Handbook for Sound Engineers

(Wang) #1

1310 Chapter 34


trying to boost a diaphragmatic absorber (again, no
more than about 3 dB at first) to help smooth system
response. At this point, a smooth, not flat, response is
the goal.


In many two-way, central-cluster systems, response
peaks will come in the neighborhood of 200–2000 Hz.
These frequencies include the efficiency peaks of the
low- and high-frequency loudspeakers and reflect the
normal midrange deficiency of many two-way systems.
These peaks are, therefore, the first place to make
system adjustments, again, working for smooth
response at first.


34.5.2.2.8 The Desired System Response Curve


Once the system response is reasonably smooth, begin
adjustments toward the desired system response curve,
as shown in Fig. 34-76. For a speech-only system, the
final curve should be relatively flat from its low-fre-
quency limit (about 50–80 Hz) up to about 1000 Hz. At
1000 Hz, the response should begin a roll-off of about
3 dB/octave to about 8–10 kHz. Response above this
frequency should roll-off more rapidly (use a system
low-pass filter if available). For a music or music and
voice system, begin the roll-off at about 2000 Hz and
allow the response to follow this roll-off to 12.5 kHz or
higher with rapid roll-off above the desired maximum
frequency.


This high-frequency roll-off is a guideline and
should be modified for each individual system on the
basis of subjective sound quality. The purpose of the
roll-off is to improve the system sound quality, since a
perfectly flat system response, as displayed on the
real-time analyzer, will sound overly crisp, and vocal
sibilants (high-frequency breath sounds) will be overly
emphasized.


The final equalization curve will probably not follow
these rolloff curves perfectly. A final system curve
within r2 dB of the desired curve is, in most cases,
more than adequate. Avoid filter settings more than a
6 dB boost or cut whenever possible, remembering that
a 6 dB boost requires four times the amplifier power
output and four times the loudspeaker power capacity.
For this reason, a final equalization curve that requires
only a r3 dB setting on any individual filter but is
within only r3 dB of the desired curve may sound better
than a final curve that is within r1 dB of the desired
curve but required r6 dB of equalization at several filter
positions (and, therefore, reduced system head room
and increased system noise).


34.5.2.2.9 Why High-Frequency Roll-Off Is Required in
Equalization

The reverberant field in most rooms is dominated by the
low frequencies. When the equalization microphone is
placed in the reverberant field (past the Dc, critical dis-
tance from the cluster), the frequency response shown
on a real-time analyzer will also be dominated by the
low frequencies. From another point of view, the low
frequencies don’t attenuate much past Dc because they
are supported by the reverberant field. The high fre-
quencies, however, continue to attenuate past Dc since
they are not supported as strongly by the reverberant
field. For this reason, the real-time analyzer will show a
frequency response that is dominated by the low fre-
quencies (bass heavy).
If equalization is then performed using this real-time
analyzer display, which is low-frequency dominated,
most people will tend to boost the high-frequency
response of the system (or cut the low-frequency
response) to make the analyzer display a more flat
response. This, however, causes the direct sound from
the loudspeaker system to be dominated by the high
frequencies and our perception of the sound quality will
then be that it is too sibilant or too harsh or lacking in
bass. The rolloff curves shown in Fig. 34-76 were
developed to avoid this problem.
The experience of those engineers who have experi-
mented with direct-sound-only equalization using TEF
analyzers (not real-time analyzers) supports the idea
that we judge the response of a system based on the
direct sound and not on the direct plus reverberant
sound. The discussion of the need for roll-off in the
previous paragraphs also supports this concept.
One additional reason to roll off the high-frequency
response during equalization is to compensate for the
presence boost that exists in many microphones. Listen
to the system using the house microphone to see if any
additional roll-off is needed for this problem.

34.5.2.2.10 Use of High- and Low-Pass Filters

A 12 or 18 dB/octave high-pass filter, at approximately
50–160 Hz, will enhance the performance of a
voice-only system by filtering out unwanted low-fre-
quency transients like dropped microphones and breath
pops. For a music or music-plus-voice system, use a
high-pass filter at 20 Hz or above. Most music systems
are actually improved by a high-pass filter at 40–80 Hz.
In addition, a vented-box-type, low-frequency enclosure
should be high-passed at a frequency slightly below the
Free download pdf