On May 18, 2012, Paul Barton, founder and chief product developer for PSB Speakers invited me and a handful of other A/V journalists to visit Ottawa, Ontario, Canada for a guided tour of the NRC (National Research Council) acoustics labs—the very lab facilities where many PSB product designs (and some NAD products) are evaluated and put through extensive double-blind listening tests. I had read about the NRC labs many times, but had never seen them first hand so that I jumped at the chance to see the labs and learn more about their day-to-day operations. Before I talk about my day at the lab, however, one critically important point of clarification is in order.
Clarification: As a matter of policy, the NRC does not invite visitors to its campus for any sort of PR-related activities. However, the NRC does allow some of the Canadian firms that use NRC facilities to bring guests on site (subject, of course, to relevant NRC security regulations). The only way that I, as a representative of Playback and Hi-Fi+, was able to visit the NRC facility, therefore, was as an invited guest of Paul Barton and of PSB Speakers.
About the NRC: Founded many years ago, Canada’s NRC is a brilliant example of government working in close collaboPSB-NRC Master ration with business to insure the long-term viability and growth of high-tech industries. The NRC was established to support advanced research spanning a wide range of technology-related disciplines, all with an eye toward making sure that Canadian firms will remain competitive, and ideally play leadership roles, within rapidly evolving technical markets. Moreover, the NRC also exists as a means of making sure that Canadian security and defense forces remain at the cutting edge of technical developments.
To these ends, the sprawling campus of the NRC comprises a total of 65 buildings, each dedicated to a specific area of research. On the drive in, for example, I saw a giant aerospace research center and an absolutely huge wind tunnel, which according to Paul Barton, has become a popular destination for certain NASCAR teams hoping to gain an edge through extensive aerodynamics testing. As I understand things, Canadian firms (including startups) can lease research time and technical assistance from full-time staff in the appropriate NRC facility at special, in-country rates. Firms from outside of Canada can also use the facilities, but pay much higher rates. In this way, NRC helps Canadian firms gain and maintain a technical edge in increasingly competitive worldwide markets.
Happily for audiophiles and music lovers, two of NRC’s buildings provide space for acoustics laboratories—labs that have been used successfully by PSB for R&D work on loudspeakers and, more recently, headphones. One lab building contains an anechoic (echo-free, sound absorbing) chamber, while the other provides an IEC standard listening room that is set up to allow extensive, comparative, double-blind loudspeaker listening tests.
Just to give us a sense for the sheer scope of research activities at the NRC, Paul Barton had made a special arrangement for journalists to see an ultra high-powered electron scanning microscope lab set up on a vibration-isolated platform in a research area located in the same building as the anechoic chamber. It was most impressive to see the giant microscope, to learn that in operation its test chamber is evacuated by a pump so powerful that the vacuum in the chamber contains fewer atoms that an equivalent volume in outer space would (!), and then to see resulting scan images out in the control room area. The researcher in charge showed how the scope could easily examine samples of materials at the molecular/atomic level. He pointed, for example, to a test screen that looked something like a close-up, black and white image of, say, a piece of tweed fabric. But in this case the “pattern in the fabric” in fact showed the alignment of individual atoms within a sample of a superconductor material under examination. Pretty heady stuff…
Inside the NRC Anechoic Chamber and Control Room: After our stopover to see the electron-scanning microscope, Paul Barton took the group into the anechoic chamber area—an area in which he has been doing research for many years. Indeed, one has almost the sense of the NRC lab being Barton’s “home away from home.”
There is a small control room area featuring the expected banks of signal generators, signal analyzers, audio amplifiers (oldie but goodie Canadian—of course—Bryston 4Bs, to be exact), a smattering of monitor speakers, various test instruments, an oscilloscope, a microphone control bay, and several computer workstations used to run tests, and then gather and analyze data. Off one side of the room, behind thick, double isolation doors, is the anechoic chamber itself, whose walls, floor, and ceiling are entirely covered with a cross-hatch of four-foot deep acoustic damping wedges (the wedges appear to be made of a golden yellow acoustic fiberglass-like material where the wedges are given shape by open weave metal mesh). The entry door to the chamber is itself covered with wedges whose shapes have been partially rounded off to accommodate the arc of the door when it swings open.
The interior of the chamber is a strange and wonderful place to behold. Extending from the entry door is a minimalist, open-weave metal “catwalk” that is suspended well above the floor of the chamber and at the end of which is found a motorized, remote-controlled loudspeaker test fixture. Visitors must be very careful when entering the chamber because, with the aim of minimizing acoustically reflective surfaces, the catwalk has been design with handrails at all (one wrong move and a guest could conceivably fall down upon the grid of acoustic wedges on the chamber floor below!). Since standing room on the catwalk was at a premium, Paul had to split us into two small groups in order to give each visitor some time in the chamber.
Inside the chamber, illumination is provided by a small set of work lights. Suspended from the chamber ceiling and positioned directly across the chamber from the loudspeaker test fixture is a geometrically precise array of costly, calibrated Brüel & Kjær microphones. The mics are arranged so that one mic is positioned directly on-axis with the loudspeaker under test, while two more mics take measurements from 15 degrees above and below the test speaker, with another pair of mics taking measurements from 15 degrees to the left and right of the test speaker.
Barton explained that this arrangement creates a measurement test “window” that will, when results from the five mics are averaged, tend to minimize if not eliminate any spurious diffraction artifacts that any one mic might happen to pick up. In addition to the mic array, the chamber provides a small closed circuit video camera (to enables testers to verify that loudspeakers are correctly positioned before tests begin), a small laser-type aiming device (which helps testers make sure the speaker is precisely aligned with the microphone array), and a strobe-type emergency warning (so that technicians working in the chamber with the door closed will have a visual means of being alerted if, say, a fire alarm goes off).
Paul invited us to close the chamber door and to observe how extraordinarily quiet the chamber really is. How quiet is it? It’s quiet enough that you might hear blood flowing through vessels in your ears and, if conditions are right, quiet enough that you might just be able make out the (acoustic) sound of your own heartbeat. Trust us on this one: Agent Maxwell Smart’s fictional “cone of silence” has nothing on this place. The chamber just plain swallows up echoes, reverberations, and sounds of all kinds. Paul pointed out that the chamber is certified to be completely anechoic from 20kHz on down to 80Hz, although it does reflect some bass energy below 80Hz. This low-frequency limit is largely a function of how big the chamber is, so that the only way to get a chamber that is anechoic all the way down to 20Hz is to build an exceedingly large (read “expensive”) one.
Happily, such a large chamber does exist, though it is apparently expensive to use and difficult to schedule for ongoing test work on audio projects. However, Paul took advantage of the larger chamber’s existence in order to create a low-frequency compensation curve for the smaller NRC chamber. His methodology was simple and elegant. First, he tested a speaker in the big chamber and then, on the same day and under identical atmospheric conditions, tested the same speaker again in the NRC chamber, and compared results. Predictably, results from the two chambers above 80Hz were essentially identical, but Paul was able to plot differences below 80Hz to create a compensation curve that could be applied for all future full-range (that is, 20 Hz – 20kHz) tests conducted in the NRC chamber. Clever, wouldn’t you say?
A Loudspeaker Test Session in the NRC Anechoic Chamber: Once Paul had given all of his guests a turn in the chamber, he offered to show us how an actual test session is conducted, using his own excellent and recently released PSB Imagine Mini small monitor speaker as the test subject. Part of the significance of this choice is that all of the journalists assembled in our small group had heard the Mini, while several (Playback included) had also reviewed the speaker. The test procedure went roughly as follows.
Step 1: Paul carefully placed the Imagine Mini on the test fixture in the chamber, connected speaker cables to the Mini, and then carefully shut both the inner and outer chamber doors.
Step 2: From within the control room, Paul turned on the laser-type aiming device in the chamber, and then, while watching the feed from the chamber’s video camera, carefully used remote controls to adjust the position of the test platform until the speaker was perfectly aligned with the microphone array.
Step 3: Paul set amplifier outputs to a level equivalent to the industry-standard of 2.83V at 8 Ohms, and then played pink noise through the speaker and conducted a check on each of the microphones in the five-mic array. During our session, this test turned up a faulty connection (thankfully on the control-room end of the signal chain), which Paul swiftly corrected.
Step 4: Paul fired up an oldie-but-goodie PC whose purpose is to control frequency response tests and to gather and analyze data. Making sure that amplifiers were connected to signal generators under control of the test PC, Paul initiated the test sequence. The process involved 20Hz – 20kHz test sweeps, where sweeps started at the upper frequency limit and worked downward to the lower frequency limit, with each sweep taking about 8 seconds to complete. The sweeps were, of course, repeated for each of the five measurement mics.
Step 5: After frequency response data has been captured, and taking care to make sure that the aforementioned low-frequency compensation curve is applied, Paul used the computer to analyze data and to generate a composite frequency response graph for the speaker. Paul demonstrated for us that it is entirely possible to view individual response curves from any of the test mics, or to overlay all five curves on top of one another, if desired. Any or all curves generated can be routed to an old-school X-Y axis plotter to produce hardcopy graphs for the designers’ logbooks.
And the Envelope, Please: The response curves for the Imagine Mini were impressively consistent from one mic position to the next, and they were stunningly flat (in essence, ± 1dB from a bit below 100 Hz all the way up to 20 kHz). It’s one thing to see results like these from monitor speakers costing thousands of dollars apiece, but it’s quite another to see them from small monitors costing around $760 per . PSB has long been known for building affordable loudspeakers that exhibit serious technical excellence, as these test results so eloquently demonstrate.
PSB’s “Special Sauce” Explained: In response to a question from a member of our group, Paul explained his thinking on what it takes to build loudspeakers that will work in the widest possible range of listening rooms.
In the real world, explained Barton, a major variable is the relative “liveness” or “deadness” of the speaker buyer’s listening room—something the speaker designer can’t possibly know in advance. According to Barton, almost any competently designed speaker that provided neutral or flat frequency response will work well in a dead (that is, very heavily damped) room, because the room will typically swallow up most of the speakers’ off-axis output no matter how good or how bad the off-axis response might be. But where the game becomes challenging is in trying to get speakers also to work well in relatively “live” or resonant rooms with lots of reflective surfaces. In those rooms, smooth off-axis response becomes critically important, because listeners will hear much of the speakers’ off-axis output reflected back toward their ears.
While noting that flat on-axis response is important, Barton emphasized that it is only a first step toward sonic excellence. The art of the design game, notes Barton, involves finding ways to make sure that off-axis response rolls off gradually and very evenly, so that there are no sharp off-axis response peaks or dips with which to contend (since such peaks or dips tend to stick out like the proverbial “sore thumb” in more lively rooms). To help ensure that PSB speakers work well both in “live” or “dead” rooms—or anything in between—Barton works very hard to optimize both the on and off-axis response of his speakers. And as you might have guessed by now, there are specific test regimens associated with this goal.
Although Paul did not have time to show us precisely how such tests are conducted, he explained that his standard practice is to measure all PSB speakers in 15-degree increments from the front side (from straight ahead, or 0 degrees, on around to 90 degrees off axis), both in horizontal and vertical axes. Then, he measures output from the rear of the speaker in 30-degree increments, again in horizontal and vertical axes. Barton keeps refining and adjusting his designs until he is satisfied that off-axis response is as smooth and evenly balanced as possible, yet without compromising smooth, flat on-axis response. This, says Barton, is the key to creating loudspeakers that will work well in most kinds of real-world rooms, a design practice we might describe as part of the “special sauce” behind the PSB brand.
In the second and final part of this blog, we’ll talk about the challenging and thoroughly eye-opening experience of participating in double blind listening tests in the NRC’s IEC-standard listening room. Until then, we wish you happy listening.