Loudspeaker designer John Dunlavy: By the Numbers... Page 3

Atkinson: I do wonder that with designers who haven't thought through the fundamentals of what they're trying to achieve in loudspeaker design, a lot of the tweaking or "voicing" that they do with their speakers is balancing problems in the on-axis response, say, with those in the power response, or vice versa. Let's imagine that someone is using an 8" midrange unit that beams quite narrowly at the top of its passband, crossing over to a 1" tweeter that has very wide dispersion at the bottom of its passband. That speaker will tend to sound bright. So what the designer then does is to pull down the on-axis response in the presence region—he ends up fudging both the on-axis response and the power response to get something that will sound flat in a typical room. But you don't have a loudspeaker that is in any way accurate. And certainly you don't have any accuracy in the time domain. The step response will be all over the place.

Dunlavy: That's a very good observation, and certainly very, very true. I think that, putting it into perspective, one must look at the range of the market and what portion of the market various designers are trying to target. Fortunately or unfortunately, the largest percentage of consumers are perhaps not as interested in objective accuracy as some of us would like for them to be. A lot of that comes from the fact that they simply don't have a great deal of experience to draw from, either in terms of listening to live music or knowing how to listen—knowing how to listen is an acquired talent. You're not born with it, I don't think. At least I wasn't. It took me a long time to figure out why speakers sounded the way they do...

Atkinson: What measurement tools do you use in your design work?

Dunlavy: We use a number of instruments. We don't trust any one in particular because different measurement systems permit you different latitudes and accuracies. We mostly use Doug Rife's [DRA Labs'] MLSSA system. The latest edition is just the most incredible measurement tool currently in the marketplace at any price. But we also use time-domain spectrometry because of the flexibility of what it permits us to do.

Atkinson: Doesn't your strict insistence on a design philosophy based on "accuracy" hurt you in the marketplace? For example, if you look at the SC-VI, this is a speaker that weighs over 500 lbs and stands almost 7' tall—yet when I listened to it, it didn't sound like a large speaker. It sounds large when the music is large, it sounds small when the music is small. But when people want a big speaker, surely they want it to sound big, to grumble away to itself in the bass?

Dunlavy: Oh sure. There will always be the boom-box advocates. And if it doesn't boom, they think there's something wrong with the bottom end of [the speaker's] response curve. But those who regularly attend concerts where there's no amplification of the musical instruments, well...the big bass drum is typically tuned to 30-35Hz. But it doesn't boom. It has a beautiful timpanic quality. It has a shimmering quality. It just kind of crawls across the stage and the audience to you. It doesn't just go boom boom boom boom boom. The drum diaphragm has a texture to it, an acoustical texture, that gets totally lost with a lot of systems and loudspeakers.

Atkinson: I've found that most big loudspeakers that I've measured for Stereophile do boom. I assume that trying to achieve well-damped low-frequency behavior is why you use sealed-box alignments for all your designs, without a port to be found.

Dunlavy: Pretty much so. As soon as you add a port, the Q goes sky-high. For a port to efficiently add to the direct sound being radiated by the driver, it had to delay the energy a full 180 degrees at that frequency where it's augmenting the frontal radiation. But that port can only radiate 50% of the energy per unit cycle. Then, the next half-cycle, it radiates 50% of the remaining energy, so you have this long decay time. That's why, when you look at the impulse or step responses of speakers designed according to this approach, they look terrible. In fact, would you buy a CD player, would you buy an amplifier, would you buy any other component in your system that exhibited that kind of response? Yet people do. And they love their systems!

I played string bass two to three nights a week for nearly 22 years prior to moving to Australia in 1981. And I can tell you, the biggest problem for a bass player playing along with even a four- or five-piece combo is playing the bass loud enough to where it stands out. It's just not a loud instrument. But [to accurately reproduce the sound of the string bass] you need a loudspeaker with a "solid" response, by which I mean time-coherent, very-low-Q bass with a minimum of boom. One has to avoid the use of ported enclosures. And as soon as you abandon ports, you must have damped cavities behind the woofer of sufficient size that the speaker can reproduce that low-end bass with a natural-sounding Q.

Atkinson: Returning to the topic of "accuracy," surely you can't have the same sort of target response for your small SC-I that you would have for the monster SC-VI? If you're going to discard two or three octaves of bass extension, doesn't that then mean you have to fiddle around with the treble response to better balance the lack of bass?