Our Anechoic Chamber: Grateful for a Dead Room
by John Meyer
In November of 1994, at Meyer Sound we embarked on the design and construction of our own anechoic chamber. After careful evaluation of the alternatives, we decided that such a room was an essential tool for designing new loudspeaker systems that would perform precisely as specified in critical sound reinforcement applications.
Most anechoic chambers of any meaningful size and quality are owned by universities, or by government agencies such as NASA and the military. I suspect most companies feel they can't justify the cost, especially these days when many have been led to believe the same results can be accomplished through computer modeling.
I strongly dissent from that view. There are simply too many variables in loudspeaker design (high Q horn design in particular) to allow highly accurate prediction through currently available computer modeling programs. Our experience has shown that some modeling predictions can be off by as much as fifteen dB, which is totally unacceptable. At Meyer Sound, we now consider deviations of even five degrees — accepted as standard tolerances by many in the industry — as being too wide for critical quality design work. For design of highly directional full-range systems, we now rely on exact measurement down to a single degree, preferably less. An anechoic chamber is still the only way to reliably gather this data.
The Meyer Sound anechoic chamber has been up and working for a number of years. As expected, this extraordinary testing tool enables us to design products with more consistent patterns, and it also accelerates our new product development cycle. A case in point is our CQ Series, which was nominated for the prestigious TEC award. Designing those systems would have been a nightmare without the chamber. More on that story later.
Anatomy of a Dead Room
The ideal anechoic chamber is a room totally free of acoustical reverberations. Any sound projected into the room, at any frequency, is fully absorbed.
Of course, no anechoic chamber is perfect. It helps to build the room as large as possible: the inverse square law dictates that sound energy will dissipate, so each square foot of internal surface will have less energy to absorb. Conversely, a smaller room will require more or better sound absorption material to have the same effect. This is particularly true of bass frequencies, and that's why smaller chambers only have effective absorption down to a certain limit-perhaps 100Hz or so.
The effectiveness of an anechoic chamber is measured in dB of rejection,which is the ratio of direct sound to reflected sound inside the room. The new Meyer Sound chamber provides better than 80dB rejection from 20kHz down to 80Hz, which is excellent performance for a medium-sized chamber. True, we still go outdoors for tests below 80Hz. But outdoor tests at higher frequencies are notoriously unreliable because of wind and temperature shifts. For the critical middle and upper octaves, an indoor chamber remains the only truly reliable tool for accurate measurement.
Built to Last
The external concrete structure of the Meyer Sound chamber was built-and built solidly-as a filling station garage back in the 1930's. About two-thirds of the self-standing structure now serves as the chamber itself, with the remaining portion housing the analysis lab.
We contacted Eckel Industries of Cambridge, Massachusetts, the world's leading supplier of materials for acoustic absorption, and asked them to design a chamber to meet our performance requirements. Next the internal walls were constructed, and a grid system was attached for the absorption wedges. Then the crew from Eckel rolled up two, 40-foot semi-trailers filled with their one meter, blunt-nosed, foam-tipped fiberglass wedges. They affixed the wedges for walls and ceiling, and attached all floor wedges vertically on roller boards to allow freedom of movement during set-up. During test sweeps, the entire floor is covered except for a small area occupied by the speaker positioner (which has absorptive treatment on all exposed surfaces). The internal working dimensions of the finished chamber are approximately six feet high, ten feet wide and twenty-seven feet deep.
Astronomical Precision. Literally.
I'm a great believer in numbers, and in the importance of precise measurement. I've found that in the field of high-Q speaker cabinet design, tolerances of ten degrees or even five degrees are of limited value. What we wanted was a system that could accurately measure response down to one degree, or less.
The traditional method of placing microphones in an arc around the speaker simply would not give us the precision we needed. Such a method is awkward to set up, it is prone to human error, and there's no way to fully automate tests for precise repeatability.
Therefore, we decided to use one fixed microphone and then find some way to rotate and tilt the speaker in tiny increments for fully automated test procedures. That approach would require a device able to move a large, heavy object with absolute precision.
Fortunately, the people who make mounts for observatory-size astronomical telescopes had the basic technology already in place. "Rotate to a tenth of a degree? No problem! Hold a speaker weighing up to 800 pounds with absolute stability? Easy!" Of course, large, computerized telescope mounts are not mass produced devices, and our custom positioner required extensive modification of previous standard designs. Consequently, it was far from inexpensive.
DFM Engineering of Longmont, Colorado did a superb job on our device, building the arm, the bearings and the servo mechanisms. They also provided a computer and software for automatic repositioning at any desired increments during test sweeps. Positioning commands are in hundredths of a degree with accuracy guaranteed down to a tenth of a degree.
To the best of our knowledge, this new positioner gives Meyer Sound the most precise tool for dispersion pattern testing anywhere in the professional loudspeaker industry.
Testing, Testing, 1, 2, 3...
In a typical test session, the speaker device is fed pink noise, and the positioner is programmed for a 360-degree sweep in one degree increments. A full sweep in one plane takes about 20 minutes. The acoustic signal is captured by a single Bruel and Kjaer microphone. We use either 1/4-inch or 1/2-inch capsules; each capsule has a built-in heater to eliminate any deviations due to moisture accumulation.
After boosting through the B&K preamp, acquired signal goes into a modified SIM® system which generates 20 averages on an 8000-point FFT analysis. Our custom software takes the raw data and converts it to 1/30 octave information using an algorithm developed in consultation with Dr. Henrik Staffeldt of the Technical University of Denmark. Data is formatted and stored on a Pentium computer running OS/2.
After running a horizontal sweep, we physically turn the speaker over on the positioner mount to do a vertical sweep, and then usually do a number of diagonals as well. Our goal is to capture a complete and absolutely accurate "fingerprint" of the device being tested.
Look at It This Way, or That Way
Our next challenge is finding a way to output all the data in a way that is meaningful. Currently we use several display programs using both Pentium and Power Mac platforms. Our printout options include polar patterns, waterfall plots, and pressure isobars ("weather maps"). The enormous amount of raw data at hand allows us to generate displays at very high resolution.
However, my feeling is that two-dimensional graphics are inadequate for representing what are inherently three-dimensional phenomena, so we are working on adapting three-dimensional computer graphic programs to accept our data.
In 1995, Meyer Sound was contracted to replace the sound reinforcement system in San Francisco's War Memorial Opera House. We had planned on using our MSL-4 systems, but it turns out that the MSL-4 cabinets were four inches too deep. They would not fit under the space allowed above the proscenium.
That meant we had to design a new, high-performance, high-Q cabinet that was slimmed down from MSL-4 dimensions. We used the chamber to precisely measure just about every type of horn ever made, pulling out designs going back to the 1930's. After a year of testing and fine-tuning various prototypes, we came up with the driver/horn/electronics package now in the CQ-1 and CQ-2. These boxes are 7.5 inches shallower than the MSL-4 and yet have only a slightly lower Q. Perhaps even more significant, the CQ Series horn lacks the typical horn signature; it has the open and natural sound more characteristic of a soft-dome tweeter.
We simply could not have developed the CQ Series in the same time frame without the chamber. In fact, I doubt we could have realized the same performance without using the chamber no matter how much time we took. It was an indispensable tool in bringing the CQ Series to the market, and I'm sure it will continue to play a central role in Meyer Sound's R&D program for years to come.
The Chamber's Brothers and Sisters
Bringing our dream of an anechoic chamber into reality was a group effort. Just about everybody in management and engineering at Meyer Sound was involved at some point, but I would like to give particular credit to Pete Soper and Fred Weed. Both played key roles in creating the chamber, and developing it into a smoothly functioning tool for critical loudspeaker testing procedures.