Infinity IRS Epsilon loudspeaker Measurements
John Atkinson measured the Infinity Epsilon and provided me with the results after my auditioning was complete.
The sensitivity of the Infinity Epsilon measured effectively to specification at a calculated 85.5dB/W/m (B-weighted). The impedance of its woofer section (fig.1) indicates a cabinet tuned to about 38Hz. The load is relatively benign, never dropping below 4 ohms. However, the inevitable rise in impedance at resonance requires me to modify slightly a couple of statements I made in the main text of the review. Note that, between 20Hz and 30Hz, the impedance varies between just over 5 ohms and 9 ohms. This means that the power available from an amplifier will be less than the latter's 4 ohm ratingwhich may partially explain the amplifier/Epsilon combination's unfulfilled demand for more power in the low-bass region on particularly demanding material. Note also the small ripples in the response above 100Hzthese are usually indicative of cabinet resonances.
Fig.1 Infinity Epsilon, woofer electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
The impedance of the upper-range drivers (fig.2), however, does drop below 4 ohms. In particular, the dip to 2.8 ohms just above 200Hz, combined with a significantly capacitive phase angle in this region, makes the Epsilon a challenging load for its upper-range amplifier. Low-powered tube amplifiers, or any amplifier uncomfortable with a load dropping below 4 ohms, should not apply.
Fig.2 Infinity Epsilon, mid/treble array electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
Fig.3 shows the FFT-derived nearfield responses of the woofer (left) and L-EMIM (right). The curve also shows the action of the woofer and midbass controls. The action of the former is quite subtle (though more significant as you get on the steep slope of the curve below about 30Hz), giving 6dB points ranging from 24Hz to 26Hz with a slight change in the amount of energy prior to the rolloff. The midbass contour control gives a boost or cut of up to 3dB but covers quite a narrow frequency range. This, though, is in a region where room resonant problems are common. The acoustic crossover to the L-EMIM, at 150Hz, is as specified.
Fig.3 Infinity Epsilon, nearfield woofer and L-EMIM responses showing effect of Woofer control set to "0" and "±5" and Mid-Bass Contour control set to "0" and "±3" (5dB/vertical div.).
The overall combined frequency response of the Infinity Epsilon, averaged across a 30° horizontal window, is shown in fig.4. Here, the response below 312Hz is the complex sum of the nearfield outputs of the woofer and the L-EMIM; the response above 312Hz was taken at 45" at a height of 37", my seated ear height. The response holds up remarkably well to below 30Hz, and is extremely smooth across the full frequency range. The shallow depression centered at about 56kHz, combined with the rolloff above 10kHz, may explain my comments about the slight lack of "air" in the Epsilon's subjective performance. Otherwise, there is little here to criticize. As JA said, this is an impressively engineered speaker.
Fig.4 Infinity Epsilon, anechoic response on L-EMIM axis, 37" from floor, at 45" averaged across 30° horizontal window and corrected for microphone response, with complex sum of nearfield woofer and L-EMIM responses below 312Hz.
The action of the L-EMIM, EMIM, and EMIT level switches with the Epsilon's on-axis response subtracted, shown in fig.5, is very subtleconsistent with Infinity's specifications. The effects of the controls were, however, audible.
Fig.5 Infinity Epsilon, effects of L-EMIM, H-EMIM, and EMIT switches, normalized to response on L-EMIM axis (5dB/vertical div.).
The horizontal and vertical response families are not shown here. The horizontal response was noncritical within a ±10° window. The off-axis vertical responses, relative to the response on the 37" axis, were very smooth. Sitting considerably lower resulted in a dip at about 3.5kHz (the crossover region between the H-EMIM and the L-EMIM); sitting higher brought up the treble region somewhat. This only became evident at heights above 45"an impractically high listening height (unless you listen on a bar stool).
The impulse response on the L-EMIM axis is shown in fig.6. Ringing is notably absent here, though it's evident that the system is not time-aligned. The latter is clearer in fig.7 (the step response taken at the same location). There's a slight delay between the arrival times of the EMIT and the EMIM, followed in another millisecond or so by the L-EMIM, and another 3ms later by the woofer. All the drive-units appear to be connected with the same (negative) acoustic polarity.
Fig.6 Infinity Epsilon, impulse response on L-EMIM axis at 45" (5ms time window, 30kHz bandwidth).
Fig.7 Infinity Epsilon, step response on L-EMIM axis at 45" (5ms time window, 30kHz bandwidth).
Finally, the cumulative spectral-decay, or waterfall, plot is shown in fig.8. The behavior here is excellent, particularly at the top of the range, where there's almost no visible hash. (Ignore the ridge just below 16kHz, which is due to our measurement computer's screen.) A bit of resonant behavior is evident in the low and mid-treble, but this is relatively innocuous, being very low in level.
Fig.8 Infinity Epsilon, cumulative spectral-decay plot at 45" (0.15ms risetime).
This is a first-rate set of measurementscertainly among the best we've measured, and consistent with the Epsilon's superb listening quality.Thomas J. Norton