1360 Chapter 35
Practical Considerations. Improvements that are theo-
retically desirable must also be practically accom-
plished. It is clear that only reasonable measurement
times can provide reliable data in an efficient manner. In
practice, an angular resolution of 5 degrees has proven
to be adequate for most needs, sometimes even lower
resolutions of 10 degrees can be sufficient. Simulation
software packages should be able to handle higher reso-
lutions as well, but only for special cases. This is partic-
ularly feasibl when measuring durations can be reduced
using multiple microphones, like ten or nineteen receiv-
ers arranged on an arc. This technique requires some
care in the measurement setup since all of the micro-
phones have to be calibrated and normalized relative to
each other.
Also the acquisition of both magnitude and phase
data requires more care than just the measurement of
magnitude-only data. However, modern impulse
response acquisition platforms provide a good means to
obtain complex data and a sufficient frequency resolu-
tion. The representation of a loudspeaker directivity
function based on impulse response wave files is thor-
oughly discussed by the Working Group of the Stan-
dards Committee SC04-01 of the AES.^7 As we will
present farther below, the utilization of phase data in
acoustic modeling has become an important factor. As
an illustration, Fig. 35-22A—D show, the magnitude
and phase data for a loudspeaker—UPL1 from Meyer
Sound Inc.—in high resolution in both MATLAB and
EASE.^35
Additionally, it is worth mentioning—and we will
give some practical guidelines in the next
sections—that, in order to obtain acceptable data for a
point source approach, measurements have to take place
in the far field of that assumed point source. Like indi-
cated previously, this may be difficult for large multi-
way cabinets or column loudspeakers.
In general it must be emphasized that the computer
model utilizing this loudspeaker data can only be as
good as the data of the lowest quality included. Nowa-
days, the accuracy of the loudspeaker data is often much
higher than that of the material data. Absorption and
scattering coefficients are usually only known in or
octave bands for random incidence. The user must be
aware that although loudspeaker direct field predictions
may be very precise, any modeling of the reflections and
the diffuse sound field in the room will be limited by the
available material data. Furthermore, it is not very likely
that there will ever be systematic, large-scale measure-
ments of angle-dependent complex directivity data for
the reflection and scattering of sound by wall materials.
To complete this practical perspective another point
of concern has to be underlined. Any data set describing
the acoustic characteristics of a loudspeaker should also
document important measurement parameters and
conditions. In particular, the point of rotation used for
the sensitivity and balloon measurements must be
defined in such a data set and indicated in the case
drawing as well, Fig. 35-23. Only when this reference
point is known will the end user be able to define
precisely the location of the loudspeaker in the
computer model and to obtain the right results.35.2.1.2 Simulation of Complex Loudspeakers35.2.1.2.1 Modeling by Means of a DLL Program
ModuleIn a first step to overcome the variety of issues related
to the reduction of complex loudspeakers to simple
point sources, the DLL approach was developed.36,.37
Technically speaking, the MS Windows dynamic link
library (DLL) is a program or a set of functions that can
be executed and return results. It cannot be run stand-
alone but only as a plug-in of another software that
accesses it through a predefined interface. The basic
idea is to move the complexity of describing a sound
source from the acoustic simulation program into a sep-
arate module that can be developed independently and
that can contain proprietary contents. In this way, a clear
cut is made between the creators for simulation software
packages and the loudspeaker manufacturers who can
develop product-specific DLL modules on their own.
However, acoustic prediction programs have
different underlying concepts and therefore, although
the DLL concept is a general philosophy, the DLL inter-
faces are different too. In consequence, a DLL built for
one simulation platform cannot be used for another.
Nevertheless, all DLL models share a similar approach
to resolve the given problems. Because they can be
programmed, they are essentially able to handle any
kind of data and realize any kind of algorithm. If the
mathematical description and/or sufficient measurement
data for the loudspeaker system exists, this information
can be encoded into a DLL. Given an appropriate theory
for how the source radiates sound, the solution can be
implemented without much compromise. Practically, the
DLL provides the data describing the radiation of sound
by a particular loudspeaker and the simulation software
employs this data to model the interaction of the source
with the room. For an example see Fig. 35-24.(^1) » 1
(^1) » 3