When we were asked to come up with an arrangement to replace the Meyer UPA's installed in the Opera House, the challenge in front of us was to achieve greater intelligibility with a minimal sound reinforcement system.
Our first step in that direction was to conduct a listening test in the Opera House using our MSL-4 product--a self-powered, arrayable, high-Q, long-throw reinforcement loudspeaker. We did our measurements on the MSL-4's to make sure that we were getting higher Q or directionality than with the UPA's--that is, more sound directed at the audience, and less rebounding off the walls. The listening tests in the Opera House proved to be successful and very beneficial. Basically, we thought that the issue was decided: we would engineer a new system based around the MSL-4's. This, however, was not the case.
The MSL-4's, as it turned out, were 4 inches too deep for mounting above the Opera House proscenium. Used as they were, the MSL-4's would partly block the view from the box seats on the either side of the stage. And, for various reasons, structural changes to the proscenium facing were out of the question.
We were now faced with redesigning the MSL-4, which we had just finished developing, in order to "fit," quite literally, this particular application. The prospect of redesigning the MSL-4 product simply to shave 4 inches off its depth, did not excite us.
Roger Gans, sound designer for the Opera, had essentially given us the go-ahead to redesign the MSL-4. But it was our thinking that, since the Opera Company was facing a major move that would take their '95 - '96 and '96 - '97 seasons over to the Civic Center, maybe they would give us the time and license to build a system the way we wanted. This is in fact what happened.
Toward More Sophisticated Test Criteria
Not long before the Opera House project began, I was invited to participate in a shootout involving one of our products at an architectural firm in Japan where they had a huge anechoic chamber. This experience convinced me that architectural acousticians operate on different level than speaker manufacturers. Architects and sound designers want to see a sound dispersion pattern they can draw in with a ruler and protractor: basically, a speaker that acts like a well-behaved floodlight. These people are using test criteria that are on another level of sophistication than those most speaker designers are accustomed to.
I am a great believer in numbers. My preference is to conduct tests in a completely controlled environment where you can gather data, and repeat tests endlessly for comparison. In an anechoic chamber, reverberance, generally destructive to intelligibility, can be totally eliminated and the characteristics of speaker can be studied minutely--before introducing it into an environment with walls. Further, in order to be intelligible, each note that a voice produces has to be above the noise level of the room. And, that can be measured too.
An excellent study exists--an articulation index, in fact, written in German--which lays out those numbers. (We've had to hire the assistance of a German-speaking professor, Dr. Henrik Staffeldt from the department of Applied Electronics at the Technical University of Denmark, in order to make use of this information.)
We Build an Anechoic Chamber
In order to participate at that level of testing and technology, I felt that it was necessary for Meyer to build its own anechoic test chamber. First we had to explain to our banker that this was a worthwhile investment--and not simply for our needs at the Opera House. Afterall, an anechoic chamber is a long-term investment. We knew it would take at least three years before the chamber led to the development of a new product and justified its cost. Once we overcame that financial hurdle, the way was clear for us to put together a very sophisticated test chamber.
The typical anechoic chamber is a very complex and cumbersome affair: a two-story tall room strung with a nylon or steel web for placing multiple microphones in order to measure sound from an essentially immovable source point. Measurements are conducted by setting up an array of microphones--which can be very expensive, and time-consuming as well. In addition, this sort of arrangement can not provide accuracy of more than 10 degrees -- and we were looking for much more precise measurements than that. In addition, when you start dealing with two speakers in an array, the interference patterns are much finer than that. The ripple phenomenon is everywhere, and well within the audible range.
We came at the problem from the other direction: we started thinking about building a chamber where we could use one, stationary microphone while moving the sound source very precisely.
I called around to various companies that built giant telescopes and mirrors to track astronomical objects. This led me to a company that built computer-controlled positioning mechanisms for the Jet Propulsion Laboratory. These mechanisms are capable of positioning very large, heavy objects, very accurately, in fact, to within a 100th of a degree. Building such a mechanism that guaranteed accuracy to within a 10th of degree--more accurate by a thousand-fold than measurements achievable in a typical anechoic chamber--was no problem at all for this company. Working with these people, we came up with a design for a speaker positioner, using their bearings and servo systems, that would hold a 900-pound object and rotate it in one degree increments, both vertically and horizontally.
Next we contacted a company to build a chamber around this positioning device that would provide very high ambient conditions, better than 85dB absorption--in other words, a room with virtually no reflections at all.
We now had the most precise anechoic chamber in the world, and we were ready to tackle the Opera House project.
Breakthrough Technology and the CQ Product
The Opera House project offered us an opportunity to develop an extremely smooth horn at a time when the industry had almost convinced itself that horns were always going to be ripply, and had really given up hope of developing a product that did not exhibit this characteristic.
What makes horns sound like horns is that they create their own reverberation. The object is to minimize this characteristic to the point where this reverberation or ripple effect becomes inaudible. This is an enormously complex problem, occurring in three-dimensions over a frequency range of eight octaves--and we're just now starting to bring massive amounts of parallel computing power to bear on the problem.
With this in mind, we sought to achieve a specific, technical goal pre-determined by the Opera House. The objective was to create a loudspeaker with a precise dispersion pattern of 50H x 40V.
After four months of testing every type of horn ever made--including designs going back to the 1930's--measuring them in 1/24ths of an octave, and a good deal of trial and error too, we were able to develop a highly directional horn, matched to a box and electronics, that became the CQ (Constant Q) reinforcement loudspeaker. It took another year to fine-tune the product.
Essentially, the technological breakthrough here was that we were able to create a horn where you can not hear the horn, one that sounds like a soft-dome tweeter, but actually out-performs a soft-dome design, which offers no directional control at all.
The CQ product is a complete system--a self-powered loudspeaker that's easy to install and use in any acoustic environment where extremely tight control and avoidance of spillover are critical. The CQ design really takes the guesswork out of speaker placement.
As for the size problem that was the original impetus for the CQ design: CQ-1s and CQ-2s are 21" wide x 30" tall x 22 1/2" deep--7 1/2 inches shallower than the MSL-4's. The CQ-1 has a dispersion pattern of 80H x 40V, and the CQ-2 a pattern of 50H x 40V. Both products have a lower Q than the more directional MSL-4.
Testing the CQ's
The CQ's were beta-tested in the acoustically problematic Bill Graham Civic Auditorium, temporary home of the San Francisco Opera Company while the War Memorial was undergoing rennovation. At the Civic Center, a flown ring consisting of one CQ-2 in the center surrounded by four CQ-1's, provided coverage for orchestra seating up through the auditorium's box seating. Then, for the longer reach up to the balcony level, we installed five UPL-1's at ground level on either side of the stage, and a cluster of MSL-4's. Designed for maximum throws of some 100 to 120 feet, the MSL-4's have considerably longer reach than the CQ's. We selected the MSL-4's because we knew they would perfectly complement the CQ's coverage and provide a well-balanced system for optimum sound. The room was further enhanced with six more delayed CQ-2's as a surround system. [For a complete account of the Civic Auditorium beta test of the CQ product, see the article, A Night at the Opera, "reprinted" here from Systems Contractor News.]
In the Opera House, six CQ's in all will be installed. The 80 degree CQ-1's will cover the downfill areas, spaced as wide apart as possible, but offering as little interference or ripple as possible. Both the 50 degree and the 80 degree product are designed to be used singly, but can be arrayed. If you find the right point, the CQ-2's in fact array very well.
Toward the Future
Currently, the CQ's are being used in a number of shows right now-- Smoky Joe's Cafe, The King & I, Disney's Lion King, and Beauty and the Beast
The CQ product opens up possibilities for other types of performances at the Opera House - ballet, for instance - where subtle sound reinforcement is desirable. In addition, the CQ product, and the research behind it, opens many possibilities for very transparent enhancement of classical music - applications where people might otherwise back away from the use a PA system. Any reinforcement application where sonic image is a concern, such as mixing and playback of 5-channel surround sound for film, would benefit greatly from a speaker with a uniform frequency response of 40 Hz to 18 kHz over the entire coverage area in both the horizontal and vertical axes, and no side-lobe spillover.