A look into main loudspeaker system trade-offs involved in dealing with balconies, accompanied by design examples...

Previously, I addressed the impact of breaking and tapering a line of loudspeakers (here). Now let’s turn our attention to balconies.

Everybody seems to love breaking main loudspeakers horizontally into left and right, but breaking them vertically into upper and lower? Not so much.

We can be repeat offenders when it comes to multi-main horizontal breaks, such as a parade route or racetracks.

Multi-mains can go vertical as well, although it’s unlikely that we’ll expand beyond two elements.
Vertical expansion is also driven by room shape, in this case a very specific room shape: the balcony. Let’s look into the trade-offs involved in battling the balcony.

We’re talking about left/right (L/R) mains. There’s no such thing as a lower center (that’s where the band is). Center mains solve balcony coverage with L/R side fills near the deck and lots of delays. The question for the left main is when to surrender to uncoupling. We all want the band to stay together, but sometimes the members need to go their separate ways (Figure 1).

 
Figure 1: Example applications that question whether or not we need to split the mains into upper and lower sections. The same main loudspeaker height and balcony shape appears in all three examples.

Single & Double Slopes

The typical listener plane is a slope rising with distance. We solve this shape with the asymmetric coupled point source. The simplest listener plane is a constant slope, a consistent rise over distance. We commonly encounter more complexity, with a steeper slope in the rear than front. The coupled point source can adapt to this shape by complementary asymmetry, even with very substantial differences in rate of rise.

Balconies add a second listener plane, which is where the trouble starts. We are now double-sloped, a shape that calls out for uncoupling. There are two primary strategies: treat the shape as a single complex slope (and stay coupled) or treat them as distinct slopes and solve them separately (uncoupled).

We can keep the main array coupled by plowing a line of best fit through the balcony details. The downsides are level variance (balcony front will be louder) and ripple variance (balcony front reflection). The upside is the extended frequency range of the coupling zone, the result of keeping all sources close. A deep balcony increases the level variance. A tall, reflective balcony front increases the ripple. That’s what we’re up against.

How deep is too deep? How do we know when the front will give us trouble? We’ll be able to wrap our heads around balcony acoustics once we see how acoustics wraps itself around balconies.

The double slope has four coverage target milestones: VTOP1 (vertical top), VBOT1 (vertical bottom), VTOP2 and VBOT2. Each has a unique angle and range relative to the mains, and each pair has a unique angular spread and range ratio relative to each other. It’s the inner pair’s relationship (VBOT1-VTOP2) that has the twist. Let’s plug in some numbers and see the results (Figure 2).

 
Figure 2: Example applications with double sloped vertical shapes. Multiple shallow balconies present a nearly symmetric (1:1) range ratio while the floor is highly asymmetric (>2:1). Coverage can be divided between the two slopes.

We start with a matched pair, two identical slopes, stacked directly on top of each other. Each has a 20-degree spread and 2:1 range ratio. Our loudspeaker is in the middle so it covers from VTOP1 (+20 degrees) to VBOT2 (-20 degrees), with a 2:1 range ratio (6 dB). In between are VBOT1 (+0 degrees) to VTOP2 (-0 degrees), which also have a 2:1 range ratio.

We can solve a 6 dB range ratio spread over a 40-degree angular spread by aiming an 80-degree loudspeaker at the uppermost seat. How do we solve a 6 dB change that happens in a 0-degree spread? OK, it can’t be 0 degrees because the balcony has to be thick enough to hold people, but it can be very, very small.

Return Ratio

“Listen up, main! Stick together. Here’s our mission: Go deep then gradually come closer for 20 degrees and then instantly go deep again and repeat.”

If you’re not convinced yet that this is mission impossible, then add range ratio until you surrender. A wider balcony front gives us more angle to work with, but with friends like this, who needs enemies?

Let’s make a single modification to the previous shape and do the exercise again. Slide the upper floor backwards so its front aligns with the lower floor’s rear.

What’s different? VBOT1 and VTOP2 are still both at 0 degrees, but they now have a 1:1 range ratio. There’s no longer a zig-zag in the middle. We would surely cover this with a single main.

It’s also not a balcony any more but it reveals the mechanism, the return ratio, the primary indicator for splitting the array.

Every inch we slide the upper floor forward increases the discontinuity between VBOT1 and VTOP2. Such sharp turns in coverage require angular isolation and we don’t have it. Return ratio (in dB) quantifies the level difference the balcony forces us to overcome.

We can saw a line of best fit through a shallow balcony with a small return ratio and keep the array together. Return ratios of 6 dB or more cannot be smoothed over (Figure 3).

 
Figure 3: Upper/lower decision examples. In all cases there is 20 degrees of coverage required above and below the speaker location. (A) Single main can cover the continuous slope. (B) The 3 dB return ratio does not splitting. (C) The 6 dB return ratio indicates splitting is best. (D) Underbalcony loudspeakers reduce the return ratio to 3 dB (no splitting required).

Secondary Options For Upper/Lower Mains

There are still options short of breaking up. We can outsource coverage to others, specifically the underbalcony area, which can be covered by delays. The area covered by the delay is taken off the custody requirements of the main. VTOP2 moves closer, reducing the range ratio and opening up angle, a double bonus. We may be able to tough it out as a single main if the delays can bring the return ratio in bounds.

The height of the mains also plays a role. We’ve looked at them in the middle. Going upward reduces the angular spread between VBOT1 and VTOP2 (as if it wasn’t small enough already). Going higher leads to occultation (the blocking of the sight line to the loudspeaker) underneath, which reduces return ratio by coverage reduction.

Delays, however, have moved from optional to mandatory. Occultation seriously downgrades the underbalcony area and should not be considered fair trade for return ratio gains. Moving the mains under the balcony line ensure sightlines to the back and opens up the angular spread (the mains can see the underbalcony ceiling now).

Return ratio shows no improvement and balcony coverage will become more challenging. The upper level slope is flattening (from the mains POV). The upper level needs more severe coverage shaping due to reduced angle and rising range ratio. Such severe asymmetry is difficult with a single slope. Asking the mains to do that upstairs and on the floor is a very tall order, especially if it’s a tall balcony.

There’s one more height-related consideration: the coverage pattern bending often termed “smiles and frowns.”

This is relevant to high and low orientation to balcony fronts (and our attempts to avoid them).

Only a vertically centered main can be precisely steered to overcome a high return ratio.

Upper or lower positions cause the coverage transition to appear at different heights across the room.

Here’s the consolation for this exercise. All is not lost when we raise the white flag and divide the array into upper and lower sections.
Instead of trying to make one array do something it hates to do (double sloping), we get two arrays being what they love to be: asymmetric coupled point sources drawing a single slope. The ripple variance in the low end might be a very small price for this payoff.

The design process is basically a two-layer cake. Upper and lower are separately analyzed and comparably power scaled.

In some cases it’s possible to move the upper main deeper into the house (since it starts at the balcony). This is free money in terms of power and signal/noise ratio, as long as the image is not compromised.

Balcony Fronts

There’s an old expression to sum up the sound engineer’s perspective on this: The only good balcony front is a dead balcony front. “Balcophobia” is a serious malady in the sound community. Designers go to great lengths to avoid what acousticians go to great lengths to install: lively balcony fronts. We’ve all been burned by this, so it’s worth a few paragraphs to put things in proper perspective. Some balconies are poisonous but others are harmless, but many of us run from both kinds.

Bad balconies are tall, lively, featureless (single angle, not diffuse). Glass and steel (bad). They also have lousy angles with respect to our loudspeakers, sending sound back on stage or onto paying patrons. Even worse (the worst) is the flat, curved balcony in a fan shaped room with the stage as it’s focal point. Been there, done that.

Good balconies are short, dead, diffuse, filled with lighting gear, multi-angled and inclined to send our sound harmlessly into the open air such as the ceiling. Size up the live surfaces of your local balcony. It’s harmless below 500 Hz if it’s less than 0.5 meters tall. The key is to not to hurt the design over avoiding something that won’t hurt you (Figure 4).

 
Figure 4: Upper/lower main design examples. (A) U-Balc reduces the return ratio enough to use a single main. (B-D) Upper lower mains are used.

Bob McCarthy has been designing and tuning sound systems for over 30 years. His bookSound Systems: Design and Optimization is available at Focal Press. He lives in NYC and is the director of system optimization for Meyer Sound.