One form of some impurity that hasrecived relatively little notice until lately is "phase distortion." For years it was almost ignored by audio insiders because it was difficult to measure, it often was masked by more obvious forms of distorions and there was little agreement as to now important it is in listening to music.
As sound-reproducing equipment hasbecome generally better, familiar distorions (such as harmonic and IM) have been greatly reduced. The more subtlesonic derangements lurking under them are thus more readily discerned. And more sophisticated test gear has been developed to detect them in lab tests.
"Phase" in audio refers to the time-relationships of the elements of a signal. For instance, a single burst of music that lasts no more than a second may contain tens of thousands of "tonal constellations" in a precise sequence of start, rise and decay. The extent to which this sequence is preserved and reproduced accurately is now regarded as important to the realism or "naturalness" of your system's performance.
To the extent that the sequence is deranged, the music you hear is distorted. Often it is heard as a blurring of inner betail during complex musical passages, Specially when played fairly loudly. You may get the whole sonic picture, but many of its finer shadings will be obscured.
At its worst, phase distortion lends music a "canned" quality. It also has been described as an "acoustic traffic jam" in which all the elements of a musical signal cannot get through the system smoothly.
When testing for phase distortion, engineers use a test signal called a "square wave." This signal is rich in harmonics and contains frequencies a decade above and below its nominal "center" frequency. Thus, for instance, a 1000-Hz square wave contains frequencies from 100 to 10,000 Hz. On an oscilloscope it looks like a perfect square.
The extent to which it emerges from the amplifier or sistem as a perfect square is an indication of phase response. If the leading edge rises vertically the response has "fast rise time" -- this is very desirable. But if it tilts, that is a sign of less-than-perfect bass response. A rounded top-edge is a sign of imperfect treble response. A wavy top edge shows "ringing" which is a spurious response produced by the equipment after the signal itself. This particular form of distortion fouls up the transient response so that short, intense sounds sound "extended" or "blurred" This is especially noticeable on drumbeats, plucked strings etc.
All the answers are not in yet, but it is now widely agreed that one basic way to get good phase response is to design a circuit (in an amplifier), or a network (such as the frequency crossover in a speaker system) to have a smooth bandpass characteristic that is wider than the nominal range of frequencies to be handled. For this reason, do not be surprised at amplifier response figures that go beyond the nominal 20 Hz to 20,000 Hz range. The extra margin on either side of the response range is a good way of insuring that the desired range of frequencies will get through cleanly. When the audio signal lacks these "margins" it squeezes through, and in the process generates phase distortion.
Good phase response can be more important to natural sound than high power. A 25-watt amplifier with good phase response will sound better than a 50-watt unit with high phase distortion. The lows will be better defined, and the treble will sound clear but not strident.
In speaker systems, the design of the crossover network is far more important than the physical spacing of the individual drivers. The latter gimmick, often touted in terms of aligning all the voice-coils so that the sound from each "starts toward you at the same time." is mostly nonsense since, to be really effective, this kind of structural correction would involve distances of a few feet rather than a few inches. The trouble is, few speaker manufacturers talk much about their crossover networks, but maybe it is time they did.