Loudspeaker designer John Dunlavy: By the Numbers...

One of the characteristic traits, I have found, that defines the loudspeaker designer is that they are loners—they seem to avoid one another's company as if on purpose. But if ever you sit down with a designer, all you need to do to open him up is to ask him what he feels to be important in loudspeaker performance.

Such was the case with John Dunlavy of Dunlavy Audio Labs. I was driving up to Colorado Springs to measure his 530-lb Signature SC-VI loudspeaker—reviewed in the August 1996 by Steven Stone—so it seemed an ideal opportunity to sit down with John and ask him what he felt to be fundamental in getting the best from a loudspeaker. The floodgates opened. One of the things I found surprising was to learn that Kansas-born John does a good deal of the basic design work for a loudspeaker without ever listening to music on it. This would seem to be high-end heresy!

John Dunlavy: Oh, no. Listening comes later. Because if you stop to think about it, no loudspeaker can sound more accurate than it measures. It may sound worse, or it may sound sweeter, prettier, but if we're talking about absolute accuracy—the ability of the speaker to reproduce as perfectly as possible whatever's fed to it—such a system can never sound more accurate than it first measures. So we try to get the greatest accuracy we can achieve from measurements. Then we begin doing what some people might call "voicing," because the best set of measurements are still open to interpretation.

You could have a speaker that makes ±1dB, for example. And you say, "Gosh, how can you better that?" But what if it's +1dB over an octave and a half, say from 1-2.5kHz? And then it suddenly jumps down to -1dB, for a total change of 2dB, for the next octave? You're going to hear a spectral imbalance. The specs might look great—golly, here's a speaker that measures ±1dB—but it's not going to sound nearly as good as a speaker that is up and down 1dB every third of an octave. In the real world we're used to hearing that—reflections from the walls of the room cause similar variations—and we tune that out.

It may come as a surprise—this is giving away a trade secret—but when I design a loudspeaker, I first design it by looking at the step response. I find that by playing around with the crossover network while observing the step response in real time, any change I make is immediately available. When I get the step response right, everything else goes along. It's implicit. It goes along for the ride.

So, to those who like to call us "technocrats" or whatever, we would suggest that those who design by "voicing" loudspeakers are working with an enormous number of perturbations. Did they have their ears examined the morning they designed the speaker? What kind of a room did they voice it in? What program material did they listen to to voice it? You go on and on and on. The permutations and combinations are mind-boggling, to say the least.

When I was a little brat and color photography was just coming in—this was back in the late '30s—grass tended to look greener than green and all reds were fire-engine red, and the sky was super-blue. And everyone was saying, "Oh, aren't these photos wonderful!" It was almost better than real! It was surrealistic instead of realistic. And people initially liked it.

But what we're finding more and more is that you get tired of that. Nature is pretty clever; we've evolved naturally under the influence of our environment, and nature has tended to provide us with a color spectrum that, over a long period, is more satisfying than surrealistic colors. The same is true with sound. We find that a good percentage—not everyone, certainly—of young people today like "boom-boom boxes" on the bottom end, they like maybe a high end that rises a bit, and that's fine. That's like the surrealistic color on old postcards. But in the long run, you grow tired of it.

Recently I read in one of the hi-fi magazines that accuracy doesn't mean anything: "As all violins sound different, there cannot be a concept of what an accurate violin actually sounds like." To me that's illogical. If I buy the latest violin recording made with a very rare 17th-century violin or whatever, I want to hear what that particular violin sounds like. I don't want to reproduce it so it sounds like a generic violin.

John Atkinson: But that's what a bad loudspeaker does. It imposes its own character on everything.

Dunlavy: Exactly.

Atkinson: To judge from the SC-VI, it seems that you regard getting the outputs of the drive-units in your speakers to arrive at the listener's ears at exactly the same time as being very important. Other than the VI's subwoofers, all its drive-units are recessed into the front baffle. How important is that to the loudspeaker's sonic integrity?

Dunlavy: We participate regularly in the recording of our symphony orchestra here in Colorado Springs. We've also recorded instruments like violins and cellos and timpani in our big anechoic chamber using instrumentation-quality microphones and equipment. And we find that in order to reproduce those sounds with a level of accuracy such that you can not literally hear any difference between the live and the recorded sound, you have to have a speaker that exhibits almost perfect impulse and step responses. The only way to do that is to time-align the drivers very, very accurately, usually within a matter of a few microseconds, then use a minimum-phase, first-order crossover network and get everything right. And you have to have an on-axis response of better—well better—than ±2dB.