Loudspeaker designer John Dunlavy: By the Numbers... Page 2
Dunlavy: That's right. You can go one way but not the other. Flat on-axis response provides very poor correlation, in general, with what you hear. On the other hand, if you have near-perfect impulse and step responses, it follows that you must also then have from that a near-perfect frequency response on-axis. And using a first-order crossover network is the only way you can achieve accurate impulse and step responses. As soon as you go to a second-order crossover, the impulse response is hideous.
Atkinson: I can count the manufacturers of loudspeakers that produce good, time-coherent impulse responses on the fingers of one hand—Dunlavy, Spica, Thiel, Vandersteen, Quad. So why don't the designers of 99.9% of loudspeakers out there care about the impulse response?
Dunlavy: I think a lot of people, especially those who don't listen to live music a great deal of the time, are not really concerned with whether the music they're reproducing matches the live performance. They're after more of an effect. And so the accuracy of their systems doesn't need to meet the same criteria that would have to be met if one wanted to make a loudspeaker where you couldn't hear the difference between it and the original performance. And that's what we're into. Certainly there's plenty of room in the marketplace for what might be called "good-sounding" speakers, "sweet-sounding" speakers, "nice-sounding" speakers, "pretty-sounding" speakers—everybody hears differently. On the other hand, there are a lot of people who regularly attend live concerts who want accurate reproduction.
Designing with higher-order crossover networks greatly simplifies the blending task between drivers. But what most people don't realize is that one of the really great difficulties in designing with higher-order crossovers is that they store energy. That's very visible when you look at the impulse response of a speaker that has a second-, third-, or fourth-order network. And the step response also looks terrible.
Of all of the measurements that we take that come more close to predicting, or most close to predicting how a speaker is going to emulate a properly recorded live performance, it's step response. Everything is implicit if you know how to interpret a step response...if my life depended upon my describing what I thought a speaker was going to sound like, all other factors being equal, I would choose step response. And feel very confident that I would be spot-on.
Atkinson: You talk about accuracy, but so far you've just defined that in terms of the impulse or step response on the speaker's intended listening axis. What about the speaker's power output into the room? That surely has as much effect on the perceived balance as the on-axis performance?
Dunlavy: That's certainly true. We pay an awful lot of attention to the power response of the speaker into the room. Because that's one of the things that permits us to determine whether we're listening to a live instrument, let's say with our eyes closed, in a typical room. We hear two things. We hear the direct sound of the instrument, but we also hear all of the reflected sound, the reflections off of all of the boundaries of the room. And the ratio between that direct sound as a function of frequency and the reflected sound determines to our ears whether we perceive it as being realistic or not.
We spent a lot of time and money, over 20 years ago, doing measurements in an anechoic chamber of the three-dimensional response patterns of 17 different musical instruments, including drums, string bass, cello—we measured a bassoon, a clarinet, a violin. If a loudspeaker's directivity pattern is incapable of emulating the aggregate, the average of the patterns of all of these musical instruments, it will never sound "accurate."
Most musical instruments are almost omnidirectional at low frequencies, as are most loudspeakers, so it doesn't pose a problem. But as you go higher in frequency, to between 100Hz and 300Hz, if you don't get the beam-width of the speaker correct in this range—and by "correct" I mean that it simulates most live instruments—it will add warmth, unnatural warmth, to the sound of voices and musical instruments. It'll make the average male voice sound too chesty, very unnatural. As you go up higher in frequency, if you have a tweeter that radiates too broad a pattern...it's going to produce shrieky sounds, it's going to sound too zippy. I think everyone's experienced that, especially from inexpensive speakers that have a rising high end.
So a good designer certainly knows that he has to pay a lot of attention to the polar response of a loudspeaker.