Floor Loudspeaker Reviews

The panels are hinged, allowing them to be tilted forward or back to accommodate any likely listening height, as well as unlikely ones like LA standing up. (Yogis please note: the forward tilt range is not great enough to allow listening from the lotus position, unless you are able to levitate.) Two screws on each base can then be tightened to lock the panel in position. This tiltability is a dandy feature, and one I wish had been available on other large electrostatics I have tested in recent years. It must also be said that it is almost a liability on the Spectras, because they have such broad vertical dispersion that the tilt feature will rarely be needed, and the adjustability makes it possible to end up with different tilt angles. Both panels must be at exactly the same angle; if they aren't, the panel top/bottom timing differences between the two speakers will cause phase interference, resulting in an apparent channel imbalance (footnote 2).

Acoustat recommends that the speakers be toed in, with the axes of their innermost panel strip intersecting at or just in front of the center listening seat. The distance from the rear wall will depend on whether or not the woofer is used. If it is, the speakers can be anywhere beyond two feet from the rear wall; if not, they should be at least four feet from the wall.

Listening
Initial listening was done with the woofers off and the panels running full-range. Equipment used was the Ortofon MC-2000 cartridge with its own step-up transformer, SOTA Star turntable, Sony '650/'703 CD combo, with Audio Research SP-11 preamp and Threshold SA-1 power amps.

Four things struck me immediately about the Spectra 3s' sound. First, their extreme high end is exquisite: velvety smooth, open, and detailed, without a trace of sizzle or tizziness. Second, their low end is fuller and more extended than I have ever heard from any full-range electrostatic. Third, they have superb lower-midrange, which gives cellos and large brasses marvelous weight and authority. And fourth, their sound is a little brash, almost but not quite to the point of being sizzly. Massed violins were not exactly steely, but they were imbued with a slight edginess. I suspected a mild peak in the 6–9kHz region, but could not measure any such. In fact, the measured high-end response was extraordinarily smooth, except for a mild discontinuity around 5kHz, where a 2dB dip was followed by a 2dB hump. I doubt if that would account for the hardness, but it could.

Imaging from the center listening seat was excellent—stable and specific. And it stayed centered from any listening seat along the entire length of my seven-foot sofa! But while off-center listening preserved the stereo stage in good balance, it shot down the imaging specificity, producing broad, vague center images several times wider than they should have been. So, while we now have, for the first time, an electrostatic speaker which allows off-center listeners to hear a balanced stereo stage, we still haven't eliminated the center "sweet spot." But it's a step forward, nonetheless.

Soundstaging was good, but not very. The Spectras had a tendency to compress width, putting extreme-left and -right images at the inside edges of the panels rather than behind the middle of the panels. With those rare recordings which (because of phasing anomalies) sometimes image beyond the speaker positions, those images never went beyond the outer edges of the Spectras. The obvious solution here—moving the speakers farther apart—didn't work, because the center imaging from the center seat fell apart, becoming vague and unstable. Clearly, what is happening here is that the ear is latching on to the full-range panel as the primary source of directional information from each speaker, but why the problem is not soluble by wider spacing is something I cannot explain. The Ohm 5s, which also crossfire their middle and upper ranges, produced definite, stable images from way out beyond their physical locations with those same recordings.

Bass output from the Spectra 3 was way out of balance with the rest of the spectrum, and no amount of messing with room placements helped. (The output at 40Hz measured 6dB above that in the midrange! Output at 32Hz was only 4dB below that at 40.) The low-end rise made the sound tremendously impressive but, probably for the same reason, LF definition was rather mediocre. Pitch delineation was only fair, there was no count-the-cycles throb from bowed basses, and bass drum was soggy.

On many recordings—some Telarcs in particular—at average listening levels of only around 85dB, sudden LF onslaughts (as from bass drum) caused the entire sound to choke down for a moment as though the step-up transformer was saturating. At one point during the Sheffield Shostakovitch 1, three rapid bass-drum strokes took the output level down by increments of about 2dB each, reaching about –6dB before the drum let up and the signal output could recover.

Acoustat's chief engineer, Jim Strickland, told me this had nothing to do with the step-up transformer, but was due to momentary loss of the high-voltage charge when the diaphragm comes very close to the stator wires. He also pointed out that that is the main reason for incorporating the dynamic woofer.

Worse, though, was the fact that at and above moderately high (90dB) midrange volumes, the diaphragms frequently bottomed out on strong bass signals, causing irksome clicking noises. Now, 90dB is probably loud enough for many listeners, but it is not going to satisfy audiophiles who like to shake the floor during fortissimos. Neither does it approach the live peak spls reached by instruments like the bass drum. (English audiophiles recognize a phenomenon called scale distortion, which results when a recording is played at a higher volume than that at which it would be heard live from its apparent distance. With most minimally miked symphonic recordings, proper scale is obtained at average levels around 90dB. With most small-group pop and jazz recordings, levels around 100dB or more are called for. The Spectra 3s cannot handle the latter, full-range.)

Next, I tried the Spectras with their woofers and internal crossovers switched in. This (of course) had no effect on their upper ranges, but it did ameliorate the LF problems immensely. I was now able to run the system at midrange peak levels of around 100dB before encountering any problems with LF stress. Unfortunately, there was also some loss of output below 50Hz which, since the 50Hz region was less affected, tended to remove some of the underpinnings of the sound. LF quality, however, improved with the cone woofers working—something that rarely happens with electrostatics. Usually, cone woofers tend to impair bass detail, but in this particular case the bass from the full-range panels was so ill-defined that the cones could only help matters.

I am aware—perhaps more so than many reviewers—of how much a loudspeaker's sound is influenced by the power amplifier used to drive it. How, then, can I can so confidently attribute the problems I heard from the Spectra to the speaker itself? Because I have found the Threshold SA-1s to be almost right down the statistical middle when it comes to bass balance and control and to neutrality of midrange. Just a shade less forward through the midrange than the least-forward tube amps, they have the midbass quality of the tightest-sounding tube amps, and a deep-bass performance completely in keeping with their midbass quality—much stronger and better controlled than any tube amp. I will not claim for a moment that they are perfect, any more than I would do so for any other audio product, but their sound is representative of all the other power amps that have garnered enthusiastic reviews in these pages.

A Change of Amplifier
At the time of testing the Spectra 3s, I also happened to have in-house a model 1.1S power amplifier from a new company called Mirror Image. As I had been impressed with what I heard from that amp during a brief pre-listen—review next month—I tried it with the Spectras.

Reverting to full-range operation of the electrostatic panels again, I repeated the same tests with the Mirror Image amp as I had used with the Threshold. Every aspect of the sound was improved! Highs were, if anything, even more sumptuously sweet and delicate, and bass was tighter, more detailed, and much better balanced with the upper range. The upper-midrange glare was substantially diminished, too, although it was still there to a moderate degree. In addition, bass/treble balance was markedly improved: the system now had what I felt to be fairly natural balance, with only a hint of heaviness, and it was now possible to run the midrange signal levels up to about 96dB before diaphragm bottoming from strong LF material occurred. (Measured LF output, relative to the upper ranges, was exactly as it had been with the Thresholds, but it did not sound remotely the same. So much for objective testing!)

Although the bass quality was much improved with the Mirror Image amp, it still wasn't great. The low end still lacked punch and impact, and while pitch delineation was now at least what I would call good, it was still not great. Now, this is not to say, necessarily, that the Spectra's low end is a lost cause. Large-panel speakers are notoriously unpredictable when it comes to bass performance in a given room, and it is certain that there are lots of rooms out there in audioland wherein these speakers would probably yield well-balanced, controlled bass performance. But better-balanced LF output will not change the fact that the electrostatic panels of the Spectrum 3s are definitely limited in their ability to handle high levels of heavy bass.

I can see why it was necessary to include the "optional" woofer. Without it, the Spectra 3 hardly qualifies as an audiophile loudspeaker. With the woofer, however, and the Mirror Image amplifier (or any other possessing a similarly lean low end), these speakers do a very respectable job at midrange levels well up toward 103dB, which is quite enough for most classical listeners but still shy of what a hard-rock nut is going to demand. (The saving grace here is that many rock recordings don't have a damned thing below 60Hz. Unfortunately, a few do. Sheffield's Track Record, for example, was bottoming out the cone woofer at LF levels of 105dB, when the measured average SPL was only around 97dB.)

Summing Up
In conclusion, I must qualify the foregoing by saying that I have an unusually large listening room (24' by 19' by 9'), and tend to listen at higher levels than many people—particularly those with neighbors nearby. The Spectra 3's output-level capabilities may well suffice for many listeners, who can then appreciate the system's incredible transparency, speed, delicacy, and midrange naturalness. But unless used with a very polite-sounding amplifier, the Spectra 3's hardness is going to be difficult for any critical listener to tolerate. It's not that there's a lot of it; it's just that I do not expect to find any hardness in a system at this price level.

To sum up, the Spectra 3 is a very good loudspeaker; it falls short in some areas of performance, but in my opinion, this does not detract from its overall appeal.

---

Footnote 2: The easiest way to make sure two panel speakers have the same tilt angle is to use a plumb line from the top of the panel to its bottom edge. (An effective substitute for the real thing can be made by tying one end of some sewing thread—still attached to its bobbin—to a small hex nut. With the bobbin on end on top of the panel, you can rotate it until the nut is at the right height above the floor.) Drape the plumb line over the top of the panel, on the side toward which the panel is leaning, and adjust until the plumb weight (or the nut) is level with the bottom edge of the panel. When the measured distance between the thread at the top of the nut and the surface of both panels is the same, their tilt angles will be the same.