Working in the Front Line Page 6

Another revealing example is the effect of printed circuit boards on amplifier sound quality. In one example, an amplifier was prototyped in hardwired form using phenolic paper pin board, with a physical layout and connection wiring precisely conforming to a correctly designed printed circuit board. Thoroughly measured and auditioned, it gave an excellent performance. Second prototypes were then built using pre-production pcbs. By intensive measurement, the two were judged to be almost identical, yet the sound quality of the second version was significantly poorer.

After some investigation, the pcb was suspected as the reason for the poor sound; several complete prototypes were therefore made with different board dielectrics; eg, bonded paper and glass epoxy. Different foil thicknesses and copper purities were also tried. All measured well, yet all showed further sound-quality differences, the work leading to identification of a satisfactory compromise.

Conventional electronic wisdom indicates that while pcb quality may be relevant above 50MHz, it is of no importance to audio amplification working at less than a hundredth of that frequency. This may be true for noncritical applications, but where sound quality matters and where sensitive critical auditioning is involved, not even the printed circuit can be left to chance.

When a single high-quality plastic film capacitor can be audibly identified under double-blind conditions, it is not so surprising that the much poorer dielectric of a pcb has an audible effect on a high-quality amplifier.

Still less welcome to the engineering establishment is the discovery that audio cables vary in their subjective accuracy; rather less than amplifiers, it must be immediately noted, but nevertheless in ways which can be described and ranked on merit.

With the finest of today's systems the best cable is fortunately close to invisibility in audio terms—the ideal condition. The results from cable reviewing suggest that the use of poor or inappropriate cabling leads to a loss of up to 30% in performance in a state-of-the-art system.

The significance of cable quality is understandably proportional to the quality of the reproducing system, and becomes irrelevant in the context of rack systems and similar fundamentally compromised systems. The primary requirement for assessing small sound-quality differences is that the reproducing system used must be of the highest available quality, chosen by a combination of trial, experience, and informed opinion. It must then be optimally set up and installed in a room possessing favorable acoustics, and fed neutral, high-quality program. Put bluntly, there is no point in attempting to quantify the perceived depth in a stereo image illusion if the system is incapable of reproducing it, or if the source material lacks the necessary recorded information.

A wine taster cannot perform when using dirty or contaminated glasses; likewise, an art critic cannot make reliable judgments when wearing shades.

Detailed comparative tests made on audio cables have brought to light a diversity of previously unsuspected and therefore neglected factors which have subjective consequences:

Dielectric: A good correlation has been observed between dielectric loss and sound quality. A vacuum insulator shows the lowest loss, followed by air, and then by a range of dielectric materials commonly used for cables of all classes. The subjective ranking correlates with their dielectric properties. Thus, foamed or predominantly air-spaced types with PTFE, polypropylene, and polyethylene dielectrics score highly, while higher-loss materials such as PVC are distinctly inferior, even to the point of generating identifiable colorations and changes in timbre.

Associated with the subjective performance of the cable dielectric is the insulating thickness, this often related to the manufacturer's voltage rating. Better sound often follows higher ratings. Solid dielectrics are common and include those plastics mentioned above, as well as higher-molecular-weight polymers, ceramic powder, silicone rubber, and resin-impregnated glass fiber. Natural thread such as cotton or silk has been tried, plus various grades of carbon-based rubber. Every dielectric can be shown to have its own distinctive sound, even when used in a line-level interconnect application of just 1m in length.