To Doug Schneider,
I noticed you will soon be publishing the Magico S5 review. I will be interested to see how you feel they compare with Vivid Audio and Rockport Technologies models. I do, however, have some questions pertaining to measurements your mag makes at the National Research Council. Do you have any articles explaining how to interpret the graphs you publish? Also, I have been rereading and doing some research on time- and phase-accurate speakers, and there seem to be some very valid reasons for building speakers that way, although more complicated and expensive for the manufacturer. What is your take on the topic, and are your top picks all phase- and time-accurate?
Our review of the Magico S5 will be published on this site on December 1, with measurements. We used to have a page explaining a bit about the measurements, but it wasn’t detailed enough, so we don’t link to it anymore. I might look into producing another, but only if I feel I can do a good enough job of it -- there’s a lot to explaining and understanding them. That said, there are a number of loudspeaker-design books and websites that have information explaining similar graphs, so that might be a better place to learn, at least for now. A good place to start is the "Loudspeaker Measurement" page on Wikipedia.org, which at least explains some terms and typical graphs.
Your question about phase and time accuracy is a good one that comes up quite often. I assume that you mean a passive loudspeaker with a first-order crossover (which provides minimal phase shift) and proper alignment of the drivers (usually a stepped or sloped front baffle). If so, it might surprise you to learn that most speakers today are not time- and phase-correct, including all of the ones I think sound the best, and that the reasons for not doing this has nothing to do with difficulty or cost.
On the surface, creating a speaker that’s accurate in terms of time and phase seems like the sensible thing to do. In fact, I’ve learned that many designers begin that way, but quickly realize that it’s not necessarily the best approach, and often end up scrapping such designs and going for a more balanced design approach, which doesn't optimize those parameters, but does optimize others that have shown to be more important when it comes to producing a better-sounding loudspeaker -- flat frequency response, wide and even dispersion, low distortion, etc.
This brings me to why it’s not necessarily a good idea to design strictly for time- and phase-accurate performance, at least with passive loudspeakers (with active digital designs, which we’ll see more of in the future, the rules will change because phase shifts can be more easily eliminated). Even if you include all the ingredients necessary for this kind of design, a speaker with a separate tweeter, midrange, and woofer, produces time- and phase-correct behavior at just one very small point in space, usually right at the listening position. That might seem sufficient, but it’s really not at all adequate when you consider that the loudspeaker interacts with the room. Basically, everything outside that very small listening window is anything but correct insofar as time and phase go; in fact, it’s often worse than with speakers that have not been designed for on-axis time and phase accuracy. This can be somewhat ameliorated by using coincident drivers, in which the tweeter is placed inside a midrange cone, but such drivers usually present problems of their own.
Another problem is that a true first-order crossover (electrical and acoustical properties combined) results in plenty of driver overlap. This means that the drivers are usually operating way beyond their optimal bandwidths, which often results in high distortion -- obviously, not a good thing. That’s why there’s no such thing as a time- and phase-correct two-way speaker: its tweeter would still have significant output below 1kHz, and its midrange-woofer up to 10kHz. Truly time- and phase-correct speakers are almost always three- or four-way designs, in which the operating bandwidth of each driver is considerably narrower. Time- and phase-accurate speakers can certainly have flat on-axis frequency response, but they usually have poor off-axis response, usually because of the driver overlap, which again is not a good thing. There are other problems as well, but this should be enough to tell you that serious problems and compromises arise when you try to create a speaker that’s time- and phase-correct, most of which can’t be solved no matter how much money you throw at them.
As for the companies you mention -- Magico, Vivid Audio, and Rockport Technologies -- none makes speakers that are time- and phase-accurate, even if the sloped front baffles of the Vivid and Rockport models look as if they might be. In all cases, they use higher-order crossovers to greatly reduce driver overlap, which have more phase shift but result in reduced distortion and improved off-axis performance, which, as I mentioned, are important. It’s not that these companies’ designers lack the skills to design for time- and phase-accuracy, or that it’s more complicated or expensive for them to do so, given that each already charges boatloads of money for its speakers. It’s just that they know that the tradeoffs in doing so aren’t worth it; they favor a more balanced approach that, in my experience, results in better sound. . . . Doug Schneider