Floor Loudspeaker Reviews

Sidebar 4: Measurements

Wilson Audio Specialties optimistically rates the WITT's sensitivity at a fairly high 90dB/W/m. In practice, I measured 87.5dB, which is about the industry average. I was able to A/B a Wilson WATT/Puppy 5 with a WITT in mono: I verified both the higher 91dB specification for the '5 and the easily-calibrated 4dB or so drop to the WITT. Both are sufficiently neutral transducers to allow this calibration to be made on music program as well.

You might logically expect, therefore, a given amplifier to play louder through the System 5 than through the WITT. That depends, however. A tube amplifier is susceptible to loading, for example, and has its output power optimized when loaded by its rated matching impedance. This means that a lower-impedance speaker may not necessarily sound louder, due to the mismatch between the peak-voltage:current ratio available from the amplifier. Such a mismatch can even result in a loss of delivered power (footnote 1). On the other hand, a strong, load-tolerant, solid-state amplifier will almost double its true delivered power into a halved impedance. Conversely, into 16 ohms it will deliver almost exactly half power.

While the WITT can be seen to be an easy 8 ohm load (fig.1), the '5 is not! (See JA's measurements in WP's review last November, p.106.) With a tube amplifier, where impedance matching matters, the practical loudness of the two speakers will in fact be very similar. Only with solid-state drive, unfazed by the WATT/Puppy's 2.5 ohm minimum impedance, will that extra 4dB of measured "voltage" sensitivity be realized as truly greater loudness.

Fig.1 Wilson WITT, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).

How favorable is the electrical loading presented by the WITT? As speakers go these days, very favorable indeed. Fig.1 show an average value of 7 ohms, not exceeding a range of 5.5–10 ohms in the critical 50Hz–10kHz range. This desirably small variation just scrapes the 5 ohm line at the 30Hz port resonance.

The WITT's mild current draw and comparatively moderate variation in impedance offer the promise of consistent sound quality with a wide range of amplifiers and speaker cables. Smaller amplifiers will not feel crushed by this loading; likewise, solid-state dreadnoughts may aspire to unexpected subtlety when so lightly loaded. In concert with the mild variation in impedance magnitude, note the equally mild reactive content shown in fig.1: the electrical phase shift ranges from just –20 to +25° between 25Hz and 5kHz, increasing to a mild 35° in the final octaves. (This latter is primarily a function of the usual normal inductive component of the tweeter voice-coil.)

Because the Wilson Audio WITT is so large and heavy, some of the in-the-air measurement I normally carry out with smaller speakers was impractical. Some loudspeakers offer the convenience of multiple inputs, which allows selective assessment of individual drive-unit ranges. Not so with the single-wired WITT. However, as the technical analysis proceeded, it became evident that the WITT's drivers operate with a significant overlap beyond their nominal ranges. This, together with the WITT's significant height, meant that results for my usual, nominal 1-meter measuring distance would inevitably carry errors owing to the differences in path length from the drivers to the microphone. Despite increasing influence from reverberant room energy, checks were made at 2m and 3m distances in order to gain a clearer picture of the overall frequency balance that would be achieved for a normally sited listener.

In such cases, the room-averaged response is very helpful. It both provides good smoothing or integration of short-term boundary irregularities and reveals more of the overall energy put into the room by the speaker. The measurement is still firmly weighted by a good proportion of the loudspeaker's first-arrival signal, this considered by many to be a loudspeaker's primary acoustic signature.

Nearfield assessment of a loudspeaker's low-frequency output can be complicated by delay and phase errors, between multiple bass drivers, or between bass drivers and reflex ports for example. The room-averaged response can give a more helpful picture of the actual in-room bass extension and uniformity. And by definition, the measurement includes the real-world low-frequency effects of boundary loading—in this respect, what you measure relates well to what you hear.

Examining first the quasi-anechoic on-axis response (fig.2) and noting that the open-cell foam grilles have no significant effect on the frequency response readings, the WITT meets ±3dB limits from 40Hz to 20kHz. Sections of the trace are particularly uniform. For example, the 2–12kHz region is textbook flat, meeting ±2dB limits. Above that region, however, the upper treble has a peak of +5dB at 16kHz (just audible on rougher music program), and an inaudible peak of +6dB at 22kHz. Any high-Q (+10dB or more) "oil-can" tweeter dome resonance is noticeable by its absence. This is presumably due to the unusual tweeter construction—a 15mm voice-coil driving the inner diameter of the 25mm concave titanium diaphragm. The "Dioxid" surface treatment of the titanium foil doesn't seem to have significantly affected the response compared with the untreated versions.

Fig.2 Wilson WITT, anechoic response on tweeter axis at 1m (5dB/vertical div.).

The WITT's 2–8kHz "presence" range is mildly depressed relative to the midrange. This lends the balance a mildly "distant" perspective, something that I noted on audition. The bass/midrange integration looks good, while the –6dB point was established at a low 32Hz. In-room, the perceived bass extended a little more, to typically 27Hz. As with the NHT 3.3, there's no need to consider a subwoofer even if the WITT does not go quite as deep as the NHT.

Exploring the response variation above and below the tweeter axis, the only significant change to be found was at 1.5kHz. The variation was around ±7dB for about a quarter-octave, confirming the broad acoustic overlap between mid- and treble drivers. Above the tweeter axis, the response featured a dip of 7dB, while below-axis, it was a touch "hot" in the same region. If your listening seat is on the low side, you might consider adjusting the front and rear spikes to tilt the speakers down a little. You won't notice any real difference in appearance, since the top of the speaker is sloped.

In the horizontal or lateral plane, the WITT showed remarkable uniformity, a tribute to its low-diffraction design and good acoustic integration. Key points are the lack of off-axis "holes" and the low level of overall loss. An off-axis notch developed at 16kHz 45° off-axis; this is normal for a 1" tweeter. These results confirm the good room sound achieved, a function of an even energy output both on- and off-axis.

The room-averaged 1/3-octave response is shown in fig.3. In my listening space, the bass range is slightly elevated by 3dB relative to the midrange. This correlates with the mildly rich, warmhearted character of this design. Just how well the usual floor reflection at this point is controlled by this speaker is shown by the dip of just 2.5dB in the lower midrange. The upper crossover point is barely visible in the measured response, so well-integrated is the design. Overall, the WITT delivers a fine result for room drive. Again like the NHT 3.3, you can wave a measuring microphone in its general direction and capture a surprisingly smooth response, another hallmark of good system design.

Fig.3 Wilson WITT, spatially averaged, 1/3-octave averaged response in MC's room (5dB/vertical div.).

Examining the individual driver outputs (with caution, since the nearfield technique was essential here), the 12" woofer nominally runs from 60Hz to 160Hz, but has a slow rolloff at the upper edge of its passband. It is still making a significant contribution to the midrange right up to 1.5kHz, after which point the rolloff steepens.

The large rectangular duct/port peaks or resonates with the enclosure volume at 30Hz, this frequency the impedance minimum of the loaded bass driver revealed by fig.1. Like the Puppy, the WITT's port output has something of a bandpass character. Due to a combination of tuning ratio and system Q, the port output is still only –3dB by 80Hz. A minor harmonic resonance could be seen at 400Hz, but with a level 20dB below the mean, this is pretty harmless.

The midrange unit;s coverage was very broad, estimated at 70Hz to 2.5kHz, –6dB, with monotonic rolloffs at either extreme. No significant resonance could be seen in the outputs of the midrange unit's reflex ports, which are located on the rear panel. They are tuned fairly high in frequency and are so well-damped that they act as a purely resistive termination. While the upper crossover point is placed at approximately 2.5kHz, the measurements revealed significant output from the tweeter down to 1.3kHz. This tweeter type is specially chosen for extended duty.

The WITT's impulse response (fig.4) shows mixed acoustic polarity between the drivers, some moderate ringing from the tweeter at the edge of audibility, and a well-damped transient output in the lower range. The Blackman-Harris–weighted Energy-Time Curve (fig.5, dotted trace) is particularly good, with the first 35dB of fast decay handled very smoothly.

Fig.4 Wilson WITT, impulse response on tweeter axis at 1m (3ms time window, 20kHz bandwidth).

Fig.5 Wilson WITT, Energy-Time Curve, unweighted (solid trace); Blackman-Harris–weighted (dotted trace) (5dB/vertical div.).

Turning to the waterfall plot (fig.6), the rapid clearing at the back of the graph is commendable. It correlates very well with the crisp transient definition of the WITT. Note that the in-band 16kHz mode isn't very serious in resonance terms in contrast to the inaudible mode at 22kHz, which shows a stronger resonant character. The WITT's phase linearity is actually pretty good, as shown by fig.7. Lacking short-term rotation, the phase lags smoothly from the mid-100Hz region right through to 20kHz; indeed, a clean waterfall display couldn't have been achieved without a fair measure of phase control in this design.

Fig.6 Wilson WITT, cumulative spectral-decay plot at 1m.

Fig.7 Wilson WITT, wrapped phase response on tweeter axis at 1m.

Power tests established that while the WITT is not as low in distortion as the WATT/Puppy 5, and even less so than the X1/Grand SLAMM, it performs with merit, nonetheless. That big bass unit in particular, efficiently reflexed to control cone excursion, need not move far to generate clean, high levels of low frequencies. I was able to feed the speaker 75W of 30Hz with negligible audible distortion, this corresponding to a sound pressure level of around 104dB at 1m, which is pretty impressive at this floor-shaking frequency. At a more typical 90dB listening level, the second-harmonic reached 0.5% in the midband, with the third at 0.25%. In the bass, both second- and third-harmonic distortion were below the audible threshold; below 1% right down to 40Hz. Over the rest of the range, the second-harmonic was typically 0.2%, with the third a commendable 0.12%. At a louder 96dB spl the results were around 6dB poorer, in line with the usual relationship in loudspeakers, which links power increase with distortion in a linear manner. Even so, the general average for second- and third-harmonic was more than satisfactory at better than 0.6%.

Peak power testing showed that amplifiers rated at up to 200Wpc into 8 ohms are suitable with the proviso that if the amplifier is severely clipped or overloaded, replaceable Caddock power resistors will begin to blow in the WITT crossover. As the input power rose, I had no intimation of compression or limiting. At 200W peak program there was no perceptible increase in distortion or loss of clarity.

In an average room, a pair of WITTs will deliver up to 105dBA of clean sound, ample for all but the most dedicated head-bangers. For comparison, with appropriate drive the WATT/Puppy 5 will get you 110dBA, the X1/Grand SLAMM 115dBA.

Although I knew that the system's full dynamic capability wasn't being exercised, I successfully drove the WITT with 45Wpc of Conrad-Johnson MV55 for civilized reproduction of orchestral program in a moderate-sized room. It's easy to get carried away by high sound levels, as they do act like a drug—the MV55 lent a sense of proportion of the proceedings. Likewise the 40Wpc Exposure 20 and 50Wpc Musical Fidelity A1000 integrated amplifiers performed well with the WITT. I would set 40W as the sensible minimum input unless you have unduly sensitive hearing.

The WITT proved well capable of revealing amplifier quality, particularly after it had been properly run in for 100 hours or so. At this point, some judicious tightening of the driver bolts will bring a further gain in dynamic precision and clarity.—Martin Colloms

---

Footnote 1: Typically, an 8 ohm-rated 100W of "tube power" will fall to 60W into 4 ohms, and to 70W into 16 ohms. (The small relative gain in level at 16 ohms is due to the lighter loading, which allows a greater swing of anode volts within the amplifier.)—Martin Colloms