1268 Chapter 34
phone, then the system will almost certainly feed back.
If the total direct plus reverberant level at the micro-
phone is at least 6 dB below the expected level from
the talker, the system will probably be stable.
Reference 6 provides a method, if complicated, for
calculating the total reverberant level at the microphone
from a complex cluster. The PAG and NAG equations
may be used with a value of N, calculated as discussed
in Section 34.3.2.10 for the complex cluster, although
the direct plus reverberant approach will probably give
a more accurate result. EASE and Modeler also provide
this capability.
The actual feedback process also involves reflections
from near the microphone, since these reflections can
increase the sound level at the microphone by as much
as 6 dB (for a reflection from a hard surface that is in
phase with the source sound and, therefore, adds coher-
ently with the source sound). Thus, such a reflection
could cause a system to feed back even when calcula-
tions would show a 6 dB feedback stability margin.
Addition of more than one microphone to the
complex cluster system will not automatically lower the
gain before feedback by 3 dB as indicated in the PAG
equation. That is because each microphone may receive
a different amount of direct sound from the cluster and
may be subjected to different nearby reflections.
Assuming similar microphone locations, however,
simplifies the calculation, and a 3 dB reduction in gain
before feedback is a reasonable assumption in most
systems. This means that the total sound level from the
cluster reaching either microphone must be about 9dB
below the expected talker level at that microphone to
achieve the 6 dB feedback stability margin. This also, of
course, assumes that the gain (volume control setting)
of each microphone is similar.
34.3.2.11 Signal Alignment and Purposeful
Misalignment in Cluster Design
Ideally, the sounds from two loudspeakers that cover the
same audience area reach the listeners’ ears at precisely
the same moment. However, experienced designers know
this is almost impossible to achieve in a real system.
One proposed solution to this problem is signal
alignment of the cluster. To implement this concept, the
system must be designed with a separate amplifier
channel, and separate signal delay channel for each
loudspeaker. The signal delay must be adjustable in
10 μs (or smaller) increments. Choose a reference loud-
speaker, usually a long-throw loudspeaker, and delay all
the other loudspeakers so their signals line up in time
with the reference loudspeaker.
Signal alignment can work well when the audience is
concentrated in a small area. However, it is very diffi-
cult to design a cluster to be properly signal aligned for
a larger audience area. Aligning the cluster for one
listener may make matters worse for another listener.
Also, correct signal alignment can vary with frequency.
As a result of these and other problems, signal align-
ment is best for very simple clusters or for clusters that
have been specifically designed for signal alignment.
Another solution to the problem of signal alignment
is to purposely misalign the loudspeakers in such a way
as to simulate beneficial reflections (see Section
34.2.3.1.2). This concept relies on the idea that a
delayed signal arriving between approximately 20 ms
and 50 ms after the original signal is beneficial to both
Lp and intelligibility.
Purposeful misalignment ideas can be used to advan-
tage in designing an exploded cluster. Following this
concept, it is acceptable for a listener to hear two clus-
ters at once if the signal from the second cluster arrives
between 20 ms and 50 ms after the signal from the first
cluster. Some designers use DSP signal delay on alter-
nate clusters in an exploded cluster to achieve this goal.
It is even possible to combine the concepts of signal
alignment and purposeful misalignment by carefully
aligning a small cluster and purposely misaligning that
cluster with another, nearby cluster. However, there is as
much art as science in the implementation of either
signal alignment or purposeful misalignment. For this
reason, the designer is cautioned to rely on traditional
cluster design principles and to not depend on signal
alignment or purposeful misalignment for the success of
a design.34.3.3 The Distributed Loudspeaker SystemThere are several types of distributed loudspeaker sys-
tems but all share a common theme. In contrast to a cen-
tral cluster where the loudspeakers are all concentrated
in one location, the loudspeakers in a distributed system
are distributed throughout the audience area in such a
way as to cover the area evenly.
Because every listener is more or less the same
distance from a loudspeaker, coverage can be very
uniform from a distributed system. In addition, if the
system is carefully designed, the potential for feedback
should be very low, and because every listener is rela-
tively near a loudspeaker, the direct/reverberant ratio is
high and intelligibility is often very good.
Thus, in the ideal case, a well-designed distributed
system can work very well. One problem with even the
ideal distributed system, however, is that the listeners