Acoustic Research AR 303 loudspeaker Measurements
Other than impedance (for which I used an Audio Precision System One) and the in-room spectral analysis (footnote 1), all measurements were performed with a MLSSA v.8.5 system from DRA Labs, in combination with an Outline computer-controlled loudspeaker turntable and a B&K 4006 microphone calibrated to be flat on-axis at the typical measuring distance I use. To minimize reflections from the test setup, the measuring microphone is flush-mounted inside the end of a long tube. Reflections of the speaker's sound from the mike stand and its hardware will thus be sufficiently delayed not to affect the measurement.
While the new AR 303's impedance does drop to 3.3 ohms in the upper bass and 3.6 ohms in the high treble (fig.1), it's not nearly as demanding a load as the original AR-3a. The shape of the curve, however, does confirm my comments that its already mellow balance will be accentuated if the speaker is driven by an amplifier with a highish source impedance. The magnitude peak at 38.5Hz is due to the sealed-box woofer resonance; there are no wrinkles in either trace to suggest the presence of cabinet resonances. Listening to the panels with a stethoscope revealed a slight amount of "cabinet talk" in the lower midrange present on the front baffle and back panel, but this was low in level.
Fig.1 AR 303, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
Although the 303 is specified as having a voltage sensitivity of 85dB/2.83V/m, my measured B-weighted figure was a little lower, at 84dB. The difference is inconsequential, though a goodly powerful amplifier of at least 50Wpc output is indicated.
Looking at the quasi-anechoic response averaged across a 30 degrees horizontal angle on the tweeter axis (fig.2), the overall response trend is smooth, but sloped-down a little from the upper bass to the high treble. There is also a distinct step in the curve just above 1kHz. This kind of response shape would usually be associated with some audible nasality. The 303, however, seemed relatively uncolored, suggesting that the on-axis peak'n'dip is probably compensated for by the opposite behavior off-axis.
Fig.2 AR 303, anechoic response on tweeter axis at 50", averaged across 30 degrees horizontal window and corrected for microphone response, with nearfield woofer response plotted below 300Hz.
In the bass—measured in the nearfield with the microphone capsule almost touching the woofer dustcap—there's a moderate rise in the midbass, peaking 5dB above the 1kHz reference level at 63Hz, before the speaker's output drops to -6dB at a low 27Hz. The 303's owner gets a lot of bass for the money. Though the owner's manual recommends placing the speakers close to room boundaries, this will have the effect of accentuating the 303's bass output even further.
Despite its wide baffle, the 303 offers reasonably well-controlled treble dispersion, as can be seen in fig.3—note that only the differences in response are shown in this graph. The more even change in balance is to the tweeter edge of the baffle, shown to the front of fig.3. If you have to position these speakers close to the side walls, this suggests that the tweeters should be on the speakers' outside edges; otherwise, place the mirror-imaged speakers so that the tweeters are on their inside edges. The wide baffle does make its presence known, however: despite the use of a small-diameter tweeter dome, the top two octaves fall off more rapidly with off-axis angle than would be the case with a minimonitor. In a typical room, this will contribute to the speaker's rather mellow tonal balance.
Note that fig.3 reveals that, in the midrange, the 303 also has rather more limited dispersion than a typical two-way design. This is probably due both to the crossover performance and to the use of a 12" LF driver. Despite my earlier conjecture, the lack of off-axis energy is a little lower in frequency than the on-axis peak around 1kHz, however. Predicting a loudspeaker's perceived tonal balance from its on-axis response and how that response changes off-axis will never be an easy task—or even, sometimes, a possible one!
Fig.3 AR 303, horizontal response family at 50", normalized to response on tweeter 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.
Vertically (fig.4), the 303 only offers its optimal treble balance over a very small listening window: on or just below the tweeter axis. Above or below that axis, suckouts appear in the crossover regions and the tweeter's top octave rolls off. This makes AR's recommended stand height strange: stands much shorter than 12" will result in a lack of mid-treble energy with the typical listener's ear height of 35-39". My own recommendation would be for stands 12-16" high, though the large 303 does look alarmingly top-heavy on a normal, center-pillar speaker stand.
Fig.4 AR 303, vertical response family at 50", normalized to response on tweeter axis, from back to front: differences in response 45 degrees-5 degrees above tweeter axis; reference response; differences in response 5 degrees-45 degrees below tweeter axis.
Even with the speakers well away from the walls, the in-room response (fig.5) still featured an excess of energy in the bass, as I'd noted in my auditioning. The benefit gained from this, however, is an impressive low-frequency extension, even the 25Hz, 1/3-octave band being almost up in level with the midrange reference level. A slight excess of low-treble energy will probably not be heard as brightness per se. Instead, I suspect it gives the ear a reference point against which the higher-treble balance can be judged. And the 303's in-room balance is somewhat rolled-off in its top two octaves, correlating with my comments on its perceived HF quality. Note, however, that the in-room balance is very flat around 1kHz despite the on-axis peakiness. It looks as though the off-axis behavior does compensate for the on-axis departure from flat in this region—at least in my room.
Fig.5 AR 303, spatially averaged 1/3-octave response in JA's listening room.
In the time domain, the 303's impulse response (fig.6) is non-time-coherent but otherwise unremarkable. Ignore the wrinkles at and after the 7ms mark in this graph, which are due to the first reflections of the speaker's direct sound from the room boundaries. The step response (fig.7) indicates that the tweeter and midrange units are connected with inverted polarity, while the woofer's output is in positive polarity, these checked by examining the step responses of the individual units. The 303's cumulative spectral-decay, or waterfall, plot (fig.8), calculated from the impulse response data, is impressively free from resonant modes. While there's a little bit of low-level hash between 6kHz and 8kHz, and the response step at 1kHz is also associated with some delayed energy, this is otherwise an excellent result, implying good resolution of recorded detail and a relative freedom from treble grain, as I found.—John Atkinson
Fig.6 AR 303, impulse response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Fig.7 AR 303, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Fig.8 AR 303, cumulative spectral-decay plot at 50" (0.15ms risetime).
Footnote 1: For my in-room spectral analyses I average six measurements at each of 10 separate microphone positions for left and right speakers individually, giving a total of 120 original spectra. These are then averaged to give a curve that, in my room, has proved to give a good correlation with a loudspeaker's perceived balance. I use an Audio Control Industrial SA-3050A spectrum analyzer with its own microphone, which acts as a check on the MLSSA measurements made with the B&K mike. I also used the Goldline DSP-30 automated spectrum analyzer (currently under review).—John Atkinson