Vertically arrayed speaker deployments increasingly rule the day in pro audio. Such systems have advantages in deployment, sightlines, and dividing the audience into different coverage zones. This month’s tech feature is the second in a series that will detail real-world complexities involved in deploying vertical arrays and practical approaches that the system technician can apply to ensure even coverage throughout the audience area.
So in this second part of the series, we dive directly into approaches for aiming vertical arrays and for controlling their coverage utilizing electronic processing. If you are new to the topic, I would encourage reading Part 1 in the September 2014 issue of FRONT of HOUSE. It sets the backdrop for everything going forward. We will now briefly recap a couple key concepts from the first article, and then dive into a field-deployable array technique.
Array Behavior — A Recap
If we were to distill our launch-off point for this article, the following two paragraphs cut straight to the heart:
“The first physical behavior we can control… is the physical location and aiming of the boxes. Aiming of the loudspeakers is both a powerful tool and a limited one. The concept that the lows, mids and highs emanate from each vertical array box in narrow beam pointed at a specific region in the audience is not supported by physics. At high frequencies (i.e., above approximately 3 kHz), where sound has wavelength smaller than, or comparable to, the waveguide dimensions of an individual box, the loudspeakers’ inherent directivity is effective at directing sound at the desired location in the audience. However, as frequencies get lower, and wavelengths get longer, each box has less individual directivity and directional control becomes an ever increasing function of the total array.
“Array processing decisions, except at very high frequencies, cannot be made in isolation. For example, if system techs require more than 300 Hz in the audience area where the third box from the top of the array is pointing, they cannot simply boost 300Hz on that box in the array and expect to create the resultant boost in that section of the audience. The level of 300 Hz for that audience region can be influenced, but it requires system techs to consider the processing of all array elements as a whole to achieve the desired result.”
These are the guiding principles of all array processing, regardless of type or configuration. At high frequencies, where the wavelengths are small, the location of the speakers and how they are aimed effectively shapes the coverage pattern. Thus, at high frequencies, changes in processing to individual transducers, loudspeaker enclosures, or zones of an array can be made with relative impunity. These changes will influence acoustic output where the enclosure is aimed with proportionately small influence on other areas of the coverage pattern. By contrast, the behavior of the entire array must be considered at low and mid frequencies. Processing on any given enclosure (or array zone) must be evaluated in the context of the overall array directivity behavior over the entire coverage zone. It is important to keep these big concepts about array behavior close at hand as we investigate various techniques to improve array performance.
The (Changing) Main Lobe
In this article, we are focusing on a particular effect where the shape of a vertical array coverage lobe, and the nominal aiming direction of the coverage lobe, changes with frequency. This effect is illustrated in Fig. 1, where the array coverage pattern at 8 kHz and 200 Hz is displayed side by side. Here all of the boxes have the same processing, and the angles for the array have been chosen to provide smooth coverage distribution over the audience area.
Going High-End
Several key features about the high frequencies emerge from this Fig. 1:
• The array is large with respect to wavelength, and provides a tight pattern distribution.
• Each vertical region along the array has a portion of the high frequency lobe directed at an area of the audience.
• The intra-array angles are meaningful in describing the high frequency coverage pattern.
• There is good control over the high-frequency aiming axis.
The Low-End
By contrast, the low-mid frequency behavior exhibits:
• A softly defined pattern distribution, as the array is not as large with respect to wavelength.
• One primary array lobe emerging approximately from the center of the array.
• Little influence on lobe shape or aiming from the intra-box angles.
• There is less control over the low-mid aiming axis.
Clearly, the low-mid aiming axis does not match well with the high frequency lobe aiming.
Back to Physics
The differences in behavior boil down to wavelength. At high frequencies where wavelength is small, the boxes have narrow directivity. The aiming of boxes by the flyware sets the pattern effectively. The wavelength at 8k Hz is 1.7 inches (4.3 cm). By contrast, the wavelength at 200 Hz is 67.5 inches (171.5 cm). This wavelength is bigger than an enclosure, and each box has little directivity control over this long wavelength. The array collectively has directivity at 200 Hz, but since it has a finite height, there is less overall control than at 8k Hz.
Further, the aiming of each box does comparatively little to control directivity in this range. Since a box has little directional control at 200 Hz, the angle of each box matters little to the overall coverage pattern at this frequency. Instead, what has some influence on directivity is the overall curvature of the array. Even here, the array’s total curvature is not likely to have much effect on the shape of the coverage lobe at 200 Hz. This is because physical array curvature affects the arrival time of sound from each portion of the array, but the arrival time difference caused by the physical spacing introduced by intra-array angles is comparatively small.
Bringing Coverage in Line
Generally, the aiming of the low mid lobe is lower than the high frequency aiming. What we would like to undertake is raising the aim of the low-mid lobe so that it falls more in alignment with the high frequency aiming. We do this by manipulating the arrival time of parts of the array, “pulling” the coverage lobe in the desired direction (i.e. upwards). Specifically we want to cause the low-mid frequencies from the top section of the array to arrive later in time than their physical aiming allows. This will draw the aiming of the low mid lobe to be more in alignment with the high frequency coverage.
Fig. 2 shows the result of what happens when we try to do this by dividing the array from Fig. 1 into three zones, and blindly applying delay. In the left-hand side of the image, we see that the low-mid coverage lobe has been tilted upwards to an aiming that better fits the high frequency coverage. However, the right side of the image shows we have paid penalties in the upper midrange coverage pattern. This is because the extra delay we added has also shifted the upper midrange arrival times in a non-advantageous way. What we need instead is to shift the arrival times in a frequency-dependent way, so that we influence the low-mids without causing harm to other frequency ranges.
We can implement this straightforwardly using a type filter called an allpass filter. Allpass filters influence the arrival time of signals in a frequency dependent way, but they do this without changing the frequency response. Technically speaking, these filters have changing phase response, but unchanging magnitude response. Allpass filters used to be somewhat exotic, but are now readily available on most high end and/or midrange digital signal processors typical of those used with vertical arrays.
With the allpass arrow in our digital quiver, we can now manipulate the arrival time of the array in a frequency dependent manner for each of the three zones. The results are shown in Fig. 3. The low-mid coverage (far left) has been tilted upwards, the upper midrange (center) coverage has been left untarnished, and the high frequency coverage (far right) is as before. By enacting this processing, we have achieved a more even front-to-back spectral balance for the audience area, one where the front of the audience has less low-mid buildup, and the back of the audience receives more low-mid energy. Further, we haven’t changed the overall output capability of the array, as we have only changed the arrival time of each zone, and not the frequency response.
Conclusion
There are other ways this aiming could be implemented, but allpass filters are the most straightforward approach. The aiming can also be supplemented by using dedicated low frequency enclosures atop the array, and influencing their arrival time to further support the overall lobe shape. It is important to note that no amount of rigging or flyware fiddling can provide the large amount of physical delay necessary to implement this re-aiming at low frequencies. Further, because each section of the array has different low-mid arrival times, it requires zoned processing that splits the array into multiple sections.
To change the directivity of the array at lower frequencies, we have influenced the arrival time of sound from across the entire array height through the use of digital filters. The aggregate result is a re-aiming of the primary low-mid lobe to correlate better with high frequency aiming decisions. The result is a more consistent front-to-back coverage for the audience and an easier day for the system tech.
