A TOUCH OF CLASS
Mar 1, 2000 12:00 PM, Bruce Borgerson
Meeting the sound reinforcement needs of the Sydney Opera House's OperaTheatre.
The Sydney Opera House is certainly one of the busiest, if not the busiest,performing arts centers in the world. On average, the facility's fivetheaters host more than six shows a day, every day of the year, requiring acrew of 20 sound technicians just to keep up with the multitude of audiorequirements for each show. Until recently, David Connor headed this staff,serving as the facility's sound and A-V technical manager before departingto serve as a consultant in sound system design.
Prior to leaving, however, he oversaw the design and installation of aneffective new sound reinforcement system in the Opera Theatre, the venue'ssecond-largest performance space. This was just one of many A-V projects atthe time, and Connor pointed out that the system upgrade process at thefacility is virtually never-ending.
"For example, concurrent to the Opera Theatre project, we were outfitting anew theatre while we were also in the planning stage for a self-containedrecording studio that will allow us to record performances being heldanywhere in the building," he said.
The new system for the Opera Theatre, a 1,547-seat, two-tiered room thatprimarily accommodates opera and repertory theater, replaced the originalsystem installed 25 years ago when the facility was new.
"In those days, sound was the little brother of the total production,"Connor said. "You'd run a few rain and thunder cues through a Revox, mixingat the stage manager's desk, feeding a small, inconspicuous house system,and that was pretty much it. Obviously, times have changed with soundbecoming more and more important to performance. Now, almost everyperformance involves some aspect of amplified sound; from subtleenhancement of opera to substantial amplification of a rock band for moderndance. This was the primary impetus behind the new system effort.
"We had reached the point where we were being asked to do much more thanwas originally conceived. We needed more of everything - more level, moredynamics and higher fidelity. Over the past five years or so, this need wasmet with temporary systems, but the results were often less thansatisfactory. Not only was sound quality less than optimal, but the theatreaesthetics were also negatively impacted.
"System design is becoming more complex and thankfully, more effective. Inthe old days, designers would just point a bunch of boxes at the audience,and the resultant sound quality was a bit of a lottery; it often varied toomuch with where you sat, and many times, the acoustics took the blame forpoor system design," Connor said. "I also don't believe that running a fewmodeling programs is really the definition of sound design either. Althoughmodeling programs are a real asset to the sound system designer, there is alot more going on that needs to be addressed. The main issue is how thesound system interacts with the auditorium and just as importantly, withitself and its immediate surroundings. As the arrays get larger, and theimpact of the auditorium increases with size, the sound you hear ischaracterized by the whole system design rather than the loudspeakerproduct you are using. Getting appropriate, consistent off-axis loudspeakerbehavior is as important in a liv!e auditorium as getting the direct field right."
With this in mind, Connor developed a specification for a new system quitepointed about performance expectations but not specific regarding theactual product to be used and then floated it to see what designs qualifiedsystems designers would propose. Over the years, he has done quite a lot ofwork as a consultant in electroacoustics, contemplating how to specifysystems and what makes them sound the way they do, particularly in largeauditoriums, and this experience manifested itself in this facet of theproject.
"In my view, one of the important factors that's been missing is realspecifications that effectively define what we hear in auditoriums," Connorsaid. "Typically, a system is specified in terms of frequency response, say20 Hz to 20 kHz, with a tolerance, and away you go, but that is only validfor one seat. It came to me that there's technology and there's what youhear, and until recently, the two weren't always related. Specificationswere often rather loose or irrelevant, particularly with regard to coveragein three dimensions. An installed system could end up meeting the spec butend up sounding horrible."
At the same time, Connor recognized that these two facets are starting tomove toward each other, especially in light of more widespread use of FastFourier Transform (FFT) measurements and a concerted effort to control whathe calls, the "direct-to-everything-else ratio".
"We are now starting to be able to measure roughly what we hear," he said.
In the simplest of terms, the main goal for this system (and most others,for that matter) was to ensure that the sound impinging on the audienceareas had technical integrity and that the sound hitting other parts of theauditorium and stage were minimal and even over the entire frequencyspectrum. This required a highly specific and controlled loudspeakercoverage pattern, particularly to get acoustic gain off the stage and tohelp stop reflections from bouncing around the room.
"My specification had several levels," Connor said. "Coverage was one, andby coverage, I mean that you can't just specify a certain sound pressurelevel. You need to specify a tolerance and the area in which that toleranceoccurs. Fidelity was another. In these terms, I'm talking about frequencyresponse, which should be flat and true. Then there's clarity orintelligibility, which is effectively a ratio of the direct sound from theloudspeakers to the reverberant sound arriving at the listener's ears.Sound pressure level requirements were also important to make sure thatthunder and rock bands could truly frighten the audience."
One other aspect that Connor emphasized as crucial is where sound energyfrom the system goes within the room other than the audience; what spectrumand level of sound feeds back to the stage, for example. This sets theavailable acoustic gain when using mics, which is particularly importantwhen omnidirectional radio mics are placed in an opera singer's wig. Suchaberrations as narrow band energy directed at the ceiling or walls and thenreflected back to the audience area also demanded attention.
"It's OK to get some bounce," said Connor. "I mean, that's the sound of anauditorium, but the problem is bad bounce. For example, with stackedsystems it's very possible that you will get 10 dB lobes in small bands,say at 1 kHz, heading up towards the ceiling and bouncing back down, andthus, your reverberant field is clouded. In other words, even though yourdirect sound may be quite good, the reverberant field combines with it, andwhat you end up with is a nasty sound.
"It was a fairly intensive performance specification. It all came down to asystem meeting these factors while being loud enough to handle a rock band,down to quite subtle amplification or enhancement of classical music.Enhancement is a much more acceptable term for people in the classicalgenre; they're sensitive to amplification issues, and this must be takeninto account."
In collaboration, Glenn Leembruggen of Elecoustics and Jands Electronicssupplied a proposal best meeting Connor's expectations. Leembruggen, nowwith Arup Acoustics, is active in systems design as well as acoustic andelectronic measurement both in the lab and on site. Much of his workinvolves the important interface between architectural acoustics andelectroacoustics. Jands is the Australian agent for JBL and a leadingAustralian systems contracting firm. Its focus was loudspeakers and theirspecific performance parameters and locations with the addition of tailoredprocessing. Remaining system elements would be largely retained with theexception of additional power amplification.
Leembruggen offered some specific views on system design that in many waysmatch up with Connor's vision, which he distilled to a focus on loudspeakerdirectivity.
"This can be achieved through two methods," he said. "The first is toconstrain the radiation using the angled faces of a structure, which is howa horn radiator operates at frequencies that are well above the horn'slower cut-off frequency. Another way is to form a line array of soundsources and harness and control the constructive and destructiveinterference that always exists when sources are separated.
"Two problems with line arrays are that they become too directional whenthe array becomes acoustically long, and they exhibit strong lobing effectsoff-axis when the interdriver spacing array exceeds one quarter of awavelength. These problems can be prevented using a combination of threetechniques. The first is to keep the array's acoustical length constantwith frequency by using low-pass filters (called tapering filters) toremove the outer drivers from operation progressively as the frequencyrises. The second is to make the spacing between drivers operating at thesefrequencies less than one quarter wavelength. The third is to adjust therelative broadband level of each drive unit.
"Currently, good control of radiation patterns is readily available above600 Hz using off-the-shelf horn devices, but if the direct-to-reflectedratio or acoustic gain is to be constant with frequency, then the samedirectional behavior must be provided below 600 Hz. This also ensures atruly flat frequency response at each listening position without theoff-axis boost in the low mids that often occurs with constant directivitysystems.
The proposal offered a main loudspeaker system largely located around thestage proscenium with five line arrays consisting of JBL Professional HLAseries loudspeakers and JBL midrange horn and low-frequency drivers. Thesearrays are divided into left, center and right zones, upper and lower.
About the crucial aspects of this approach, Leembruggen said, "There aretwo main electrically tapered arrays in the Opera Theatre that provide bothdirectional control and extremely high SPLs at the lower frequencies. Alow-frequency subwoofer array operates up to 160 Hz and is crossed over toa midrange array that operates up to 1 kHz. Each of these arrays isunusual. The subwoofer array is 16.5 feet (5m) long and consists of six JBL18 inch (450 mm) 2242H subwoofers spaced asymmetrically. The midrange arrayconsists of five JBL HLA 4895 low/mid horn elements and is 12.5 feet (3.8m) feet long. Both arrays use tapering filters set up in the JBL DSC260processor.
"By making the spacing between the drivers in the subwoofer arrayasymmetrical relative to the center of the array, the radiation pattern ismade to compensate for the different distance loss between close and farseats. It also provides attenuation of the energy that is radiated towardsthe ceiling by some 15 dB above 63Hz.
"In the HLA array, the low/mid elements are turned 90 degree, allowing theformation of two side by side individual arrays, one of the low-mid horns(with the 14 inch or 357 mm driver) and another of the mid-range horns(with the 10 inch or 254 mm driver). Turning the elements 90 degree alsoallows the narrower vertical pattern of the horn above 600 Hz to minimizeirradiation of the side-walls of the theatre. The low-mid array is crossedover to the high-mid array at 320 Hz. At frequencies above 600 Hz, most ofthe signal is fed to the innermost mid-horn, and the directivity of thearray is controlled by the directivity of the horn itself rather than thearray.
"The off-axis behavior of the HLA array was designed to match that of theassociated high-frequency horn, a JBL2352 fitted with a 2447J driver. Inthis way, compensation of the distance loss from near to far seats by theloudspeaker directivity was achieved over the full frequency range. The HLAarray could have been made even more directional, but at the expense ofconsistent frequency response across all seats."
Leembruggen suggested that distance loss compensation is not usuallyundertaken in the low and mid frequencies and is the key to obtaining aflat frequency response at each listener.
"The subwoofer and HLA arrays are located side by side so that lobing inthe crossover region is prevented," he said. "A JBL 2352 high-frequencyhorn is also located right next to the innermost mid horn of the HLA array,again to minimize crossover lobing.
"To confirm that our mathematical model was reasonably accurate, JBLProfessional made acoustical measurements of the electrically tapered HLAarray at Summit Labs. Because the array was to be flush mounted in theproscenium wall, we did not attempt to predict the radiation behind thearray."
Figure 1 shows Leembruggen's prediction of the array's polar pattern andthe measured pattern.
"Because the measured and predicted performances are close," Leembruggensaid, "the mathematical model was judged to be pretty good, and a furtherrefinement was made. The innermost horns were raised 6 dB in level,resulting in a smoother and more controlled polar pattern while slightlydecreasing the directivity."
Figure 2 shows the predicted polar patterns for the array with and withoutthe 6 dB increased drive level to the inner horn pair.
Leembruggen said, "Another way of showing the improvement in directivitybelow 600 Hz yielded by the array is to look at the off-axis frequencyresponses. Figure 3 shows the measured off-axis frequency response of asingle 4895 HLA low/mid element relative to the on-axis response. Normallythis plot is shown for horizontal angles, but with each element turned 90degree, it is now the vertical response. Figure 4 shows the predicted offaxis response for the HLA array with the 6 dB higher drive level to theinner horns. The increase in directivity up to 600 Hz over the single hornelement is quite evident with substantial directivity improvements below300 Hz".
Extremely limited space was available with all loudspeakers being housed incavities carved out of the walls. Jands undertook substantial mechanicaldesign to allow the loudspeakers to fit the cavities while providingfacilities to allow fine-tuning the loudspeakers' orientations. Key toJands' work were Peter Grisard, who handled the project management and themechanical design, and Peter Twartz and Kim Hasanic who oversaw thedifficult electrical and mechanical installation. From the Sydney OperaHouse side of things, the project was handled by Tony Lawrence. The wholesystem was commissioned by Leembruggen and Jands using an MLSSA analyzerand much listening.
Amplification for the main system is provided by a complement of AustralianMonitor AM1600, 1K2 and AM1200 amps as well as the onstage foldbackloudspeaker system consisting of JBL 4892s and ServoDrive Tech 7subwoofers. A 32infinity16infinity8 Amek/TAC SR9000 console is installed inthe audio control room. Recording of an event in the Opera Theatre can beaccomplished from the control room or the recording studio via a fullcomplement of tie lines and video links. A wide range of wireless, dynamicand condenser mics are available from Shure, B&K, Neumann, Sennheiser andAKG. Outboard processors and effects are also available with EQs from BSS,Klark-Teknik and Amek along with such effects generators as TC ElectronicsM5000, M3000 and M2000, the Lexicon PCM-80 and the Yamaha SPX990.
Leembruggen charactarized the sound as "really lovely."
Connor said, "Even at the very back of the theatre, the sound is extremelypresent and intimate and the tonal balance is excellent. The system is sowell balanced that during commissioning, we were able to hear changes tothe equalization of only 0.5 dB in most third-octave bands. Leembruggen andJands did a superior job in meeting what I had envisioned in thespecification. JBL has extremely good components, and the HLA system is avery responsive horn-loaded system. This became even better with theexpertise applied in the design and installation. I'm comfortable that thesystem is capable of meeting the needs of the room for the next 20 years ormore."