
Let’s say the sound system in the house of worship you’re working on goes into feedback whenever microphones pass under the loudspeaker array.
Worse yet, there are “soft spots” in some sections of the audience area.
Choir mics “squeal” before they are loud enough and the podium mic “rings” annoyingly for some presenters. You know that the system should be equalized to eliminate these problems.
So you install an equalizer and the feedback is reduced, but the soft spots persist and the system just doesn’t sound good.
But that’s why you, the consumate audio profesiponal, are there.
After some careful listening tests, a “problem area” within the room is chosen for the measurement mic placement.
This is a place in the seating where people complain that they can’t hear, or a place where the mic consistently goes into feedback, such as directly under the loudspeaker array. The measurement looks something like that shown in Figure 1.

The response clearly shows an acoustic “comb filter” that results from a time offset between two sound arrivals at the measurement position.
The measurer first makes certain that the secondary arrival isn’t simply the result of a bad mic placement (floor bounce, etc.) or loudspeaker placement (ceiling or wall bounce, etc.).
After ruling out these two possibilities, it becomes apparent that the multiple arrivals are due to the overlapping patterns of two loudspeakers being used to provide audience coverage.
Standing at the mic position and simply looking at the array, noting that you are clearly within the coverage pattern of two loudspeakers suspended over the stage, confirms the suspicion. Sound travels at a single constant speed.
Yet, in this case, there are two loudspeakers.
Therefore every location in the room that is receiving direct sound at equal level from both loudspeakers (except for the center line where the distance to each loudspeaker is exactly equal), will receive two signals arriving at different times.
This time offset causes the comb filtering.

An acoustic comb filter can produce undesirable coloration of the sound and loss of definition. It can even change where the sound seems to be coming from, ruining the “imaging” of the system.
The possible “options” are:
1. Set the analyzer resolution to smooth the comb filtering, and then adjust the equalizer for the desired response. This is not a solution. It just masks the problem.
2. Ignore the comb filtering and simply “notch” the frequencies that are prone to feedback. Even though this is a common approach, it is treating the symptom and not the problem.
Excessive frequency notching can ruin the sound of the system. Why filter out sound that needs to be there?
3. Conclude that humans aren’t all that sensitive to narrow notches in the spectrum, so the comb filters are just something that we can live with.
This is rationalizing the problem and is simply not true. It’s usually the explanation provided by someone who is responsible for the problem in the first place!
4. Get out the old one-third octave real-time analyzer. You can’t see the comb filters on it.
For many years, audio professionals did not have high-resolution analyzers that could identify arrival time problems. The system response looked fine on a one-third octave analyzer, but it still sounded bad.
Today’s analyzers are vastly more powerful and can reveal much more about the nature of a sound problem.
5. Inform the owner that the current loudspeaker placement has created some problems that cannot be “corrected” electronically. The only real solution is to relocate the existing loudspeakers or redesign the array.
Unfortunately, the sad reality is that only the last option is likely to fix the system.
An acoustic comb filter is a symptom of a more significant problem. When two loudspeakers are placed in close proximity, the resultant distance offset will cause “lobing” in the speaker’s radiation pattern.
Lobes can be described as “fingers” of sound pressure “maximums” in the three-dimensional space surrounding the array.
The fingers are separated by nulls or axis of minimal sound pressure level. The fingers typically cause problems with microphones, since a mic is likely to feedback when it is placed within a lobe.
The nulls cause problems for the audience, since parts of the audio spectrum that are critical for speech intelligibility (understanding the words) are cancelled at some listener’s seats.
When a series of these lobes and nulls exist, the visual representation of the frequency response at one listener position will resemble the teeth of a comb, with a sequence of peaks and valleys.
This is a far cry from the “perfect” system response that would look more like a flat line. As such, a comb filter is the symptom of a spatial problem that has resulted from a loudspeaker selection and placement choice.
To illustrate, look at the simulations shown below (Figure 6), which show such a condition performed with the EASE sound system design software package.
Two loudspeakers with low directivity control have been separated by two feet.
The resultant does not represent accurate sound reproduction and can cause the afore-mentioned problems with acoustic gain and speech intelligibility.
Please note that it is certainly possible to build quality “arrayable” loudspeakers, and there are a number of good examples in the marketplace. However, all of them have several parameters in common:
1. Large physical size
2. Horn-loaded components
3. Aggressive pattern control to minimize interaction with adjacent loudspeakers
If these loudspeaker requirements present problems for a particular venue due to the required large physical size, then smaller loudspeakers can be used (usually in greater number) if they are placed sufficiently close to the listeners (i.e. exploded arrays or distributed systems).
Figure 2 (on page 38) outlines the options, and there aren’t many.

Radio broadcast engineers have understood for years the importance of proper antenna array design to control lobing in RF radiation to steer their signal to certain areas within the listening range and away from others.
For instance, if a station is licensed to radiate 50 killowatts of power, they can use an antenna array to steer the radiated signal up and down an interstate highway rather than out across a sparsely inhabited area. In fact, if they do it wrong, they can be in violation of federal law and therefore subject to prosecution.
Loudspeaker array designers must work with the same physical laws and principles as antenna designers. The only difference is that they can’t be prosecuted for bad sound.
Balloon plots are useful because they show the three-dimensional radiation pattern from a loudspeaker or group of loudspeakers located at the center of the balloon.
The plot describes what is happening at a single frequency. The plots can be generated for multiple frequencies to more fully describe the performance of an array. The balloon plot of a “perfect” loudspeaker would be the same, regardless of frequency.
Comb filtering in the magnitude response (a measurement at a single point in space) is evidence of lobing in the spatial radiation of the array.
Pat Brown teaches the Syn-Aud-Con seminars and workshops. Synergetic Audio Concepts (Syn-Aud-Con) has been a leader in audio education since 1973. With nearly 15,000 “graduates” worldwide, Syn-Aud-Con is dedicated to teaching the basics of audio and acoustics. For more information visit their website.