Magnat, a prominent West German loudspeaker manufacturer, is introducing a novel four-way, three-piece speaker system that is claimed to be the first speaker to achieve true omnidirectional sound dispersion. The system supplied to us for test was a preproduction model, as the speaker is not scheduled for release until later this year.
The Magnasphere Nova consists of two satellite speakers and a powered subwoofer module. The subwoofer, like similar products in its external appearance, is approximately cubical and has a glossy black piano finish. Its output emerges from slots about 2 inches high and 20 inches long near floor level on the front and rear of the enclosure.
The satellites, however, are unique in appearance and design. Each consists of three black, perfo-rated-metal spheres stacked vertically on a slender black-steel column. Each sphere contains a pair of magnetic drivers mounted back-to-back and radiating in opposite directions. Since each driver pair operates in phase, the radiation pattern is approximately the same in all directions except in the plane of the sphere’s vertical “equator,” where the drivers are joined.
The bottom sphere of each satellite, the “woofer,” contains two 4-inch polypropylene-dome radiators that operate between 100 and 1,400 Hz. Each dome, constructed of two different thicknesses of polypropylene and mounted in a cutout in a plastic sphere slightly less than 8 inches in diameter, is driven at its 2-1/4-inch diameter by an inner “cone” that has a smaller voice coil at its apex.
The middle sphere, which radiates from 1,400 to 3,400 Hz, is essentially a smaller version of the woofer sphere. It contains two 2-inch soft-metal-dome radiators mounted back-to-back within a 4-inch spherical case, which in turn is surrounded by the perforated-metal cage. Immediately above it is the tweeter, with a pair of 1-inch drivers in a 2-1/2-inch cage (all these dimensions are approximate).
Passive crossover networks for the satellite drivers are located within their spherical enclosures, which are joined by eight-pin DIN connectors. A separate control box contains an active electronic crossover circuit, with a slope of 24 dB per octave, that excludes the low-bass frequencies from the satellites, supplying them to the subwoofer module through a cable and DIN plugs. The controller has front-panel adjustments for setting the subwoofer’s output level and the low-bass crossover frequency (either 80, 100, or 120 Hz).
An unusual feature of the Magnat system is its time-delay correction. Any three-piece system such as the Magnasphere Nova is very likely to have large time differences between the outputs of its subwoofer and satellites. In part, this results from the placement of these components in the listening room, where they are often many feet apart, and in part from the inherently greater delay in the output of a subwoofer compared with that of a midrange or high-frequency driver. The audible significance, if any, of these time differences has not been established, but the Magnasphere Nova provides a way for the user to compensate for them, if only approximately.
The signals supplied to the satellites are routed through a high-pass Bessel filter whose delay can be adjusted for 0, 32, or 64 milliseconds (ms) by a switch on the control unit. Since typical delay differences in a three-piece system like the Nova will fall in that range, presumably a user could reduce the differences substantially by delaying the satellite speakers’ output by 32 or 64 ms.
The Magnasphere Nova subwoofer contains four long-throw drivers, each with a cone diameter of 7-2/5 inches. They are mounted in opposing pairs on a board that divides the inside of the cabinet horizontally. Each pair is driven by the signal from one stereo channel, although the acoustic outputs of all four drivers are effectively combined at the cabinet’s ports. Magnat describes the subwoofer as having a “compound principle” of operation. Apparently each opposing pair of drivers operates in “push-pull” mode, with their cones moving in the same direction, instead of in opposition as in the satellites. The subwoofer’s amplifier, built into the enclosure, is rated to deliver 180 watts per channel continuously, or 500 watts peak output, with less than 0.05 percent distortion.
The top of the tweeter sphere is about 47 inches from the floor, the bottom of the woofer sphere about 30 inches. The stands, about 1-5/8 inches in diameter, have black cast-iron bases. Each complete satellite weighs about 28 pounds. The subwoofer module is about 21 inches wide and high, and 18 inches deep. Although its weight was not specified, we can testify that it is very heavy!
The control unit is a black-finished metal cabinet that measures 16-3/4 inches wide, 11 inches deep, and 3-3/8 inches high. Its front panel contains a power switch and a small button that causes a motor to open and close the lower half of the panel, which covers the other controls. The panel also has a pair of overload lights to warn of excessive signal levels. Price: $5,500 (estimated) for the system.
Some details of the Magnasphere Nova system, especially the control unit and its specific adjustment ranges, might be changed in the final production models. Since the radiators used in the satellites have been in the Magnat product line for some time, however, they can be expected to remain as described.
We placed the satellites about 8 feet apart and 4 feet from the wall behind them. The subwoofer was located approximately on the same line as the satellites and midway between them. We set the subwoofer output level by ear for the best possible bass response without excessive boominess on vocal sounds, and the 80-hz crossover setting was used for most listening and measurements. We tried the other settings as well, but the lowest of the three frequencies generally produced the best results.
Our frequency-response measurements were made in the usual way, plotting room-response curves for both satellites on the same chart and averaging them to smooth out room standing-wave effects. We measured the subwoofer’s response with each of the crossover settings, placing the microphone at one of the output slots. Bass distortion was measured at the same point, using the 80-Hz crossover setting.
We measured the impedance of a satellite unit, which is the only part of the Magnasphere Nova that is actually driven by the user’s regular system power amplifier, over the full audio frequency range. The system’s sensitivity was measured at a 1-meter distance, with only one satellite operating, using 2.83 volts of pink-noise input. Since there is no fixed relationship between the signal input level and the subwoofer’s acoustic output (their relative levels are set by ear), we measured the drive signal needed to produce a 90-dB sound-pressure level (SPL) at a 1-meter distance from the subwoofer slot at 50 Hz (on the flat portion of the subwoofer’s frequency response) and then maintained that same input level from 100 Hz downward when measuring bass distortion.
Our quasi-anechoic FFT response measurements with the IQS signal-analysis system were made 1 meter from a satellite. With the microphone close to each of the spheres, we also measured their individual frequency response on the forward axis and at 45 and 90 degrees off-axis (the latter angle is close to the solid band that unites each pair of hemispheres). Plotting the axial and off-axis response curves on the same printout enabled us to assess the actual directivity of each section of the satellite speaker. In addition to our usual group-delay measurements, applied to a complete satellite array, we made a group-delay measurement on the output of the subwoofer alone.
The satellites’ combined room response was very uniform, ±3 dB from 200 to 20,000 Hz, indicating a relatively flat acoustic power-output characteristic. Below 200 or 300 Hz the response became irregular because of room reflections. The close-miked subwoofer response was flat within ± 2 dB from 25 to 80 Hz using the 80-Hz crossover setting; the upper limit extended to 110 and 140 Hz with the 100- and 120-Hz settings, respectively. From 30 to 72 Hz, the subwoofer’s output varied only ± 1 dB.
Splicing the Nova system’s bass and room-response curves was not as easy as it is for most conventional single-cabinet speaker systems. The reference level of the bass curve could be placed almost anywhere relative to the room-response level, and the room’s standing-wave effects eliminated the usual clues to the relationship between the two. We finally joined the curves with the flat region of the subwoofer response at the same level as the smoothed satellite room response, which left the region from 100 to 200 Hz undefined. Since this frequency range is strongly affected by speaker placement and room dimensions, our choice was not unreasonable. The other parts of the curve give a fair indication of what the Magnasphere Nova system itself is capable of.
The composite response, which had only a few minor irregularities, varied ± 3 dB from 23 to 20,000 Hz. The bass distortion was between 2 and 3 percent from 25 to 100 Hz, climbing to 8.5 percent at 20 Hz. The distortion consisted entirely of second and third harmonics.
The impedance of a satellite, using the 100-Hz crossover, varied between 3.7 and 7 ohms over the range from 200 to 20,000 Hz (the nominal system rating is 4 ohms). The impedance rose smoothly below 200 Hz, to about 60 ohms at 20 Hz, reflecting the isolation provided by the system’s crossover network.
The system’s sensitivity appeared to be rather low, with each satellite producing only 80 dB spl at 1 meter with a drive signal of 2.83 volts of pink noise. This figure does not include, however, the considerable contribution of the subwoofer at frequencies below 100 Hz. In actual use, the system did not require an unusual amount of drive power even at very high levels.
The fft measurements were consistent with our other results. Since they extended beyond our usual 20,000-hz measurement limit, they confirmed that the satellites’ high-frequency response was virtually flat up to about 27,000 Hz. One of the more interesting fft measurements was the group-delay response to the subwoofer, which averaged about 30 milliseconds from 35 to 95 Hz and increased to 70 ms at 20 Hz. This suggests that the optional delay compensation in the Nova’s system controller could, in theory, correct for much of the difference between the subwoofer’s delay and the typical delay of less than 0.5 ms across the rest of the audio range that we measured from the satellite output.
The fft measurements also proved that the spherical radiators are not really omnidirectional, although their effective dispersion is wide enough to give them many of the subjective qualities of an omnidirectional radiator. A close-miked measurement of the tweeter sphere at angles of 0 and 45 degrees off its forward axis showed that the two response curves differed by 3 to 6 dB over most of the tweeter’s operating range. Although this is good dispersion, the response at 90 degrees off the forward axis was down more than 20 dB over most of the range. As would be expected, the midrange radiator was better, with less than 3 dB separating the axial and 45-degree curves over its full operating range. Still, the midrange output at 90 degrees off-axis was typically 18 to 20 dB below the axial level. The woofer sphere showed virtually no response change over the first 45 degrees, but at 90 degrees it was a uniform 12 dB down over most of that driver’s operating range.
The Magnasphere Nova system was a very “listenable” speaker. The sound from the satellites seemed to float in the air, with no clues to its origin as long as we were a few feet away from the speakers. It not only filled the space between them but extended behind and above them as well.
We were unable to hear any difference in the sound as a result of switching the delay compensation between 0 and 64 milliseconds, but there is little evidence to show that bass delays in that range are audible. While this feature looks promising in principle, it may be of little practical value.
The subwoofer, of course, cannot be located by ear, although it is certainly obvious to the eye. Too much subwoofer output can make some programs bottom-heavy, and we chose what seemed to be a reasonable level for our listening tests, setting the bass control at less than half of its maximum rotation. Even this middling level produced a solid, almost tactile deep-bass output with suitable program material.
We soon found that listening to the Magnasphere Nova tends to distract one from trying to dissect and analyze its sound output. That is a characteristic of most very good speakers, and the Nova is unequivocally one of the better-sounding ones we have heard.
Our only clear disagreement with the claims made for the Magnasphere Nova concerns its supposed “omnidirectionality.” The satellites are really sets of three dipole radiators, with virtually no radiation to the sides and equal outputs front and rear. They have the audible qualities of a good dipole and are much more compact than most. Whether you find them attractive as home furnishings is a matter of personal taste. And they are certainly expensive. But in the final analysis, it is the sound of a speaker that justifies its cost, and Magnat’s Magnasphere Nova system passes that test beautifully.