InnerSound Eros Mk.III electrostatic loudspeaker Measurements part 2
The high-pass rolloff features a steep 30dB/octave slope, this arising from the fourth-order crossover slope being increased by 6dB/octave by the usual dipole action. (At the frequency drops, there is an increasing amount of cancellation of the front wave by the antiphase backwave.) This is just as well; without the crossover, the panel has a significant peak at 86Hz—which, as I noted above, is the frequency of its "drumskin" resonance. This is so strong that you can hear it coloring your voice when you speak close to the Eros. It's fair to point out that the panel will not be fed an electrical signal that will excite this resonance. However, the panel is in close proximity to the woofer, which puts out high levels of energy in this region, and it will be excited. All things being equal, I would have thought it audible with music. Certainly I could hear it with pink noise as a slight "hummy" quality.
Returning to fig.8, the woofer (red trace) rolls off with an approximate 18dB/octave slope above 400Hz. (With the crossover's Midrange control set to "90," as it was for this measurement, the woofer's level appears to be too high in absolute level, but 3dB of this will be due to the nearfield measurement technique.) Its output is fairly smooth within its passband, but there is a notch apparent at 120Hz. This is also the frequency of a small peak in the port output, which, like the bumps in the impedance graph, suggests a resonant problem in the transmission line. The port's higher-frequency response is disturbed by some peaks. Lower in frequency, it offers only a small degree of bass extension compared with a reflex design.
How all these responses sum in the nominal farfield is shown in fig.9. Again, not too much should be made of the apparent downward slope above 1kHz, which will, to some extent, be due to the proximity effect. However, the response trend is overall very smooth.
Fig.9 InnerSound Eros Mk.III, Midrange = "90," Bass = "6," anechoic response on mid-panel axis at 50", averaged across 30 degrees horizontal window and corrected for microphone response, with the complex sum of the nearfield responses plotted below 300Hz.
Regarding the phasing of the drive-units: When I connected them as the manual recommends, I got a sharp cancellation notch at 490Hz. To get the much smoother measured farfield integration between the panel and the woofer shown in fig.9, I had to invert the woofer polarity. As the mostly excellent manual states, however, Eros owners should experiment with this in their rooms and choose the woofer phase that sounds best to them.
The traditional problem with an electrostatic speaker that uses a flat panel is that its radiation pattern beams severely in the frequency region where the wavelength is smaller than the speaker's size. This is why MartinLogan and SoundLab use curved panels, and why Quad uses a multiple time delayed-ring approach. Fig.10 reveals, however, that the Eros is capable of delivering a full measure of high frequencies in the horizontal plane only if the listener sits exactly on-axis. This explains Larry Greenhill's comment that the sweet spot "seemed only millimeters in diameter," and it is fair to note that this restriction of the speaker's HF dispersion is deliberate. In the vertical plane, the panel's 41" dimension is large enough to result in some line-source behavior, which means there should be only moderate changes in balance over quite a wide range of listener ear heights (fig.11). Nevertheless, LG found that not to be the case.
Fig.10 InnerSound Eros Mk.III, lateral response family at 50", normalized to response on mid-panel axis, from back to front: differences in response 90 degrees-5 degrees off-axis, reference response, differences in response 5 degrees-90 degrees off-axis.
Fig.11 InnerSound Eros Mk.III, vertical response family at 50", normalized to response on mid-panel axis, from back to front: differences in response 15 degrees-5 degrees above axis, reference response, differences in response 5 degrees-15 degrees below axis.
In the time domain, the step responses of the panel (fig.12, red trace) and woofer (blue) are both positive-going, though the woofer's output arrives a while later and is opposed by the panel's output, which is negative-going by that time. This is why I got better measured integration between the two disparate drive-units by reversing the woofer polarity. Note the excellent right-triangle shape of the panel's arrival, which suggests a time-coherent presentation, at least over its passband. The cumulative spectral-decay plot (fig.13) features a much cleaner treble region than is usually the case with panel speakers. There appears to be some delayed energy apparent in the midrange, however, though the resolution in this graph is not good enough to reveal what is going on.
Fig.12 InnerSound Eros Mk.III, step response on mid-panel axis at 50" of electrostatic panel (red trace) and woofer (blue). (5ms time window, 30kHz bandwidth.)
Fig.13 InnerSound Eros Mk.III, cumulative spectral-decay plot at 50" (0.15ms risetime).
One I'd figured out how to use it, the InnerSound crossover-amplifier got a clean bill of health. The Eros Mk.III itself, however, is an enigma. Its clean treble decay and smooth response suggest that a listener sitting in the speakers' tiny sweet spot will experience excellent sonic clarity and a superb ease to the music's presentation, at least in the midrange and above, that could prove addictive. But the woofer's use of transmission-line loading is nothing I could be enthusiastic about; it introduces the possibility of upper-bass coloration and as LG found, it does not significantly extend the speaker's low-frequency limit.—John Atkinson