Audio equipment is, by definition, ultimately intended to provide sounds that will be heard by a listener. Despite rather more than a century of experience in the electrical transmission and reception of audible signals, the relationships between the electrical waveforms into which sound patterns are transformed and the sound actually heard by the listener are still not fully understood.
This situation is complicated by the observable fact that there is a considerable variation from person to person in sensitivity to, and preferences in respect of, sound characteristics, particularly where these relate to modifications or distortions of the sound.
Also, since the need to describe or indeed the possibility of producing such modified or distorted sounds is a relatively new situation, we have not yet evolved a suitable and agreed vocabulary by which we can define our sensations.
There is, however, some agreement, in general, about the types of defect in electrical signals that, in the interests of good sound quality, the design engineer should seek to avoid or minimize. Of these, the major ones are those associated with waveform distortion, under either steady-state or transient conditions; the intrusion of unwanted signals; relative time delays in certain parts of the received signal in relation to others; or changes in the pitch of the signal, known as “wow” or “flutter,” depending on its frequency. However, the last kind of defect is only likely to occur in electromechanical equipment, such as turntables or tape drive mechanisms, used in the recording or replay of signals.
It is sometimes claimed that, at least so far as the design of purely electronic equipment is concerned, the performance can be calculated sufficiently precisely that it is unnecessary
to make measurements on a completed design for any other reason than simply to confirm that the target specification is met. Similarly, it is argued that it is absurd to attempt to endorse or reject any standard of performance by carrying out listening trials. This is so since, even if the results of calculations were not adequate to define the performance, instrumental measurements are so much more sensitive and reproducible than any purely “subjective” assessments that no significant error could escape instrumental detection.
Unfortunately, all these assertions remain a matter of some dispute. With regard to the first of these points—the need for instrumental measurements—the behavior patterns of many of the components, both “passive” and “active,” used in electronic circuit design are complex, particularly under transient conditions, and it may be difficult to calculate precisely what the final performance of any piece of audio equipment will be over a comprehensive range of temperatures or of signal and load conditions. However, appropriate instrumental measurements can usually allow a rapid exploration of the system behavior over the whole range of interest.
On the second point, the usefulness of subjective testing, the problem is to define just how important any particular measurable defect in the signal process is likely to prove in the ear of any given listener. So where there is any doubt, recourse must be had to carefully staged and statistically valid comparative listening trials to try to determine some degree of consensus. These trials are expensive to stage, difficult to set up, and hard to purge of any inadvertent bias in the way they are carded out. They are therefore seldom done, and even when they are, the results are disputed by those whose beliefs are not upheld.
Instrument Types
An enormous range of instruments is available for use in the test laboratory, among which, in real-life conditions, the actual choice of equipment is mainly limited by considerations of cost and of value for money in respect to the usefulness of the information that it can provide.
Although there is a wide choice of test equipment, much of the necessary data about the performance of audio gear can be obtained from a relatively restricted range of instruments, such as an accurately calibrated signal generator, with sinewave and square-wave outputs, a high input impedance, a wide bandwidth AC voltmeter, and some instrument for measuring waveform distortion—all of which would be used in conjunction with a high-speed double trace cathode-ray oscilloscope. I have tried, in the following pages, to show how these instruments are used in audio testing, how the results are interpreted, and how they are made. Since some of the circuits that can be used are fairly simple, I have given details of the layouts needed so that they could be built if required by the interested user.