[SoundStage!]Paradise with James Saxon
Back Issue Article
March 1999

Numbers

Audio journalism has come a long way. When I read my first equipment report in Hi Fidelity magazine twenty years ago, I was bemused by it. First, there was a long section about how the component looked and functioned mechanically. Because I usually read about receivers, I had to wade through a catalog of all the knobs and switches. Then, the report devoted numerous paragraphs to how the product measured. Apparently, audio reporters of yore moonlighted for NASA and owned laboratories full of test equipment. Invariably, every bit of gear was used in order to make a full report. This lead to a presentation, with cursory explanation, of the college-level electronics math which fleshed out most of the report. Finally, near the very end of the article were one or two sentences commenting on how the unit sounded, upon which I pondered until the next issue. To me, this was the meat and potatoes of the whole effort and it was thin fare, indeed. In describing a receiver’s sound, a reviewer might conclude, "The Fuse Burner 220 produced 55 watts per channel of clean power and drove to distraction most of the loudspeakers we had on hand." If he were reflecting on the performance of a pair of loudspeakers, he might sum up in a similar vein, "The Compression Model 42 played clear and clean at ear-piercing levels." So much for sound quality.

In those days, sonic descriptions were hard to come by and were usually variations on the word, "clean." In several years of searching, I never once read of a component or loudspeaker that sounded "dirty." If I had, I might have bought such a product in order to learn the difference between a meritorious "clean" sound and a leprous "dirty" one.

Instead, I contented myself with reading test reports as if they were tea leaves, open to free-form interpretation. Continuous power ratings, input and output impedances, distortion percentages, signal-to-noise ratios, and power-handling capability appeared to be the stuff of good sound. Exactly how good was tough to say. Unable to grasp the significance of the measurements per se, I nevertheless made comparison charts of them. By means of a numerical grid I hoped to decipher a sonic pecking order based upon statistical ratings and an interpolation of how "clean" a component was said to have sounded.

Concepts such as harmonic definition, inner detail, focus, and dimensionality were outside my ken. Those were terms being refined at the time by "subjective" audio critics such as J. Gordon Holt, Harry Pearson, and Peter Aczel, persons as yet unknown to me. For most of us casual hi-fi nuts , "stereo" meant channel separation, not holographic imaging. The setup in my loft apartment was typical: The left speaker was on the floor facing one’s feet; the right speaker fired downward from an eight-foot-high loft thirty feet away. This gave full meaning to the term "channel separation." Rock groups played as far apart as the width of one’s dwelling. In fact, the closest approximation to a stereo image I heard in the 1970s was the signal that came from a single Bose 901. Such a wide monaural sound field held charms to my ears. Except for the need to purchase two 901s, I might have become a Bose fan then and there.

None of this is meant to condemn the way stereo reviews were once written. In fact, under the crush of so much "subjective" audio verbiage currently in vogue, I sometimes long for the comforting era of objective test reports. Despite the use of audio adjectives constructed, ironically, from visual parlance, most stereo reports seem uninformative. By sifting sentences and parsing phrases, I get an idea of how a reviewer thought a component sounded in his system based on his hearing acuity and predilections. I just don’t know whether the component would work for me. More and more, I wish a few numbers were posted as a guide. I miss my charts.

These musings derive from a kind of time capsule I recently found in Roberto Herrera’s audio workroom. Roberto, as I mentioned months ago, throws nothing away. Still, I was surprised to see amid a confusion of tape decks and stainless-steel integrated amplifiers a mint-condition issue of Stereo Test Reports, copyright 1980, casually awaiting my caress. The cover still had the sticker price taped to it, 59 colones. In today’s terms this would be less than a quarter, but back then was $8.00, a sizeable sum for a newsstand publication. Roberto saved the issue for two reasons: (1) audiophiles are pack rats; (2) he uses the published measurements as a guideline to servicing old equipment, most of which wanders in without service manuals.

Hefting the magazine, I realized that long ago I had once consulted the very same publication for buying guidance. As I flipped through the pulp pages, the thrill of rediscovery transported me to a different place and time. Here were the famous names of my audio youth, Crown, Phase Linear, SAE, Stanton, Pickering, Akai, ESS, Acoustic Research, KLH, and -- lo and behold -- Thiel. The technical terms and phrases used to portray the equipment were the very ones that had beguiled me decades before. In a flight of association, I also remembered the perfume of a certain girlfriend I had dragged to stereo stores with me, at which point I had to sit down.

Recovering from that bit of nostalgia, I recalled the stereo equipment I owned in 1980: a JVC receiver, Dual turntable and Crazy Eddie house brand loudspeakers. New Yorkers will recall trooping into El Loco Edwardo’s with a copy of Consumer Reports, demanding to audition highly rated Advents or AR speakers only to learn they sounded like fuzzballs. Years later, after having bought the clearer-sounding Crazy Eddie brand, one learned that Eddie’s Advents had been pierced in the tweeters for demo purposes. If anyone were deaf enough to prefer the punctured Advents, he received an original pair of "unmodified" speakers that were better than the house brand we smart guys bought and paid for. Most New Yorkers are too smart for their own good, including Crazy Eddie who went bankrupt.

Leafing through Stereo Test Reports, I saw several references to the scandal involving manufacturers’ power ratings. My JVC receiver was an example of the lawlessness that prevailed in the early days of transistor amplification. When I bought the receiver in 1971, a sticker on its faceplate proclaimed a power output of 140 watts. Since I paid $400 for the unit, this power rating equated to a cost of $2.85 per watt. However, the wattage figure was an overall total. Close reading of the owner’s manual disclosed that the receiver had an International High Fidelity ("IHE") continuous power rating of 35-watts per channel into 8 ohms. In other words, it actually cost $11.00 per watt. For such an investment, I could have bought a Marantz!

In 1978, the US Federal Trade Commission leveled the playing field by officially adopting the IHF Amplifier Standard for power rating. Suddenly, 25 watts looked sensible again, paving the way for such ground-breakers as the Mark Levinson ML-2 and the Bedini 25/25 class A amplifiers. Ironically, honesty in power rating was accompanied by a release of downward pressure on distortion figures. The point oh, oh, oh HD measurements that each new model proclaimed bottomed out as the new honesty in power output took on significance. But this is the point: Stereo Test Reports was about examining the numbers, challenging the manufacturer’s claims, and quantifying the merits of high fidelity components on the test bench. Although I was as innocent of their significance as a chimp using tools, I found test results to be comforting. Oddly, the verbal interpretation of the numbers seduced me. After reading Stereo Test Reports, I wanted to buy something!

By contrast the current crop of slick, subjective audio magazines seldom kindles my flames of desire. Most often, I find the reviews annoying and repetitive. Occasionally, I care enough to write a letter complaining about a choice of words or a specious conclusion. But where’s the itch to buy? It’s curbed by verbal excesses, and a lack of supporting data.

Here’s an example of a passage from Stereo Test Reports that would inspire my greed to buy. Notice how the words and numbers tie together in a dance of seven veils:

"The input stage incorporates a low-noise FET, which reduces A-weighted noise to minus 100dBW -- among the lowest values we measured. The output transistors must be among the fastest in the industry, for this amp has an extraordinary bandwidth (-3dB at 500kHz) and unmeasurable (at least to us) distortion right out to 20kHz....

"The unusually broad bandwidth of the M-A01 is reflected in a very high damping factor at 20kHz. At 79, it’s the highest we can recall. In the low and mid-frequency region the damping factor is constant at 270, indicating that the feedback factor is maintained over a much wider than average portion of the audio band....

"Every once and a while (maybe once a year) we find an amplifier that just seems to sound a bit better than others -- cleaner and with much greater detail. Frankly, we’re not sure why. The Mitsubishi M-AO1 is just such an amp. We could point to several factors.... But when we do so, we feel like the three blindfolded people trying to describe an elephant by touch; each "sees" only part of the picture. Suffice it to say that, in its power range, the Mitsubishi M-AO1 is as close to perfection as we have experienced."

Wow, where can I get me one of them Mitsubishis? The combination of statistics and phraseology still warms my innards. After such a succinct rave, who needs another three or four pages describing the sound?

Of course, the objective of high-end audio reporting is not to encourage sales of stereo equipment. A report that drives one to the limits of audio lust is one to suspect, even if it involves numbers and testing. For example, the passage above from Stereo Test Reports misses the truth by a wide margin. According to the author, Mitsubishi’s M-AO1 was all the amplifier one would ever want. Yet, if the amplifier sounded anything like most transistor amplifiers of the day, I doubt the few souls who purchased the M-AO1 continue to use it or have kept it as an investment. I may be wrong, yet I can ’t recall ever seeing a Mitsubishi amplifier of any kind available for sale or audition anywhere. If the amplifier was so great, why was it so unpopular?

Despite the appeal of a numbers-based test report, I am not advocating a revival of Stereo Test Reports. For all I know, STR continues to thrive upon the easily-impressed. So it goes. Rather, I am in full agreement with the philosophy of SoundStage! that our mission is to describe how a component sounds without engaging in advocacy or hype. Measurements and specifications divulge few hints about a component’s performance in a given system. Most audiophiles would probably agree that numbers hardly disclose the way a component sounds.

Yet, for me, purely subjective reporting with its constantly recurring adjectival flourishes has lost the balance that measurements might provide. So many products sound "transparent," offer "rich harmonic detail," or produce "a lifelike three-dimensional soundstage" that I long for the simple comfort of numbers. I wish I had a chart. Then I could consult the prose to see if the numbers mean anything, or review the numbers to see if the words mean anything.

Having been burned over the years by subjective tales of the latest and greatest, I find myself more and more interested in test results, particularly if they might verify a performance trait. For example, the pet parameter that I like to ponder is the signal-to-noise ratio. It occurs to me that amplifiers with conservative S/N ratios sometimes sound better than amps with higher ratios (the Mitsubishi M-AO1 had a S/N ratio of –100dB, an order of magnitude better than the –80dB figure claimed for the new Mark Levinson No.334!). Is it useful to our audio knowledge to understand how the components of this ratio are derived by various manufacturers? Should we test signal-to-noise claims ourselves to see if the numbers are fudged?

For instance, the aforementioned Mark Levinson No.334 retains the same signal-to-noise ratio its predecessors specified over the past decade, >80dB. Yet, to my ears, the No. 334 sounds "quieter" than any Mark Levinson amplifier in memory, providing a contrast of music to stillness which is lovely to behold. I think the greater contrast results from a lower noise floor. If so, is the Madrigal company sandbaggging on the new amplifier’s S/N ratio? Would an independent test verify my listening impressions? Calls to Madrigal prove unavailing, and I don’t have the means to measure for myself.

Yet, this is the type of sleuthing in which I would like to see SoundStage! engage. It may be, as some suggest, that we measure the wrong things in audio. Yet I consider it naïve to regard as meaningless company claims in owner’s manuals. Furthermore, I would like to see us verify independently product specifications in at least in two areas: (1) amplifier power ratings into various impedance loads and (2) impedance curves of loudspeakers. From my experience, the amplifier-speaker interface is the most critical nexus in audio. Still, most reviewers can only speculate how the sound of an amplifier or a speaker they are reviewing may be affected by the impedance curves involved. Especially when tube amplifiers are used in a system, the impedance-to-power relationship seems critical. Most conscientious reviewers will mix and match products in order to produce a harmonious pairing, but there is no predictability to this method.

I hope over time SoundStage! will have the wherewithal to publish measurements, at least those verifying specifications. I think we might see a pattern emerge. Those companies that produce reliable specs might also produce reliably good-sounding gear. Or maybe there’s no correlation between the "almighty specification" and good sound. At least we should see for ourselves.

As we become attuned to testing and measuring, I think our technical findings might help us to understand better why a component or loudspeaker sounds the way it does in a given system. After reading the standard audio adjectives millions of times, I would like to see a few numbers to back them up. Can one measure "transparency" in components, or harmonic definition? So far, no one has. This should not deter us from trying. If tests correlating to transparency or harmonic detail can be found, we could support our contentions with more exactitude, and perhaps spare ourselves from having to read the same sonic adjectives ad nauseam. I think technical testing would be a welcome addition to our arsenal. For me, the inclusion of a bit of "objectivity" would renew the fun of reading subjective reviews. And now, back to Stereo Test Reports and a description of the original THX component, Tom Holman’s Apt 1, the most famous amplifier that nobody owned, but George Lucas was watching.

...James Saxon
jimsax@soundstage.com

 

[SoundStage!]All Contents
Copyright © 1999 SoundStage!
All Rights Reserved