Quality Lies in the Details

Time to 'fess up: How many of you actually read the "Measurements" sections of Stereophile's equipment reports and understand what's being measured, and why? I suspect that many readers skip over the technical assessment of the reviewed product and make a dash for the "Conclusion."

This is unfortunate, as there is lots of useful information under the "Measurements" heading that has real-world implications for the purchaser of that component. However, without an understanding of measurements and the significance of the results, the graphs and measurement text can become a confusing mess of alphabet soup.

I can relate: I get glassy-eyed when I read in Rider, a motorcycle touring magazine, the following technical description of a particular bike: "A liquid-cooled, 1084cc, 16-valve V-four with longitudinal crankshaft—a flying Vee—delivers its power to a cassette-type five-speed drive shaft. Each cylinder head's pair of double-overhead cams are driven off their ends by an idler gear and toothed belt, and the cams actuate shim-under-bucket tappets with 16,000-mile adjustment intervals." Say what?

To bridge this gap, Stereophile will be publishing a series of articles on how we measure products under review, and what those measurements reveal about the product's performance. The emphasis of these articles will be on how the product's technical performance affects its musical performance in your system. I'll start the series with CD-player and digital-processor measurements, which will be followed at irregular intervals by explanations of loudspeaker and amplifier measurements.

While it's true that measurements are not a reliable indication of how a product will sound, some aspects of measured performance do correlate with listening impressions. Moreover, measurements often reveal a product's idiosyncrasies that may make it unsuitable for certain applications or with some other components—I'll give examples when they appear. Finally, it's important to know which products are well-engineered and which aren't. Again, the better-measuring products won't always sound better, but a good-sounding, well-engineered component inspires more confidence than one that performs poorly on the bench. In short, some of the important information in the review could be overlooked if its relevance isn't understood.

I'll go through each digital-processor measurement in the typical order in which it appears in a review, explain what aspect of the component's performance is being measured, and, where appropriate, outline the measurement's relevance in selecting components. Examples from Stereophile's equipment reports will be included to demonstrate poor and good performance.

Output level
Let's start with a digital processor's maximum output voltage level. (I'll use the term digital processor, although I'm also referring to CD players—the measurements are almost all the same.) The processor under test is driven by the digital code representing a full-scale or 0dBFS 1kHz sinewave. This signal can be generated by the Audio Precision System One's digital signal generator—see the sidebar—or by playing a test CD and feeding the transport's digital output into the processor. The analog output voltage of the processor or player is measured by the Audio Precision System One, and displayed on the computer screen and noted.

Knowledge of a processor's output level is useful in several ways. First, a processor with a very high output, such as any of the Theta or Kinergetics units, wouldn't be ideal for driving a preamplifier that had high voltage gain (amplification—another useful thing to know). Although it's unlikely that the processor would overload the preamplifier's input (there's another important measurement), you would end up using the preamp's volume control at the very low end of its range. Most volume controls have their greatest channel imbalance (one channel becomes louder than the other) when turned down. In addition, setting a "just right" volume is more difficult with high-output processors and high-gain line-stage preamplifiers; the volume control becomes overly sensitive. This is particularly true of preamps with detent level steps instead of a continuously variable volume control. It's also useful to have a rough idea of a processor's output level when auditioning at a dealer—very few dealers match levels.

hollowman's picture

Can one correlate intersample overs to the square wave phenomenon JA is referring to here? That is, Philips, Wadia, Pioneer's Legato Linear, and Meitner do not "compress" the overs?