Editor’s note: This is a continuation of the discussion of factors of a “good” sound reinforcement system that begins here.
A common complaint of audience members who have become familiar with a particular sound system, typically in a venue with a permanently installed system and which they visit repeatedly (such as a church) is that there are locations where sound is obviously better than at others.
In other words, sound is not uniform throughout the audience area. In fact, this is a common problem with all systems, but in many cases, such as a one-time concert, the audience has no opportunity to compare the sound quality at different locations. The only evidence of the problem may be that various audience members disagree significantly on how good or bad the sound was.
It’s remarkably difficult to provide uniform sound coverage throughout an audience area from a configuration of loudspeakers. The audience layout is often irregular and complex, and just as often, the dispersion from the loudspeakers is poorly delineated and frequency-dependent.
Add the necessity for multiple loudspeakers and physical restrictions on their locations and orientations, and it’s surprising that reasonably uniform coverage can ever be achieved!
These facts of life beg some questions: How uniform does coverage need to be? What variation limits are acceptable and what are desirable?
Variations with frequency add another dimension of complexity. Program material may result in another variable; perhaps the coverage uniformity requirements for speech are different from those for music. Almost no research has been carried out on these important questions.
As with other performance characteristics, there is no accepted method for measuring coverage uniformity. Worse, it’s not clear how it even might be measured subjectively.
One fairly obvious possibility for an objective technique is to feed a constant-level broadband signal over the system and sample it in many audience locations, or record it via microphone traverses through the audience area. Such samples can then be analyzed for uniformity in various frequency bands.
Figure 1: Variations across a seating area in the level of sound from a loudspeaker system displayed at several frequencies. (click to enlarge)
Let’s have a look at Figure 1, where broadband pink noise was played through a system, with the resulting sound recorded on a traverse through the audience area.
The recording was fed through filters centered at various frequencies and plotted on chart paper, so that the x-axis indicates the location within the traverse in every frequency band.
Again in Figure 1, we can see that there are significant level variations within each frequency band at different locations, and that these variations are different at each frequency. Thus, not only is the level different at each location, but so is the frequency balance.
It should further be noted that these measurements were made on a system that was carefully designed, installed, and adjusted to produce the best possible results in this situation. In addition, the room was rather reverberant, which might have reduced the level variations.
The subjective impression was that the coverage was quite uniform, which suggests that this measurement technique might be sensitive enough to reveal any audible variations.
Feedback & Suppression
When a system’s microphone can pick up any sound from the system’s loudspeakers, which is almost always the case, feedback is possible. The characteristics of this phenomenon—and ways to suppress it—are fairly well understood, if only modestly effective.
However, note that some audible effects of the typical suppression techniques have not been well studied.
Feedback frequencies are typically suppressed by means of narrow band-reject (notch) filters added to the system electronics, with each filter tuned to a primary feedback frequency. Such filters ring measurably, with the decay time of this ring varying according to the filter characteristics.