a long-lasting clinch, but soon modern hi-fi equipment was, by definition, transistor equipment. "Solid state" became a code phrase for quality. Transistorized preamps reached levels of distortion and signal-to-noise ratios (S/N) unimaginable in the tube days. And amps' power levels soared.
But somehow a quiet, contrarian voice persisted. Some audiophiles kept their beloved vacuum-tube gear, for sentimental or financial reasons, or just because they insisted, in the face of evidence to the contrary, that the tube stuff sounded better. Over time, the voices grew in number. And just as the pursuit of power for power's sake was reaching its screeching peak in the 1980s, some serious and respected sound critics began to say that the solid-state emperor had very skimpy underwear. Today, the cream of the audiophile community--those fortunate few who can afford to spend the price of a Jaguar for a stereo--take it for granted that an expensive line stage, power amp or even digital-to-analog converter will use vacuum tubes, not transistors. A Usenet group, rec.audio.tubes, is devoted to the faith. There is even, according to the World Wide Web, an Oregon Triode Society dedicated to the appreciation of simple amplifiers made from triode tubes.
The informed skeptic might protest that even average solid-state amps are measurably superior to the best tube equipment, but that argument may not hold water. It depends on where you stand in the debate over how to measure an amplifier.
"During the 1960s and '70s, when solid state took over in the United States, our high-fidelity magazines, unlike those in the rest of the world, bought into the concept that if amplifiers measured the same on bench tests, they would sound the same," said Mike Elliot, president of Counterpoint Electronic Systems Inc. (Carlsbad, Calif.). "It was kind of like reviewing wines by mass spectrometry."
Lost, say many designers, was the understanding that bench tests such as frequency response, total harmonic distortion, intermodulation distortion and S/N were never more than point tools in the design process. Producing a great-sounding amp for a given set of speakers meant an exhaustive, iterative process of listening to music, making subjective observations, doing bench measurements to try to confirm the subjective data and then tweaking the design. Until the great stereo marketing push of the 1970s, designing hi-fi hadn't meant designing to the numbers.
The real goal
In conceptual terms, the problem for an audio amp-and-speaker combo is to reproduce the pressure wave that struck the microphone when the recording was made. All you want is a completely linear recording and playback system. But with recording and playback equipment, recording engineers, mastering equipment, amplifiers and speakers all adding non-linearities, there is no hope of that simple ideal. Instead, the goal, nearly as ambitious, is to produce a sound wave that the listener will find pleasantly reminiscent of the original. Unfortunately, psycho-acoustics has been unable to determine just where compromises are possible. Moreover, "Fashion is a part of why people buy tube equipment," said Luke Manley, president of Vacuum Tube Logic, a Chino, Calif., manufacturer of audio gear. "A lot of people like to see the glow of the tubes, and you design the cosmetics of the equipment for that. People will buy tube sound because they like the look of the amps."
Some designers of tube equipment admit that this is one fashion statement that's getting out of hand. "There is some very expensive equipment out there that is very bad," said one. "I'm afraid tubes are getting to be a fad in the bad sense of the word."
In support of the fad explanation, double-blind A-B-X experiments-where listeners are asked to identify an amp as one or the other of two devices they have just heard-often result in null results. That is, most listeners can't tell whether X is tube or transistor with significantly better than 50:50 odds.
However, even some solid-state enthusiasts are cautious about such tests. "The experimental design and statistical analysis of most of these tests has been very superficial," said John Nye, vice president of solid-state equipment vendor Clayton Audio (St. Louis). Nye, a professional economist, is a heavy user of advanced statistical techniques. "Better-designed tests are often prohibitively expensive, and the best people to organize them aren't always trained in statistics," he said. "There have been lots of cases where an organization has designed a rather crude experiment, been unable to reject the null hypothesis with 95 percent confidence, and then-totally incorrectly-concluded that there was no audible difference between the sources. Often in these experiments some of the listeners-a statistically significant group, perhaps-were hearing a difference."
Results of listening tests where the audience knows which equipment is playing, or has been coached to listen for particular characteristics, often reveal dramatic differences. "When people know what they are listening to, they say that tubes produce a more liquid, luscious sound," said Counterpoint's Elliot. "Perhaps what you see, or what you expect to hear, conditions your listening experience. But that is a legitimate part of the product."
In fact, said Elliot, the subjective impressions listeners draw from tube equipment can be quite reproducible. "I can package up a new design and ship it to a variety of listeners in different countries. They will listen to the equipment independently and report back their subjective impressions. And those subjective statements will be similar."
This suggests there are tangible aspects to the way tubes reproduce music-not just a halo effect from the beautiful orange glow of the heaters, or a placebo effect from the shock of paying $15,000 for an amp. Is there any basis in electronics theory for such an assertion? Many say there is.
Some listeners may simply prefer the high distortion and limited frequency response that tube gear produces, some observers say. Certain listening tests suggest that moderate levels of some harmonics-usually even, as opposed to odd, harmonics-make a system sound more "musical" to some ears.
But many tube lovers dismiss this theory. "I don't think that is supportable anymore," Elliot said. "We are building tube circuits that have 0.004-percent distortion, and people still like their `tube' sound." In fact, certain characteristics of vacuum-tube circuits may cause the subjective differences. The most often cited is the way tubes manage extreme transients. "The way tubes handle transients and the way they clip make them so suitable for music reproduction," said Manley. "The way transistors clip when they saturate is very hard on the ear. And it is surprisingly hard to control saturation in an amplifier-even our 1,250-watt amplifier will clip occasionally."
Others suggest that tube designs-particularly single-ended (Class-A) designs, where a single device handles the full swing of the signal from most positive to most negative-achieve greater linearity than solid-state designs. There are good mathematical arguments to support such an idea. Most of the distortions that befall a waveform are due to non-linearities of one sort or another. Some come from the fact that no amplifier stage, whether tube or transistor, has a perfectly linear transfer function over a wide input range-there are always higher-order terms. A linear region that's a little wider and a little more linear would, in fact, result in less distortion.
There is also the matter of crossover distortion. In Class-B amplifiers, one device handles the positive-going part of the waveform, while another handles the negative. This mode of operation reduces the amount of power in the stage when the input is near zero. But it forces the two devices in the complementary pair to cooperate very closely-and very linearly-when the input is near zero. Little offsets or a tendency to get non-linear near zero input can contribute audibly to distortion.
It has been argued that crossover distortion is more easily handled with tubes, and that better saturation behavior and improved linearity may in fact be characteristics of good tube designs. If so, these virtues are being revealed by fixes to some of the historic problems of tube gear-power supplies and output transformers.
"Tubes have become more viable now precisely because of new transformer, power-supply and power-capacitor technology," said Vacuum Tube Logic's Manley. "With vacuum-tube rectifiers and older storage capacitors, it was very difficult to store enough energy in the power supply to respond to big transients. But with the technology we have now, it's not unusual to have 2,000 microfarads of capacitor there-huge compared to what you could do in the early stuff." Slew rates of 25 V/microsecond are possible, he said.
"Transformers have also improved dramatically. If you wind very carefully and use the new insulation techniques, you can get very tightly coupled windings, and have excellent frequency response," Manley said. Some purists have taken tubes to a sort of esoteric minimalism. If a tube operating in a highly linear manner sounds best, they reason, let's eliminate anything that will color the sound. Class-B design, decoupling capacitors, unnecessary resistors must go. Even some of the internal frills, like heaters and screen grids, that improve the behavior of tubes are discarded, accused of spurious coloration.
The result is an almost religious pursuit of simple, single-ended triode amplifiers, most often using the venerable Western Electric 300B power triode as an output device. Such amps have been nearly a cult item in parts of Europe and Japan, and are now spreading in the United States. The claimed advantage is incomparable musicality, derived from the way the simplicity of the circuitry unmasks the innate beauty of the triode-amplifier stage. But the concept has detractors. "I don't do single-ended designs," Manley said. Such designs "saturate their output transformers very quickly. In practice, they are limited to 10 W or so, and have very poor frequency response."
Ultimately, it may be that attention to the listener's experience, rather than the choice of a particular technology, determines the results. Nye of Clayton Audio suggested as much. "The decision to use tubes or solid-state design is partly personal preference and partly economic," he said, noting that "there is no inexpensive approach to tube design. In the end, the designer's comfort with the technology may count for more than the decision to go one way or the other."
Copyright ® 1996 CMP Media Inc.