Equipment Directory Specs
The Annual Equipment Directory published each October by Audio is certainly a service to the buying public since it contains so many specifications on so many pieces of equipment. But the reader clearly is invited to make comparisons among brands, and within a brand, using the published specifications as guidance. For example, a listener could rightly want to know how much money needs to be spent to achieve a certain level of performance, or where more money spent yields ever decreasing returns.
Unfortunately, in several key areas there is much left to be desired to make such comparisons useful. Obviously it is the manufacturers who must supply accurate data to Audio; no one organization could hope to make the 40,000 measurements necessary to publish such a comprehensive listing. But there are often conflicting standards which result in measured, but not real, differences. Even within one standard, setup conditions can also influence the results.
Outlined below are some of the more obvious areas where comparisons must be made with a grain (should I say a shaker?) of salt.
Phono Signal-to-Noise Ratio. Phono signal-to-noise is the specification with the largest deviation from reality. This is clear in just glancing down the range of stated signal-to-noise ratios; the range encompasses figures from 60 to 160 dB! This should be and is unbelievable, and the principal problem with it is that it sours people on the ability of competent engineers to make real, valuable measurements.
We often hear the argument voiced that measurements don't mean anything because the instruments do not measure anything like the way we hear. In fact, IHF Standard A-202 specifies a "dummy" source impedance and appropriate psychometric weighting to closely associate measurements and listening.
The extreme range of the signal-to-noise specifications should give away the fact that manufacturers are not uniformly using the standard. While probably none of the measurements is an outright lie, the range itself should tell us that the results are not credible for comparison. And this indeed is the case.
The problem is that some manufacturers are measuring with a shorted input, while others employ the cartridge-like source impedance specified by IHF A-202. The real problem here is that one can actually design for better performance with a short circuit than with a cartridge, then measure with a shorted input and get phenomenal signal-to-noise ratios which are not in any way achievable in practice.
The only way to obtain a signal-to-noise ratio, re: 5 mV, "A" weighted, with the IHF source impedance connected, would be to reduce the temperature of the source impedance to within 2 millionths of a degree of absolute zero! Any actual source impedance has associated with it a certain amount of noise due to the random motion of electrons in a conductor at room temperature (similar to the Brownian motion of electrons in the atmosphere). The only way to eliminate this noise would be to reduce the temperature of the source impedance or cartridge to absolute zero (one visualizes mad-scientist experiments with tanks of liquid nitrogen connected to tubes running down the tonearm to supply the cartridge with cold). The noise level of the IHF standard source-impedance terminated with 200 pF and 47 kilohm is 78 dB below a 5-mV, 1-kHz input when "A" weighted. Since any realizable circuit will add some noise, signal-to-noise ratio numbers close to this have likely been measured in accordance with the standard and may be compared. Of course, some manufacturers are more conservative than others: Some give the limit of production, while others give the average.
Conclusion: One should look on all numbers greater than 78 dB as probably representing a shorted input condition and therefore not accurately reflecting the real-world condition of having a cartridge source impedance connected. Remember that design for a shorted input can yield worse results when a cartridge is connected than design for a cartridge source impedance.
Frequency Response. In the section on power amplifiers, "frequency response" was taken by many manufacturers to mean "power bandwidth." This was because of the inclusion of the phrase "at rated power" in the description of the column for frequency response, which implies that the question really asks for the "power bandwidth" rather than the "frequency response." Unfortunately, a number of manufacturers ignored the "at rated power" statement and provided a specification for low-level frequency response. This is clear because some of the claims stretch out to 250 kHz; I am certain they didn't intend this to mean that the unit will produce rated power at rated distortion, or lower, at that frequency! Furthermore, a specified frequency response without a statement of the deviations from flat in dB is practically useless.
Another influence on frequency response measurements, especially of preamplifiers and integrated amplifiers, is the fact that the setup conditions may change the response dramatically. For example, one could choose to rate the response only with a tone-defeat switch engaged. Or one could rate the response with the tone controls in but adjusted to give the flattest response, whether that response corresponded to the position marked flat or not. Also, naturally enough, most frequency response measurements are made with all filters disengaged, even though the action of the filters is intended to be infra- or ultrasonic and engaged in normal operation of the unit. Thus, the manufacturer who gives response under the standard operating conditions is penalized.
Wow and Flutter. Audio clearly calls for wow-and-flutter measurements to be made to the internationally recognized DIN 45 507 standard (which corresponds to the American IEEE and many other standards). Yet, it is obvious to anyone who has ever made these measurements that the quoted specifications greatly exceed those that could be expected to be realized in practice.
For example, in a recent Boston Audio Society test clinic, I measured one turntable-arm combination of Japanese origin which produced 0.12 percent DIN-weighted peak wow and flutter. The specification for this combination listed in Audio was 0.03 percent. Neither of these numbers would be important if both were inaudible.
But Stott and Axon of the BBC found that the threshold for the perception of flutter among the most sensitive group of listeners was in the range of 0.12 percent peak. Thus, what we measured is probably audible wow and flutter to sensitive listeners with piano music program material, but 0.03 percent weighted peak is probably inaudible to listeners on music (although it could probably be heard on pure tones). The difficulty is that there is a competing standard called JIS. This Japanese standard produces numbers which are highly optimistic compared with DIN-weighted peak measurements.
Some manufacturers have signaled their use of the Japanese standard by adding the term "W rms," standing for weighted rms to the number, but others, however, clearly have not. In our measurements on 28 cartridge-arm turntable combinations, the best measurement we got was 0.044 percent. The worst was 0.17 percent; the average was 0.09 percent.
Another difficulty is that the wow and-flutter measurements on a turntable are influenced by the choice of arm and cartridge. If, for example, the peak in the frequency response (which occurs as a result of the resonance between the tonearm, plus cartridge effective mass, and the cartridge compliance) happens to land near a frequency equal to the eccentricity of a motor pulley, the flutter will be exacerbated. So a wow-and-flutter measurement without reference to a tonearm and cartridge is not too meaningful.
Also, in our measurements we used an aluminum-backed lacquer master which is very flat physically. We expect that slight warps on vinyl pressings would increase measured wow and flutter. If the frequency of the warp should land near the frequency of the tonearm/cartridge resonance, we might expect a fairly large factor of increase. So what we were measuring (and the way things are usually measured) may be hopelessly optimistic compared to practice.
These are but a few of the more blatant areas where specifications from one manufacturer do not correlate with specifications from another. And, of course, any such survey which can devote only limited space to an individual product does not tell the entire story on specifications for a unit. For example, most engineers believe that distortions other than total harmonic distortion are more annoying to the listener, yet THD persists as most important nowadays by government decree. The only hope of covering all of the various distortion mechanisms which are known is to conduct a large variety of tests; such a variety could not be published in a format like Audio's Equipment Directory, yet a variety is almost certainly more important than the THD numbers published.
-Tomlinson Holman, Apt. Corp., Cambridge, Mass., USA.
References
1. Maxwell, John, "Phono Cartridge Noise," Audio,March,1977, pp. 40-42.
2. Holman, Tomlinson, "Noise in Audio Systems," AES Preprint No. 1418 (N-2) presented at the 61st Audio Engineering Society Convention, New York, November, 1978.
3. "Audible Effects of Mechanical Resonances in Turntables," Bruel & Kjaer Application Note.
The Editor Replies:
Touché, Mr. Holman, and thank you for expressing one small part of the difficulties we have in putting together our Annual Directory. We do indeed ask for power bandwidth in an ambiguous manner, and it shall be changed this year.
While in the past we have viewed many specifications with a raised eyebrow, we have also treated them as axiomatically true so long as they were responsive, that is in dB, rather than watts, if an answer in dB were being asked for. If the maker supplying the answer apparently knew what the question was, we have left the answer alone. Questioning the manufacturers' specifications, we have felt, is more properly left to reviewers in our Equipment Profiles section.
Each year, we hear from a number of irate readers (not to mention public relations agencies) asking why their favorite manufacturer has been omitted from the Directory. For the past three years, we have been fortunate to have enough space to reply that "neither makers nor products are deleted; we simply received no material." This year the deadline was some seven weeks after the Directory forms were sent out; phone calls were made and mailgrams sent as the absolute deadline approached. In the absence of material, we have felt that no listing is better than repeating that of an earlier year, and we do, of course, publish an Addenda. We are mightily sorry to leave anyone out, but we think it would be worse to include outdated specifications and old price information.
(adapted from Audio magazine, Feb. 1980)
Also see:
Forum (AM Clear-Channel Proposal) (May 1979)
Forum by Richard Heyser and John Curl (Sept. 1979)
Forum by Leonard Feldman (Jan. 1986)
= = = =