Wilson Audio Specialties Alexandria XLF loudspeaker Measurements
Sidebar 4: Measurements
I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the Wilson Alexandria XLF's frequency response in the farfield, and an Earthworks QTC-40 for the nearfield and spatially averaged room responses.
Performing measurements on such a large, heavy speaker as the Alexandria XLF poses two main problems. The first is practical: shipping the speakers to my test lab was out of the question, so I had to take my test gear to Michael Fremer's house. The second is that the assumption in any farfield acoustic measurement is that the distance from speaker to microphone is much greater than the speaker's largest dimension. There is also the fact that the speaker needs to be well away from any boundary so that reflections of its sound from that boundary don't corrupt the measurement. With a speaker as large as the Alexandria, neither of these conditions can be met without a very large (and very expensive) anechoic chamber.
For the impedance and in-room measurements, the speakers were in their usual positions. We then (carefully) fitted the supplied wheels to one Alexandria and (very carefully) wheeled it outside, to Michael's driveway, for the rest of the acoustic measurements. (It was a clear, windless day.) However, while this eliminated the wall and ceiling boundaries, it was impractical to raise the speaker's 655 lbs off the ground. The reflection of the woofers' output thus curtailed the anechoic time window I could use for the analysis, reducing the measurements' resolving power in the midrange.
My estimate of the XLF's voltage sensitivity was 92.6dB(B)/2.83V/m. While this is slightly below the specified 93.5dB, it is still much higher than normal. Despite the Alexandria's imposing bulk, it will play at high levels with only a few watts. (During the in-room measurements, performed at a reasonably loud level, the darTZeel amplifiers' meters never indicated more than 5W peak.)
Wilson specifies the Alexandria as having a nominal impedance of 4 ohms. Fig.1 confirms this specification, the impedance magnitude (solid trace) varying between 4 and 8 ohms over almost the entire audioband. Though there are minimum values of 3.7 ohms at 19Hz, 3.2 ohms at 510Hz, and 2.35 ohms at 35kHz, the electrical phase angle (dotted trace) is low at these frequencies and generally benign overall, meaning that the speaker will not be a difficult load for the partnering amplifier to drive.
The traces in fig.1 are free from the wrinkles that would indicate the existence of enclosure panel resonances. I had neglected to take my accelerometer and its preamp to Michael's, so I'm unable to offer my usual cumulative spectral-decay plots of the walls' vibrational behavior. However, other than the vertical "wings" that flank the midrange-tweeter-midrange (MTM) array, all the enclosure walls were acoustically inert, to judge by the knuckle-rap test. This was confirmed by listening to the walls with a stethoscope while swept sinewave tones were playing.
The Alexandria's MTM array needs to be focused on the listening position, using the precision adjustments on the speaker's rear. I measured the height of MF's ears as he sat in his listening seat: 37". I then measured the height of the Wilson's front-firing tweeter56.5"and the distance from the tweeter to his ears: 94". Once we had the speaker outdoors, we placed the microphone at exactly this position and ran some response measurements. Unfortunately, the reflection of the upper frequencies from the ground followed the direct sound by just 3 milliseconds, meaning that the resolution of the measured response was 333Hz; ie, the datapoints lie at 333Hz and its multiples.
Fig.2 shows the Alexandria's response above 300Hz, with the speaker angled away from the microphone by 10°, which was the offset in Michael's listening room, and without the grilles, which Michael left off for his auditioning. Slight peaks in the upper midrange and mid-treble are balanced by a slight lack of energy in the presence region. Whether the peaks are audible as added detail and brightness or the presence-region dip is heard as "politeness" and a laid-back, forgiving nature will depend very much on the music being played, which in turn will determine which frequency band the ear latches on to as being its reference.
To get a more detailed look at the Wilson's behavior in the frequency domain, I moved the microphone forward, along the line connecting the height of MF's ears to the height of the tweeter, until it was at my usual distance of 50". The black trace in fig.3 shows the Alexandria's output averaged across a 30° horizontal window centered on the tweeter. The response is generally very flatflatter than at the listening distancebut with the lack of energy between 2 and 4kHz still apparent. The output of the soft-dome Convergent Synergy tweeter smoothly extends at full level almost to the 30kHz limit of this graph, whereas the inverted titanium-dome tweeter used in earlier Alexandrias, as well as in the MAXX 3, peaked at the top of the audioband.
The green trace in fig.3 shows the output of the midrange drivers, taken in the nearfield. It rolls off smoothly below 150Hz, crossing over to the woofers just above 100Hz. The bottom woofer has a radiating diameter of 12.5", the upper woofer 10". However, their outputs, measured in the nearfield, were virtually identical, so fig.3 shows their summed output (blue trace). With a passband covering 40120Hz, the excessive level of the woofers compared with that of the midrange units is entirely a function of the nearfield measurement technique, which assumes a 2pi or half-space environment extending to infinity in both the vertical and horizontal planes. There is a peak between 700 and 800Hz in the woofers' upper-frequency output, but this is suppressed by the crossover.
The saddle centered at 19Hz in the impedance-magnitude trace suggests that this is the tuning frequency of the large rectangular port, which, in MF's review samples, was open to the speakers' rear rather than to the front. However, the minimum-motion notch in the woofers' summed output actually occurs at 21Hz, while the port's output, again measured in the nearfield (red trace), peaks just below 20Hz but doesn't roll off until above 70Hz. The black trace in fig.3 is the summed output of all the lower-frequency radiators, taking into account acoustic phase and the different distances from a nominal farfield microphone position. Though it peaks between 45 and 125Hz, this again will be almost entirely a function of the nearfield measurement technique. The port doesn't fully reinforce the woofers' output at its tuning frequency, which is appropriate, given that when the speakers are in a room, the boundary effect will increase the port's output level to give a response that should extend down to 20Hz at full level.
Figs.4 and 5 show how the Alexandria XLF's response changes to the sides and above and below the listening axis. (Because the speaker is too bulky and heavy to position on my speaker turntable, I've shown the changes over a limited window.) Laterally, the speaker's output shows very little change in its output up to 15° to its side. In the vertical plane, the Alexandria XLF's response doesn't change significantly from 5° above to 15° below the listening axis. This speaker appears to be much less fussy regarding the exact listening axis than the Wilson MAXX 3s that previously enjoyed pride of place in MF's system (see figs.4 and 5 here).
In the past I have argued that perhaps the most meaningful measurement of a loudspeaker is of the pair's spatially averaged response at the listening position. Using SMUGSoftware's FuzzMeasure 3.0 program and a 96kHz sample rate, I average 20 1?6-octavesmoothed spectra, taken for the left and right speakers individually, in a vertical rectangular grid 36" wide by 18" high and centered on the positions of the listener's ears. This largely eliminates the room acoustic's effects, and integrates the direct sound of the speakers with the in-room energy to give a curve that I have found correlates reasonably well with a speaker's perceived tonal balance.
The red trace in fig.6 shows the spatially averaged response of the Alexandria XLFs in MF's listening room, driven by darTZeel NHB-458 monoblocks via TARA Labs speaker cables. It is even between 300Hz and 6kHz, with small peaks balanced by similarly small dips. Above 6kHz the response smoothly slopes down, due to the increasing absorptivity of the room furnishings in the high treble. At low frequencies, the Alexandria's in-room response is basically identical to that of the MAXX 3 (blue trace)hardly surprising, given the similar woofer configuration and positioning in the room. A lack of energy in the upper bass is followed by small peaks at 80 and 50Hz, the frequencies of the lowest resonant modes in MF's room. However, the Alexandria has a little more low-bass energy, and the upper-bass dip is less extreme. Perhaps of more importance, the new speaker shows a little more energy in-room in the upper midrange and low and high treble, giving a more evenly balanced response overall.
A couple of weeks before driving to Michael's place I had auditioned another pair of Alexandria XLFs, at Manhattan dealer Innovative Audio, where they were driven by Dan D'Agostino Momentum monoblocks and a preproduction sample of D'Agostino's Momentum Ultra-Analog preamplifier, wired with Transparent Reference XL speaker cables and balanced interconnects. The ports were open to the rear, as they had been in MF's room. With a Meridian MediaSource 600 feeding data to a dCS Scarlatti D/A converter, this system gave the best sound I have heard at Innovative. I therefore asked Innovative's Scott Haggart if I could measure the Alexandria XLFs' spatially averaged response in the big room where I had auditioned them. Innovative's owner, Elliot Fishkin, was amenable; the result is the green trace in fig.7. (The red trace is, again, the speakers' spatially averaged response in MF's room.)
Above 250Hz, the XLFs' response in the Innovative listening room meets very tight limits: ±1.5dB. Again, small response peaks are balanced by small dips, but the response trend between 250Hz and 6kHz is a little flatter than in Michael's room. This may well be due to the greater distance to the listening position at Innovative: 128" vs 94".
The slight slope down above 6kHz will be due to the increased absorption of the room furnishings in this region, but I suspect that the plateauing of what would otherwise be a smooth rolloff between 10 and 18kHz in the Innovative room will be due to the effects of the XLF's rear-firing tweeter. The sharp rolloff above 18kHz in both rooms will be due to the 1" front tweeter's inevitable increasing directivity at frequencies at which it is larger than the wavelength of the sound it is emitting.
The spatial averaging has not entirely eliminated the effect of the low-frequency modes in the Innovative room, but the buildup of bass energy due to the proximity of the room boundaries will be mitigated by the overdamped woofer alignment. The effect will be to add some bass weight without obscuring clarity. The low-frequency extension is excellent, the level at 20Hz being the same as at 1kHz.
The lower-midrange dip at Innovative is both narrower and higher in frequency than the corresponding dip in MF's room. I suspect this is due to interference between the direct sound from the midrange units and the reflection from the sidewalls. (The floor-bounce cancellation of the woofers' output will be above their passband and is thus inconsequential.) The 200Hz dip is inevitable given the room dimensions, and I suspect that much of the fine-tuning of the speakers' positions is to arrange for this cancellation notch to have the least effect on music.
In the time domain, the Alexandria XLF's step response at the listening position (fig.8) indicates that the tweeter and woofers are connected in positive acoustic polarity, the midrange units in inverted polarity. However, the decay of the tweeter's step smoothly blends into the start of the midrange units' step, and the decay of the midrange units' step smoothly blends into the start of the woofers' step. This indicates optimal crossover design, which, in combination with the adjustable geometry of the upper-frequency drivers, will give the smooth blending of the drive-unit outputs in the frequency domain claimed for the Aspherical Group Delay technology.
The cumulative spectral-decay plot (fig.9), taken at the closer 50" microphone distance in order to push back boundary reflections, shows a very clean initial decay, but with then some low-level hash in the mid-treble. A small ridge of what appears to be delayed energy is visible at the computer monitor's line-scan frequency just below 17kHz. This is spurious and should be ignored. However, there is a notch in the on-axis response, and a more powerfully defined ridge of delayed energy at a lower frequency, 14,875Hz, both of which appear to be real.
As I said at the beginning of this section, there are practical limitations when measuring so large a loudspeaker. While I am confident that my measurements regime fully characterizes the performance of a small speaker (such as KEF's LS50, which I reviewed last month), with a speaker as large as Wilson's Alexandria XLF, the measurements offer suggestions rather than certainties. Note, also, that I don't measure distortion, which in this high-sensitivity speaker is likely to be very low. But overall, this is an impressively well-engineered design. As well as auditioning the Alexandria XLF at Innovative and in Michael's room, last January I had the opportunity to perform extended comparisons between the XLF and the earlier Alexandria X-2 Series 2 at Wilson Audio's headquarters. There is no doubt in my mind that this is the best speaker yet to come from the Utah company, which makes it a very fine speaker indeed.John Atkinson