Choose Your Battles: Examining Dispersion & Pattern Control In Line Arrays

Inverse Proportions

Single line array elements in the vertical plane are so-called “Proportional Q” loudspeakers (Figure 6). Their coverage angle is inversely proportional to frequency. From omnidirectional in the low-end all the way to their nominal coverage angle, as little as 5 to 15 degrees at 16 kHz.

There’s a rule of thumb for piston drivers, responsible for the low end, that states that at the frequency whose wavelength matches the driver diameter, the coverage angle will be roughly 90 degrees (axisymmetric). For any array of this length, driver complement and physical configuration, this implies that the entire “problem” region (160 Hz to 1.6 kHz) surrounding the beaming frequency is joint custody of all loudspeakers that each exhibit coverage angles of 80 degrees or more. There’s little to no isolation (level separation) from neighboring sources!

In the absence of isolation, EQ or level adjustment have limited merit. Lowering the level of a single module either by gain or EQ has primarily a global effect, affecting the entire coverage area.

Figure 6: Vertical coverage angle of a single 9-inch line array element.

The tables in Figure 7 show that attenuating the B module exclusively does nothing for the grand total other than reducing overall headroom. Eliminating the B module all together only lowers the global level by five-eights (or -4 dB) at the most.

Figure 7: Diminishing returns.

However, if there’s level separation, EQ will have a more profound effect. Figure 8 shows the effect of a notch when competing with another source. The red trace represents the sum of both signals. It’s readily apparent that the notch becomes more effective when the other source loses market share.

Note that this also applies very much to mixing. In the absence of crosstalk or “bleeding” – ergo separation – EQ has a profound and tangible impact. The equalizer will feel very responsive. Contrary, tons of crosstalk will require more drastic measures that are typically accompanied by detrimental side effects. The latter very much applies to our array as well!

Figure 8: EQ versus level separation (PEQ at 1 kHz)

Figure 9 depicts what happens to the beam should we decide to resort to EQ in the B module exclusively. It’s readily apparent that attenuating the offending frequencies has no profound effect on the beamwidth. In fact, we observe the onset of a vertical interference pattern featuring power alleys and valleys.

If we’re not careful, it’s will be as if the entire B module has been eliminated all together (Figure 9, Plot 4). This essentially changes a coupled point source into an uncoupled point source, where the physical displacement has serious implications.

The uncontrolled spurious side-lobes seen in Figure 9 could blow up in our faces. What if they end up hitting specular surfaces (e.g., a balcony face or rear wall), introducing discrete echos? The audience on a balcony in the custody of another system? Environmental pollution?

Figure 9: Attenuating beaming frequency with EQ (in zones).

The principal precept in health care is: “First do no harm.” Unfortunately, not everything can be simply remedied with a “level band-aid” like EQ or gain. In this instance, that approach clearly has limited merit and is likely to cause more harm.

If these loudspeakers are virtually equally loud, because in this part of the spectrum they are still omni- to hemispherical and fail to steer clear of each others’ territories (immune to rotation), so they better arrive in time. This is the root cause for this beaming phenomena.

Figure 1 clearly shows path differences. The difference between these trajectories translate into phase offsets and that’s when “stuff hits the fans” (plural, and pun intended). There’s a time problem that requires a “time band-aid” and the reason it’s so audible is because the levels are matched!