Page 191 - Master Handbook of Acoustics

FIGURE 10-4 A demonstration of comb filtering in which direct sound from a loudspeaker is

acoustically combined with a reflection from a surface at the diaphragm of an omnidirectional
microphone. (A) No surface, no reflection. (B) Placing the microphone 0.7 in from the surface
creates a delay of 0.1 msec. (C) A delay of 0.5 msec creates cancellations much closer together. (D)
A delay of 1 msec results in cancellations even more closely together. A linear frequency scale is
used.

In Fig. 10-4B, the loudspeaker faces a reflective surface; the microphone diaphragm is placed
about 0.7 in from the reflective surface. Interference occurs between the direct sound the microphone
picks up from the loudspeaker, and the sound reflected from the surface. The output of the microphone

shows the comb-filter pattern characteristic of a 0.1-msec delay. The combination of the direct and
the reflected rays shows cancellations at 5 and 15 kHz and every 10 kHz.
Placing the microphone diaphragm about 3.4 in from the reflective barrier, as shown in Fig. 10-
4C, yields a 0.5-msec delay, which results in the comb-filter pattern shown. Increasing the delay to
0.5 msec has increased the number of peaks and the number of nulls fivefold. In Fig. 10-4D, the

microphone is 6.75 in from the reflective barrier, giving a delay of 1.0 msec. Doubling the delay has
doubled the number of peaks and nulls. If t is taken as the delay in seconds, the first null is 1/(2t) and
spacing between nulls or between peaks is 1/t.
Increasing the delay between the direct and reflected components increases the number of
constructive and destructive interference events proportionally. Starting with the flat spectrum of Fig.
10-4A, the spectrum of B is distorted by the presence of a reflection delayed by 0.1 msec. An audible

response change would be expected. One might suspect that the distorted spectrum of D might be less
noticeable because the multiple, closely spaced peaks and narrow nulls tend to average out the
overall response aberrations.
Reflections following closely after the arrival of the direct component are expected in small rooms

because of the short delays created by the room. Conversely, reflections in large spaces would have
longer delays, which generate more closely spaced comb-filter peaks and nulls. Thus, comb-filter
effects resulting from reflections are more commonly associated with small-room acoustics. The size
of various concert halls and auditoriums renders them relatively immune to audible comb-filter
distortions; the peaks and nulls are so numerous and packed so closely together that they merge into
an essentially uniform response. Figure 10-5 illustrates the effect of passing a music signal through a
2-msec comb filter. The relationship between the nulls and peaks of response is related to several

musical notes as indicated. Middle C, (C ), has a frequency of 261.63 Hz, and is close to the first null
4
of 250 Hz. The next higher C, (C ), has a frequency twice that of C and is treated favorably with a
5
4
+6-dB peak. Other Cs up the keyboard will be either discriminated against with a null or favored
with a peak in response, or something in between. Whether viewed as fundamental frequencies or a
series of harmonics, the timbre of the sound suffers.