A dartboard of 30 things I’ve learned in 30 years of being hyper-focused on one thing...
August 07, 2014, by Bob McCarthy
It was 1984 when I first sat in front of an FFT (Fast Fourier transform) analyzer during a concert and made decisions about tuning sound systems. I’ve been doing it ever since, and have written a lot of articles and a few books about this, but these are usually very focused and detailed.
This time I’ll go a completely different direction: a dartboard of 30 things I’ve learned in 30 years of being hyper-focused on one thing. No proofs or explanations. Just the ramblings of the old man of the FFT.
1) Make sure your analyzer is working right. The old saying “physician heal thyself” applies here. Be certain you have a working diagnostic tool before you say anything about the sound system.
2) Make sure the client knows what you can (and can’t) do. Make it the same is the big theme here.
3) Find out what can (and can’t) be done. Can the loudspeaker angles be adjusted? Can we break out the bottom box on its own DSP channel? Will this heritage hall allow us to cover the walls with fiberglass? Of course not! Can we hang the PA lower? You will never know if you don’t ask.
4) Don’t be a jerk. Be inclusive and collaborative. This is a team sport. Don’t embarrass or humiliate anyone. If possible, find a way to make everybody look good when major changes are needed. You can save the day after proving that the design or install is a total fail but you will never work in this town again if you humiliate the client while doing it.
5) Working with acousticians. They are experts in acoustics but don’t assume they understand sound system optimization. Their perspective doesn’t begin until the sound starts hitting the walls. It’s a statistical world of reflection paths, cavities and resonances. There is nothing statistical for us. It’s one-on–one relationships of speakers to particular surfaces. A statistically minor surface in exactly the wrong place is a huge deal for us.
Careful where you point that thing (the measurement mic).
6) Sound designers and mix engineers. These folks have great ears and an artistic vision. Listen closely to their wants and incorporate them into the tuning as much as possible. Let them know when they ask for the impossible so that a realistic plan can be put forward.
7) Get the stupid out of the system. Did the installer follow the plans? Is the loudspeaker path blocked? HF driver polarity reversal? Why is left side different from right? There’s no glory in going through a thorough step-by-step verification, but plenty of shame in EQ’ing a loudspeaker that’s miswired. I would rather leave a system verified and untuned than tuned and unverified. Allow lots of extra time for tuning systems with external amplifiers. So many more opportunities for wiring errors, e.g., I found a 3-way system with LF amps driving the MF speakers (for seven years).
8) Quickly assess the physical. Time and resources are limited. Moving the array will take a big chunk, so figure out right away if that needs to happen. We can do front fills, surrounds, etc. while the main rig is down. Don’t wait on the physical stuff.
9) Keep it simple, stupid (K.I.S.S.).Take care of the big stuff and leave the fashion show for later. There are so many exotic tricks that can do amazing things at one location but very few that improve things for a wide area.
10) Advance planning. Have a flow chart, plan and section showing loudspeakers. Seating charts are nice for mic placement. Know how you will interface into the signal processing. Work with the designer to know how the subsystems are supposed to relate (e.g. the underbalcony feeds go with the center vocal system, not the L/R music).
11) Keep the magic out. If it’s too good to be true, it is. If the laws of physics are being violated, they aren’t. There is something else going on. Find it.
12) Use good ingredients. Any chef will tell you this. Good loudspeakers, good processing, good analyzer, good mics and (I’m pushing it now) good acoustics.
13) Have a plan A for every loudspeaker. The main covers row 9 to 23. Down fill gets row 9 to 3 and front fill row 1 and 2. Overbalcony delays take the top 3 rows. Plan A lasts until the first shot is fired.
14) Have a plan B for every loudspeaker. The mains can’t stretch all the way, so either we aim them down and make the overbalcony loudspeakers cover more rows or aim them up and extend the reach of the down fill. Now let’s see if the down fill makes it the front fill meeting point.
15) Have expectations. Visualize the trace that should come up on screen before you measure it. If it’s different, then figure out why. Were my expectations wrong, or is something else going on here?
16) Know physics. Be fluent in time/frequency and phase conversions. This is critical for everything to do with optimization. Be fluent in 20 log dB scale conversions to linear ratios. This governs loudspeaker transmission over distance (the inverse square law) and summation of sound sources. Understand summation. Be capable of comb scene investigation (CSI). Read the comb filter fingerprints to see whodunit. It’s all there in the linearly spaced peaks and dips. If we identify the correct reflection we might get the acoustician to fix it.
The sound system was all ready for tuning except for one small detail: Plexiglass blocking half the loudspeakers. Somebody is going to have to move (guess who?).
17) Prove the design. This is not a passive pursuit where we observe loudspeakers and turn processor knobs. Prove the loudspeaker aim, splay angle and uniformity of the transitions from zone to zone. If the design is disproved it can be improved (along with your ability to design the next one).
18) Use your time wisely. I don’t mind if most of the time budget is spent on aiming, splaying and spacing the loudspeakers. That’s where the big payoff is. The most effective uniformity aid is optimized speaker position. EQ is easy with good loudspeaker locations and can work well for a large area. EQ is also easy with bad locations but only helps a much smaller area of the room.
19) FFT analyzers must be dual channel with minimum 24 points/octave. Amplitude, phase, coherence and impulse response. It’s critical to have complete confidence in your analysis system.
20) Place microphones strategically. We’re looking for specific answers to specific questions. A mic in the middle (ONAX) does EQ. Mics at the edges tell us if it’s aimed right (top and bottom for vertical and side for horizontal). A mic between two subsystems tells us if splay angle or spacing is right.
21) Order of operations. Follow the optimization equation (A+B=A). Aim and EQ the mains (A). Aim/splay/space, EQ, level set and delay the subsystem (B). EQ the combination of A+B and move on to the next.
22) Slogans about EQ are BS. We’ve all heard “the best EQ is the least EQ” and “he who EQ least, EQ best.” I guess that’s why they don’t put EQ on mix consoles and DSPs anymore. My versions: “the best EQ is the best EQ” and “he who EQs best, EQs best.”
23) Don’t be the EQ police. If the client wants flat then make it flat (and as close to the same everywhere). If the client wants +10 dB in the LF range, then fine (and as close to the same everywhere).
24) Filter choices. Parametric EQ for single systems. Usually 3 or 4, never more than 6 and rarely narrower than 0.5 octave. Go after the three (or so) biggest issues that cover the widest frequency spans and audience areas. All of the little stuff is gone in two seats. Use multi-band shelving filters for combined system EQ to settle down the LF coupling. There are more exotic filters out there but I never find the need.
25) Be on the alert for latency gone wild. Devices that give different latencies in different configurations, or on different outputs, or just because. Measure all the roads that lead to Rome and check their latency.
26) Adding extra delay for precedence. I never do it. It’s not needed if the levels are set right, and it can cause destructive side effects (most notably turning up the delays).
This system measured really strange. HF is good, then next row it’s bad, and then good again. Magic? The answer was revealed by crawling into the dusty soffit and finding the loudspeakers blocked by steel struts (for years).
27) With measurement mics, the best-case scenario is high quality, high stability, high matching and large quantity. When I can’t have it all, I favor a pile of lesser mics over a single perfect one. They must be stable and level matched, but a dB here or there of frequency variation is OK. Moving a single (perfect) mic around opens up a huge tolerance for errors. There is huge benefit to multiple stable mic positions when making changes in a multi-stage array.
28) Point the mics at the loudspeakers. Within ±30 degrees is fine but everybody worries if they’re off 5 degrees, so it’s easier to just say OK and aim them just right.
29) For mic height, the key thing to remember is we’re looking to represent an area (not just a seat). I generally use standing head height to keep clear of nearby empty seat back reflections, and will move down to sitting when warranted (e.g., front fill). I also go ground plane when measuring empty flat floors that will later be populated. The logic is this: Place the mic at the height that closest resembles the future show condition.
30) Never let “Nightfly” be used to evaluate your work. If the PA sounds good with those types of tracks, there’s a serious problem. Hi-hat should not lead the band.
Bob McCarthy has been designing and tuning sound systems for over 30 years. His book Sound Systems: Design and Optimization is available at Focal Press. He lives in NYC and is the director of system optimization for Meyer Sound.