EgglestonWorks Andra II loudspeaker Measurements
For logistical reasons, the samples I measured were different samples from the ones auditioned by Paul Bolin. My estimate of the Andra II's sensitivity was slightly but inconsequentially lower than specified, at 88dB(B)/2.83V/m. This is still a dB or so above the average sensitivity of the speakers I have measured over the past decade, but more important, the EgglestonWorks is a very easy speaker to drive. Fig.1, which shows its electrical impedance magnitude and phase, reveals that it drops below 4 ohms only briefly in the octave between 50 and 100Hz, and that the phase angle is generally low. The small peak at 39Hz indicates the tuning of the sealed-box enclosure—although the twin woofers are mounted in an isobaric arrangement, this doesn't affect the nature of the loading—suggesting only moderate bass extension for such a large speaker.
Fig.1 EgglestonWorks Andra II, electrical impedance (solid) and phase (dashed). (2 ohms/vertical div.)
However, the moderate 12dB/octave rolloff of a sealed-box speaker, in conjunction with the low-frequency gain offered by a typically sized room, will mean that the Andra II's perceived extension will be greater than this measurement implies, as PB heard in his auditioning In addition, the nearfield responses of the drive-units, shown in fig.2, indicate that the Andra's output peaks significantly between 40 and 90Hz. About 3dB of this rise will be due to the nearfield measurement condition, but the rest is real, according to the pink-noise responses I took (not shown). Compounding the nature of the speaker's low-frequency behavior is the fact that there is significant overlap between the woofers and the midrange units. The latter appear to extend an octave lower than both the specified 130Hz crossover frequency and the nearfield measurements supplied me by Albert von Schweikert, the leader of EgglestonWorks' design team.
Fig.2 EgglestonWorks Andra II, nearfield response of midrange units (blue), woofers (red), and the complex sum (black), the last spliced at 300Hz to the farfield response.
If I reduce the woofer level in this graph (red trace) by 10dB and imagine the midrange output (blue trace) without the 3dB boost below 200Hz or thereabouts due to the nearfield technique, I get something that looks very similar to AvS's graph. However, if I follow the recommendations of DRA Labs' Doug Rife and scale the individual drive-units' outputs in the proportion of the square roots of their radiating areas, I get the relative levels shown in fig.2.
The black trace in fig.2 shows the complex sum of the lower-frequency nearfield responses spliced at 300Hz to the Andra II's farfield output. I have no explanation for the swaybacked shape of this trace between 150Hz and 600Hz, unless it really is plotted 4-5dB too high in level. The same response trace is shown to the left of fig.3, which shows the speaker's farfield response averaged across a 30 degrees horizontal window on the tweeter axis. Unlike its predecessor, which featured a large suckout at the upper crossover frequency, the Andra II is impressively flat from the low treble upward, with just a small trough visible in the presence region.
Fig.3 EgglestonWorks Andra II, anechoic response on tweeter axis at 50", averaged across 30 degrees horizontal window and corrected for microphone response, with the complex sum of the woofer and midrange responses plotted below 300Hz.
Allowing for different microphones and measurement techniques—I use DRA Labs' MLSSA, AvS uses LMS from LinearX—my curve agrees pretty closely above 1.3kHz with that supplied me by AvS. (The small dip between 5 and 6kHz appears in both our curves and is due, AvS says, to diffraction.) But between 300Hz and 1.2kHz our responses are different. Both show a fairly narrow suckout in the upper midrange, but mine is centered on 1kHz instead of the manufacturer's 800Hz. AvS mentions in his cover letter that this dip is due to a suspension resonance in the Morel midrange units; perhaps it varies from sample to sample of this otherwise well-regarded drive-unit.
Because of the Andra II's bulk, I was unable to lift it onto my Outline speaker turntable for the acoustic measurements, and had to examine its lateral dispersion by rotating it by hand on a small dolly. Therefore, rather than take measurements at 5 degrees intervals all the way out to 90 degrees off-axis, as I usually do, I performed response measurements at 15 degrees intervals from 15 degrees onward. The result is shown in fig.4. Though the off-axis traces are smooth, a slight off-axis flare can be seen in the presence region. In typical rooms this will balance the slight depression in the same region. The Esotar soft-dome tweeter becomes quite directional above 10kHz, which will lead to a slight lack of top-octave air in very large rooms.
Fig.4 EgglestonWorks Andra II, lateral response family at 50", from back to front: differences in response 90 degrees-5 degrees off-axis, reference response on tweeter axis, differences in response 5 degrees-90 degrees off-axis.
In the vertical plane (fig.5), the speaker's response doesn't change much for a listening axis between the upper-midrange unit and the top of the enclosure (34"-46"). If you stand, however, a suckout appears at the upper crossover frequency.
Fig.5 EgglestonWorks Andra II, vertical response family at 50", from back to front: differences in response 15 degrees-5 degrees above tweeter axis, reference response, differences in response 5 degrees-10 degrees below tweeter axis.
The Andra II's step response (fig.6) indicates that the tweeter and midrange units are all connected with inverted acoustic polarity, the woofers with positive. The large ripple at the 5.5ms marker is the first reflection of the woofers from the floor—I couldn't lift the speaker off the ground for this measurement—and the small one just before the 8ms marker is that of the midrange units. As a result of these early reflections, I had to aggressively window the EW's impulse response when I calculated its cumulative spectral-decay plot (fig.7). The on-axis dip at 1kHz is associated with some delayed energy, implying that it is indeed due to a resonance, as suggested by AvS. There is also some low-level delayed energy around 5kHz. But other than these phenomena, the plot is generally clean.
Fig.6 EgglestonWorks Andra II, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Fig.7 EgglestonWorks Andra II, cumulative spectral-decay plot at 50" (0.15ms risetime).
Finally, the impedance plots were free from the small wrinkles and discontinuities that would imply the presence of cabinet resonances. Using an accelerometer, I wasn't able to detect any such resonant modes at a significant level
Other than its bass performance, which reminds me of that of the larger Alón speakers that have also performed well in Stereophile reviews, there is nothing in the Andra II's measured performance that contradicts Paul Bolin's very positive listening impressions. My thanks to Albert von Schweikert for supplying me with an almost complete set of graphs. It is always a great help for me to see the designer's own measurements.—John Atkinson