Impact Airfoil 5.2 loudspeaker system Measurements
Given the Airfoil 5.2 system's bulk, it made sense to take advantage of a road trip I was making through the Southwest last fall to measure the speaker in Brian Damkroger's Albuquerque listening room. The advantage of this was that I could actually get a handle on the speaker's in-room behavior, which its unusual design made imperative. The disadvantage was that I couldn't take all the measurements that I usually do in my Brooklyn test lab. (Because of my need to keep my traveling rig as portable as possible, I used the Mitey Mike II from Joe D'Appolito rather than my usual B&K 4006. All measurement were performed using DRA Labs' MLSSA system.)
Because I had to test the speaker in-room, I could get only a rough idea of the 5.2's voltage sensitivity, which appeared to be around 80dB(B)/2.83V/m—considerably below the specified 90dB figure. However, this will be ameliorated in-room both by the Airfoil's line-source behavior and by its wide horizontal dispersion.
The main Airfoil tower's plot of impedance magnitude and phase is shown in fig.1. The speaker is a moderately demanding load, with an impedance dropping below 4 ohms for much of the treble region, and reaching a minimum value of 3 ohms just below 2kHz. Above that frequency the impedance gradually rises, presumably due to the Bending Wave driver's voice-coil inductance. While the impedance phase angle reaches an extreme value of 67.3 degrees capacitive at 66Hz, the magnitude at this frequency is a fairly high 13.9 ohms, which will mitigate the effect of the phase.
Fig.1 Impact Airfoil 5.2, electrical impedance (solid) and phase (dashed). (2 ohms/vertical div.)
Deciding on what axis to measure the speaker's frequency response was rather arbitrary, given the nature of the drive-unit. In the end, I chose an axis perpendicular to the driver at the point where it is driven by the voice-coil. (With the speakers as set up by the manufacturer, this was the axis that actually pointed toward Brian's listening chair.) The result at a microphone distance of 50", averaged across a window 15 degrees to either side of this axis, is shown to the right of fig.2. The peaks and dips in the lower midrange are due to room effects. (I used a pink-noise signal to assess the response in the crossover region between the Bending Wave driver array and the two coupling woofers.) The suckout centered at 600Hz is not a room effect, but appears to be characteristic of the speaker. It is very position-dependent, however, and did not affect the Airfoil's measured room response at the listening position (see later).
Fig.2 Impact Airfoil 5.2, anechoic response on listening axis at 50", averaged across 30 degrees horizontal window and corrected for microphone response, with the nearfield woofer and subwoofer responses plotted below 400Hz and 280Hz, respectively.
Eyeballing the trace in fig.2 indicates that the Impact's response gently slopes down as the frequency increases. This will be partly due to the Proximity Effect—the speaker's radiating dimensions are not small compared with the speaker-microphone distance, which means that, even at 50", the mike is still in the nearfield. But it's possible that the top octaves are also depressed in absolute terms—BD did comment on the depressed top octaves in his auditioning notes. And note how ragged they appear; above 5kHz or so, the Bending Wave driver appears to be operating in breakup mode.
To the left of fig.2 is shown the nearfield response of the powered subwoofer module. It exhibits the classic bandpass behavior of a coupled-cavity system, except that it rolls off at 24dB/octave above and below the passband rather than the usual 12dB/octave. Because the subwoofer is powered, its level in this graph has been arbitrarily chosen to give a crossover point at the specified 80Hz.
The Airfoil's lateral dispersion over the ±15 degrees window used to derive fig.2 is shown in fig.3, with the response differences on the drive-unit side (facing away from the listener) to the graph's rear. You can see that the depth of the 600Hz suckout depends very much on the axis, and worsens toward the speaker's rear. Note also how position-dependent the pattern of peaks and dips in the top two audio octaves is—again, the bending-wave drivers are operating in breakup mode at these frequencies and the overall in-room effect may well be quite flat. In the vertical plane (not shown), the line-source behavior of the drive-unit array means that there are very few response changes as the listener moves up and down. (The curves in figs. 2 and 3 were taken at the exact midpoint of the nine-element array, 42.5" from the floor.)
Fig.3 Impact Airfoil 5.2, lateral response family at 50", from back to front: differences in response 15 degrees-5 degrees off-axis on drive-unit side, reference response on listening axis, differences in response 5 degrees-15 degrees off-axis on baffle side.
Fig.4 shows a 1/3-octave-smoothed response, spatially averaged in a grid 36" wide by 18" high centered on the position of BD's ears in his listening chair, with the speaker towers and subwoofers in the positions used in his auditioning. The first thing to note is how smooth the trace is—many of the peaks and dips visible in earlier graphs are so position-dependent that they even out with the spatial averaging. And note how extended the deep-bass response is; in-room, the Airfoil subwoofer's output is down by just a couple of dB at 20Hz. But note, too, the broad peak in the upper midrange and the generally suppressed top octaves. I know BD didn't find the Airfoil to sound colored, but this measurement reveals that it is not a neutrally balanced transducer.
Fig.4 Impact Airfoil 5.2, spatially averaged, 1/3-octave-smoothed, farfield response in BD's listening room.
I must admit that, listening to music, I didn't hear nearly as much emphasis of the upper midrange as I would expect from this graph. There was also an astonishing separation of the acoustic objects on the recordings we auditioned from the speakers' physical locations. Having an array of identical drive-units reproducing everything from the lower midrange up certainly has a lot going for it.
In the time domain, the Airfoil's impulse response (fig.5) is time-coherent and in the correct positive polarity, though the tail of the response is overlaid with the ringing associated with the upper-frequency peaks seen in the amplitude-response graphs. The step response (fig.6) is essentially perfect, with a steep rise away from the time axis followed by a good right-triangle-shaped decay.
Fig.5 Impact Airfoil 5.2, impulse response at 50" (5ms time window, 30kHz bandwidth).
Fig.6 Impact Airfoil 5.2, step response at 50" (5ms time window, 30kHz bandwidth).
The cumulative spectral-decay or waterfall plot of the Airfoil's bending-wave array (fig.7), windowed to eliminate reflections of the sound from the floor, ceiling, and walls of BD's listening room, is a mess, with multiple ridges of delayed energy visible throughout the treble. To be fair, this behavior appears to be typical of speakers using planar or other unconventional drive-units, and in the past I have wondered if what you see in graphs like this is really the effect of multiple arrivals at the microphone position. Certainly the Airfoil's treble doesn't sound grainy or lacking in clarity, which is what you'd expect from a conventional speaker offering a waterfall plot like fig.7.
Fig.7 Impact Airfoil 5.2, cumulative spectral-decay plot at 50" (0.15ms risetime).
Summing up, from the evidence of its measured performance, the Airfoil 5.2 is an enigma.—John Atkinson