Handbook for Sound Engineers

Loudspeakers 631

more musical relationship to the fundamental, going
upward in frequency in successive octave steps.
Because of this, it may be true that even-order distortion
products are more readily tolerated (or even preferred)
by many listeners.
Because of the relatively high levels of distortion in
all loudspeakers and the wide variety of ways in which
a signal can be distorted, there is no consensus in the
industry as to the best means for characterizing the
distortion performance of a loudspeaker. The audible
significance of the nonlinear distortion caused by a
loudspeaker is best judged in person, and the results
may or may not correlate well with commonly
measured data.

17.9.7 Characterization for Design Purposes

Prior to beginning work on a new design, the designer
will (or should) develop a set of performance criteria for
a loudspeaker. Parameters typically considered are
bandwidth, available acoustic output, directivity, and
efficiency. Often, the designer must interpret subjective
information provided by others and translate it into
performance specifications.
The more complete the data regarding target perfor-
mance, the more satisfactory the finished loudspeaker is
likely to be. For this reason, targeted applications for a
new design should be well understood and well defined.
Possible performance definitions include characteristic
isobars, lower and upper cutoff frequencies, tolerance
for nonideal amplitude response (off-axis as well as
on-axis), maximum acoustic output, maximum distor-
tion level relative to fundamental, and phase versus
frequency criteria. A reasonably good idea of minimum
acceptable performance is also useful for purposes of
cost engineering. Once the overall loudspeaker perfor-
mance envelope is finalized, requirements for individual
component performance can be established.
The specific measurements that should be performed
on an individual component will depend on the trans-
ducer or radiator in question and on the nature of the
loudspeaker of which it will become a part. Impedance
versus frequency measurements are always essential.
For woofers, this will allow determination of the param-
eters necessary for the design of an appropriate enclo-
sure. For horn/driver combinations, the designer needs
to know the frequency of mechanical resonance. It is
also possible to identify internal horn reflections and
diaphragm breakup problems in an impedance curve.
Finally, the behavior of the component as an electrical
load is required for the design of passive crossovers.

Measuring a representative set of transfer functions

is an essential part of the component characterization

process. What constitutes a representative set will

depend on the component. A woofer will need compara-

tively few measurements if it is well behaved and to be

used only at low frequencies, a horn will need a signifi-

cantly larger number of measurements, and an array of

two or more components operating over the same range

of frequencies will require still more measurements.

Some devices cannot be suitably characterized by a

reasonable number of measurements. The measurement

process will determine if a component is usable in the

intended application and will aid in early prediction of

the ultimate performance of the completed loudspeaker.

The extent of testing needed to evaluate the most

complex component is likely to be required for the

complete system. The degree to which target perfor-

mance objectives have been met should be established

at the prototype stage. Any necessary modifications

may then be made and the system retested. This process

may continue through as many iterations as necessary

for the loudspeaker to perform as desired.

17.9.8 Characterization for the User

Once a loudspeaker is in production, performance data

will be required for (a) giving potential buyers informa-

tion for comparative purposes, and (b) use in the design

of sound systems. Loudspeaker performance data

provided to the sound system designer should be suffi-

ciently comprehensive for acceptably accurate predic-

tion of the performance of a sound system.

Unfortunately, the volume of data required to fully char-

acterize a loudspeaker is, as we have discussed, quite

large. If hard copy were generated with all pertinent

information, most loudspeakers intended for profes-

sional use would require a small book.

There are many ways to provide transfer function

information about a loudspeaker. Keeping in mind that

our extended definition of transfer function for a loud-

speaker intrinsically includes directivity information,

possible formats include:

1. Amplitude response curves calibrated to a constant
   level reference (e.g., dB-SPL) with a specified
   signal input (e.g., 2.83 Vrms) at a variety of angles.
   This format has the advantage of explicitly showing
   the direct-field response that listeners in various
   locations relative to the loudspeaker will hear, Fig.
   17-54.


2. Amplitude response measurements as above but
   normalized to the response at a particular angle,