Floor Loudspeaker Reviews

All dipoles have traditionally shared many features, as well as most of their disadvantages. Dipoles typically consist of light sheetlike diaphragms driven more-or-less over their entire area, which gives them big advantages in transparency, "quickness," coherence, and lack of mechanical energy storage (compared to conventional cone drivers). This last advantage is a big one, which is what I think most people describe as "speed" in the bass. Big sheets also have a bunch of disadvantages, principally beaming and ringing: there simply is no free lunch—you can't have a big sheet in perfect control, one that starts and stops in perfect response to the input. None of these advantages or disadvantages is shared by the Mirage M-1, since it uses relatively conventional, if refined, drivers. (The "trilaminate" and "polyflex" mentioned in the specifications refer to what other companies call "doping.")

The dipole feature that is available to the Mirage, though, is the pleasing room effect you get when the loudspeaker radiates similar amounts of energy to front and rear. In any reasonably diffusive room (not to be confused with absorptive), the rear-firing energy returns to you by a fairly circuitous route (rear-side-floor-you, rear-ceiling-side-you, etc.). Since this is just what happens to you in a concert hall, though obviously with greater path length, the result is to turn your listening room into a mini-concert hall, adding ambience to dry or studio recordings.

This effect is almost always pleasant, but, like the adjustment of speaker position necessary with both dipoles and bipoles, must be played with for best results. First, the wall or walls from which the backwave reflects inevitably colors the reflected sound. A good example would be a hard plaster wall with no interruptions; almost inevitably, the recorded acoustic you hear will be colored to sound more like an empty church, and less like what it really was. Corners can be problems: I've heard effects where the corners act like resonating boxes, adding a honk to the sound. Additionally, you, like JGH, may rebel against the exaggeration of recorded acoustic. A finely tuned recording, in which hall ambience has been well balanced against direct sound, will sound pristinely like the original recording site when played back on a revealing high-end system using wide-dispersion, extended-frequency-response directional speakers; dipolars can expand on this effect, but it will frequently not be as correct at retrieving the original experience.

With dipoles, there is also a penalty. At low frequencies, the in-phase rear-firing wave comes back to meet the front wave 180° out of phase after reflection off the rear wall. This will result in a cancellation of the sound below a frequency dependent on the distance to the rear wall. Without a doubt, this makes all such speakers more room-sensitive. Just for starters, you must position the speaker so as to locate the back-to-front wave cancellation at the most favorable spot in the frequency range for that speaker, your room, the rest of your system, and your taste (footnote 1). Admittedly, this can yield stunning results, but the problems it causes in some rooms can drive you mad. (JA's and JGH's experiences with the Magnepan 2.5s—see Vol.11 No.6—are good examples in point.)

This is where the "bipole" M-1 has a significant advantage over dipoles. With the distance between the M-1's two woofers insignificant in relation to the wavelengths produced, they reinforce each other rather than cancel, acting in much the same way as a subwoofer with two drivers. (The top and bottom woofers of the IRS Beta, for instance, are much farther apart than the two in the Mirage.) Since, at low frequencies, normal-sized drivers act essentially as omnidirectional radiators, the rear-facing woofer behaves as if facing front.

In fact, significant design work has been done on the Mirage to smooth out the effect of the woofer reinforcement. First, the front and rear drivers are positioned differing distances from the floor to vary the point at which the driver-floor interaction (which happens with all speakers) happens. Second, to minimize the effect of a front-to-rear driver cancellation that occurs in the 200–300Hz range, the rear woofer is rolled off "early" at the top (100Hz). Because its output has diminished by 200–300Hz, the cancellation effect is lessened, and frequency response (according to Mirage) is smoother.

This is only one of the many design complexities incorporated into the M-1. Some of these simply come from resolving the problems of a bipolar design; others result from Mirage's commitment to flat off-axis frequency response, in line with the listening results from Canada's National Research Council. This subject has been thoroughly covered by both JGH in his review of the PSB speakers (Vol.11 No.5) and by JA in his review of the Image Concept 200 (Vol.11 No.8).

Briefly, Floyd Toole, chief researcher in this area for the NRC, has found in blind tests that one of the most significant indicators of good loudspeakers (as judged by their "ordinary person" listening panels) is extension and smoothness of off-axis frequency response. Nor are they the only ones to so believe; I brought up in my review of the Thiel speakers (Vol.12 No.1) Jim Thiel's belief in flat off-axis response, and I am sure there are others.

At that time, I stated that a gradual rolloff in HF off-axis might be better than flat, but in truth I haven't experimented with the two different schemes. I do know that flat off-axis response can get you in more trouble in an inherently bright room—but the ease with which the Mirages set up and played in my room would seem to indicate that my earlier comments about the Thiel may have been made in a state of some ignorance. (If you remember, the Thiels also set up quite easily in my room.)

Dipolar designs are inherently unsuited to flat off-axis response, except for products that function more or less as a line source, with a close-to-360° response pattern. In a typical dipolar radiator, the polar response will closely resemble a figure 8, with almost no response in the plane of the diaphragm, and dramatically reduced response at 60° off-axis. According to Ian Paisley of API, the Mirage yields an almost circular polar response pattern, with slight indentations at the sides. (I have not yet consulted JA's measurements to see if this is borne out.)

In order to yield these curves, Mirage uses as complex a crossover as I've heard of—essentially the opposite of what Jim Thiel does in his speakers. For the front-facing "speaker," the woofer crosses up to the midrange at 12dB/octave, and the midrange down to the woofer at 18dB/octave, both occurring at 300Hz. The mid crosses up to the tweeter at 18dB/octave electrically, but, counting the natural 6dB/octave rolloff already taking place in the midrange driver by 2.3kHz, 24dB/octave overall. The tweeter's low-frequency rolloff is similarly set at 24dB/octave, all electrical. The rear speaker is the same except that the rear-woofer crossover starts at 100Hz, and takes place at 6dB/octave. Relative to a positive-going waveform, both woofers move out from the speaker, while both midranges and both tweeters move in. I'll be most interested to see the M-1's squarewave and pulse response via JA's tests.

In addition, the crossover includes a "trap" filter to drop even further the 110Hz range of the midrange drivers. Why? The enclosure for the midrange drivers consists of a sealed, bifurcated tube running between them. At 110Hz the tube has a resonance, generated by the simultaneous pushing and pulling of the drivers on the inside of the tube, which would allow the drivers to "run away" a bit, so this additional circuit drops their impedance at that point.

At this point in writing a review—one night before absolute drop-dead deadline—I can't tell whether the detail provided above is interesting to you or not. I can tell you, though, that you won't find any of it in the M-1 instruction manual, which, though thoughtfully provided in both French and English, in neither language provides important information you don't already know beyond the address, phone number, and person to contact. Of those, the phone number listed has been changed and the contact doesn't work there anymore, though I'm sure those items will be corrected in future manuals.

For a $4000/pair speaker that contends for the best I've heard, this manual stinks. You learn about speaker phasing, speaker placement, selection of proper wire ("We recommend the use of heavy gauge (#10–#12) wire, for example, Monster Cable, or other specialty speaker wires..."), cabinet care, and a general section on advisable listening rooms and speaker placement. But if I'm a consumer who just spent 4000 bucks, and don't get to read about driver choice, crossover slopes and components, design desiderata, output limitations, frequency response, or even height, width, depth, or weight, I'm going to be unhappy. What is this, already?

Obviously, some of these things are most useful for reviewer types trying to get reviews out at the last minute, but my impression is that most Stereophile readers would also like to have at least some info of this kind. Perhaps API can spring loose some money to compose, in both languages, an adequate introduction to their flagship product.

---

Footnote 1: Those with dipole speakers and a willing friend can easily demonstrate this for themselves. Play pink noise or inter-station FM tuner noise through one of your speakers when it is placed as near to the rear wall as possible. The sound will be intensely thin, well-undernourished below the treble region. Then get your friend slowly to pull the speaker away from the wall (keeping it parallel). You will hear the cancellation region move down as the distance between the speaker and the wall increases—it sounds just like the familiar jet-plane type, comb-filter "phasing" sound featured on, for example, the Small Faces' "Itchycoo Park" track—until, assuming your room is large enough, you will hear the pink noise achieve a final homogeneous character as the cancellation frequency drops below the speaker's bass roll-off point.—John Atkinson