Study Hall

Similar Experience: The Quest For Minimum Variance In A System

As engineers, we need to go one step further and not only get the sound level up, we need to keep it consistent from seat to seat, or technically from ear to ear, of the audience members. Here's a way to go about it.

The purpose of sound reinforcement, or amplified sound, is to allow everyone at the event to hear and understand the content better than if it was a pure acoustic source. As engineers, we need to go one step further and not only get the sound level up, we need to keep it consistent from seat to seat, or technically from ear to ear, of the audience members.

There is really no other way to do this. There is no good way to create special or “better” seats in an open room with a sound reinforcement system. Even if you put expensive monitors on a shelf right in front of a seat, it might not sound good with the leakage from the rest of the system.

What we should be aiming for is creating what’s called “minimum variance” of SPL (sound pressure level) across the audience. If done correctly, every person will have a similar experience and that is what they are paying for.

Stereo effects are laid aside for this discussion; something panned, also known as “uncorrelated,” is something that takes additional planning, and when chosen to be used, automatically introduces severe limitations and will compromise your knowledge of who is hearing what in what position.

Near & Far

So what is our measurable goal for this minimal variance of SPL, anyway? Different companies will have different standards, I usually work to keep the variance under a change of 6 dB. This means that the seat most to the side of the loudspeaker, the seat directly in line with the loudspeaker and the seat in between two loudspeakers must stay within tolerance. My discussion here will show how to calculate, design and verify this is happening.

Keep in mind that the seat closest to the loudspeaker and the seat the farthest from the loudspeaker should remain within 6 dB as well. In my example, I didn’t have time to verify the front to back variance with the analyzer but I will show you how to calculate it with a laser measure and a math formula. The answer is approximate.

My example: a client that had two projector screens and four loudspeakers in a ballroom for a dinner event. The ballroom was 134 feet (40.8 meters) long and at its widest point it was 72 feet (22 meters).

Using my room mode Excel calculator, I found that the optimum coverage angle for each of the four loudspeakers would be 55 degrees. This would result in minimal coverage overlap or gaps – the loudspeakers I had access to were 75 degrees but had less than 3 dB of overlap, which is within tolerance. (Remember that angular dispersions typically describe the loudspeaker’s high frequency dispersion, and the low frequency range is usually wider.)

Each loudspeaker had to be 33.5 feet (10.2 meters) apart. The stage was in the middle of the room, with Figure 1 showing how they were placed and spaced.

Figure 1.

This was the final design and without too much trouble I was able to set up the loudspeakers like this onsite. The reason the outside ones are farther forward is because the ballroom wasn’t a rectangle and the walls sloped in.

Checking The Work

Next came the verification process. To start, I took a measurement of the frequency response of the left-most loudspeaker, with the microphone at the left-most seat, mid way back. This is called OFFAX (off-axis, shown in Figure 2).

Figure 2.

The next position was the same depth but directly in the line of the left-most loudspeaker. This is called ONAX (on axis, shown in Figure 3). Normally I would start with this mic position, but the tables were being set up and I had to work around the hotel staff. The green trace is still OFFAX and the red is the ONAX compared. There is very little variation between each, which is great.

Figure 3.

The next position is called XOVR (crossover, shown in Figure 4). This is the position equidistant from two loudspeakers, and depending on the amount of overlap, there could be cancellation or addition in SPL here. Addition is fine, but it needs to be within tolerance.

Figure 4.

There are a couple of different things to think about when dealing with an XOVR position. The first is to take a measurement of each loudspeaker by itself, without moving the mic, to compare angular coverage. After a few adjustments of loudspeaker aim, I got each one to match in response at the XOVR individually (Figure 5).

Figure 5.

The thing to consider at this point, with this setup, is that the outside loudspeaker’s sound will be arriving sooner than the inside – simply because it’s closer. We need to take the difference in arrival time given by the analyzer, delay the outside loudspeaker on the console, then run the sweep with both of them on to see if there is addition of SPL across the spectrum. The gold line shows that we do indeed have summation of SPL all along the frequency axis (Figure 6).

Figure 6.

The process then continues down the line until all positions are measured. There are 4 ONAX, 3 XOVR and 2 OFFAX positions. This is the minimum amount of measurements needed to verify this design. (You can take more if time allows.) Figure 7 shows the composite of all seven measurements.

Figure 7.

As you can see, in the low end, there is more than 6 dB of variance. This is actually okay because our ears aren’t as sensitive to that range as they are to the high end, especially around 3 kHz. From about 1 kHz to 5 kHz is where to be the most strict about 6 dB of change. (It’s pretty well covered in this case.)

Another Aspect

Armed with this information, I was happy that across all seats, everyone was hearing the output of the system very similarly.

As far as the front to back, here’s the other way to approximate it. Take the distance from the grille of the loudspeaker to the closest seat ONAX. Then find the distance to the farthest seat. In a regular calculator type 20*LOG (Distance one/Distance two). This is the SPL drop in dB.

There’s another design aspect I didn’t mention. The color palette of the SPL graph showing mic positions and loudspeaker layout was not an absolute scale. It was actually showing relative level. Everything that is green is within 6 dB of level at the displayed frequency. Since tables and seats are not shown, you must verify onsite.

This introduction is fairly brief and only considers loudspeakers on stands. When you’re talking about flown loudspeakers, line arrays, subwoofers and seating with elevation, this all gets more complicated.

The goal, however, is always the same: that the audience receives the most consistent coverage possible. I did get a compliment on how good it sounded from the client, who was busy walking around the room all night. I hope they noticed that it sounded good everywhere!

Study Hall Top Stories