Loudspeakers: Objective evaluations - Part 4: Measurements of real-world consumer loudspeakers

Let us pause and have a look at some real loudspeakers designed for surround use. The first example is a bidirectional on-wall design that can be used in any of several configurations. First, each directional loudspeaker can be fed a separate signal. Second, only one of the loudspeakers can be used. Third, both can be used, connected in phase and fourth, both can be used connected out-of-phase. In this example product, the real engineering effort went into the performance of the individual front- and rear-firing systems and how they merged when operated together in phase. An out-of-phase setting was provided, but performance in this mode was not optimized.

Figure 18.19 illustrates the behavior of the three directional modes available on a particular bidirectional surround loudspeaker. In (a) the monopole mode produces a good-looking family of curves, distinguished by the tight grouping of all of the curves, from on axis, through listening window and early reflections, to sound power. Only above 10 kHz is there any misbehavior, probably because the listening axis is about 60° off of the tweeter design axis. Cabinet diffraction may be involved. The sensitivity of this configuration is low because only half of the transducers are used.

FIGURE 18.19 Measurements on a bidirectional surround loudspeaker with switchable directivity patterns. The listening axis is perpendicular to the (side) wall. Note that the DI in these and all other diagrams in this book is calculated as the difference between the listening window and the sound power curve. Normally, the listening window curve is only slightly different from the on-axis curve. However, when the front- and rear-firing drivers of these loudspeakers are connected out of phase - as in (c) - there is a great amount of acoustical interference at and around the listening axis. This is what creates the "null." In that mode, the shape of the on-axis curve applies only to one specific microphone location. Because it is the result of acoustical cancellation, the curve shape changes considerably with small angular changes. Note the greatly different listening window curve representing a spatial average over ±30°. Every listener would experience a different directivity index if it were measured in the conventional manner. The directivity index in such a loudspeaker has little practical meaning.

In Figure 18.19b, the bipole mode has substantially higher sensitivity and is capable of considerably more sound output because all transducers are in use and they are all operating in phase. The family of curves describes a good-sounding loudspeaker. The DI describes a loudspeaker that is almost (hemispherically) omnidirectional. This seems to fit the description of the loudspeaker required in Figures 16.7 and 16.10.

When the drivers are operated in the dipole, out-of-phase mode, sensitivity plummets because the drivers are working against each other. This acoustical interference can be seen at work in the irregular shapes of the direct sound curve and the listening window curve. This is the penalty of the physical separation of the front and back radiators. What happens within the broad null is disorderly, as the sounds move in and out of phase with each other, depending on frequency and horizontal angle. In terms of sound quality of the three directional options, it is evident that the dipole mode is not competitive. No more will be said about this performance because the next figure includes four loudspeakers that were optimized to perform in the dipole mode, that being the only mode available in them.

Figure 18.20 shows the dipole from Figure 18.19c and four other dedicated dipole surround loudspeakers, three of which have THX certification. Readers by now can probably guess how these loudspeakers sound. They are far from ideal.

Apart from speculations about what their absolute sound qualities may be, it is evident that they are all very different from one another. In terms of specifying the performance of these loudspeakers, it has been common practice to use sound power, the total sound radiated by the loudspeaker, most of which must be reflected by the room surfaces before arriving at the listener. Sound power alone is not a reliable measure of sound quality. Even in an ideal, highly-reflective, frequency-neutral room, these loudspeakers cannot sound or measure the same.

The large and inconsistent differences between the on-axis curves and the listening window curves confirm the complexity of events within the acoustical interference region at and around the listening axis. It is further obvious that these loudspeakers revert to conventional wide-dispersion devices at frequencies below about 500 Hz. Because it is frequencies in the range from about 100 Hz to about 1 kHz that generate the desirable perception of envelopment, it is fortunate that much of that capability remains intact. Let us put these data into the context of what is required of surround loudspeakers:

• Localizable sound effects directed to a single channel. With the on-axis and listening window curves so variable and so different from each other and the other curves, sound quality suffers. These curves are also substantially attenuated in the mid-upper frequencies, so the direct sound - the one that determines direction - may not be as loud as a first reflection from a large surface. The author recalls more than one installation in which sending a signal to the side surround loudspeaker produced localization at the rear wall. Because the precedence effect is substantially nullified if the direct sound and delayed versions of that sound have different spectra, localization will be less than ideal.

• Enveloping ambience and music, involving both/all surround channels. Envelopment is a perception generated by sounds in the frequency range from about 100 to about 1000 Hz that arrive from the sides 80 ms or more after a similar sound from the front (see Figure 7.1). The essential concept of the dipole prevents this from happening because more sound is radiated toward the front and rear walls than toward the listener. Sounds that arrive from those directions are less productive at producing envelopment (see Figure 8.6). However, close inspection of the curves reveals that these bidirectional out-of-phase loudspeakers combine to mono at low frequencies to preserve low-frequency output. From about 500 Hz down, they all exhibit somewhat conventional wide-dispersion behavior. As a result, some impressions of envelopment are preserved, although only at low frequencies, and sound quality has suffered in the process.

• A vocal or instrumental component of a "middle of the band" musical recording. If the customer decides to listen to music, the requirement is for comparably good, preferably identical performance from all loudspeakers. Dipoles don't have it.