KEF Reference 207 loudspeaker Measurements

Sidebar 3: Measurements

Two practical problems hindered my acoustic measurements of the KEF Reference 207. The first is common to large speakers in that their bulk makes it difficult to lift them off the ground. The need to window out the inevitable early reflection of the speaker's sound from the floor restricts the midrange resolution of the FFT-derived graphs. The second was specific to this review in that the air temperature on the day I measured the KEF was just 40 degrees F. (Quasi-anechoic measurements require a large room, so I perform them outdoors to eliminate boundary reflections.) I first hit the speaker hard with high-level pink noise for several minutes to make sure the voice-coils were warmed up, but the physical properties of the cone and surround materials may well have been affected by the temperature.

The 207's sensitivity was significantly higher than average, at an estimated 92dB(B)/2.83V/m. While the speaker is specified as being an 8 ohm design, its impedance magnitude (fig.1) ranged between 3.3 ohms and 6 ohms for much of the audioband. This suggests that a 4 ohm rating would be more appropriate, though the electrical phase angle is relatively low most of the time. Using the jumpers to apply the low-frequency boundary compensation raises the bass impedance by an ohm or so. However, plugging in the HF contour jumper gave very little rise in impedance. The sharp impedance peak at 46kHz is presumably due to a notch filter used to tame the supertweeter's dominant diaphragm resonance.

Fig.1 KEF Reference 207, electrical impedance (solid) and phase (dashed) with boundary and HF compensation in (top) and out (2 ohms/vertical div.).

The impedance traces are free from the small discontinuities that would indicate the presence of panel resonances. In fact, neither listening to the walls of the elliptically curved bass enclosure with a stethoscope nor measuring its impulse response with an accelerometer revealed any problematic modes. Even the cast-aluminum Uni-Q pod seemed dead to the touch.

The individual outputs of the six drive-units and two ports are shown in fig.2. Looking first at the higher-frequency behavior, assessed in the farfield on the Uni-Q tweeter axis, the crossover between the upper-midrange unit and the tweeter is set as specified at 2.7kHz, but is disturbed by a small peak in the former's output almost exactly at crossover. The tweeter shelves up by a couple of dB in the region where it is about to hand over to the supertweeter, but the range nominally covered by the supertweeter actually appears suppressed. (As I mention in my auditioning notes, the supertweeter appears to come in earlier than the specified 15kHz.) The acoustic crossover slopes are actually a little steeper than the specified 24dB/octave, presumably because they consist of a fourth-order electrical slope to which must be added the acoustic rolloff of each drive-unit.

Fig.2 KEF Reference 207, acoustic crossover on Uni-Q axis at 50", spliced to the nearfield upper and lower midrange, upper woofer, and upper port responses (black traces), plotted below 500Hz, 600Hz, 325Hz, and 150Hz, respectively, with the lower woofer (red) and front port (blue) responses plotted below 325Hz and 200Hz, respectively.

Lower in frequency, the responses were all taken in the nearfield, with the relative levels calculated according to the ratio between each driver's effective diameter. As the two woofers are independently loaded by their own ports, I've plotted each separately. This gives a false impression regarding the lower crossover frequency, as the summed output of the woofers will be 6dB higher than either on its own. I'm not too bothered by apparent crossover discrepancies, however, because this graph is more useful in showing how each driver behaves both within and outside its passband. The upper woofer (black trace) has a slightly more restricted upper-frequency limit than the lower woofer (red), but its port is tuned lower, to 26Hz compared with 32Hz, judging by the two minimum-motion points. Similarly, the front port (blue) has a more extended passband than the one behind the Uni-Q sub-enclosure (black). The upper-midrange unit hands over smoothly to the lower-midrange below 400Hz, with the latter covering a limited two-octave range. Slopes appear close to 24dB/octave.

Fig.3 shows the 207's overall response averaged across a 30 degrees horizontal window on the Uni-Q axis, spliced to the complex sum of the low-frequency nearfield responses. (Both acoustic phase and the different distances of each driver to a nominal farfield point are taken into account.) The 3dB rise in the midbass level will be due to the nearfield measurement's assumption of a 2pi (hemispherical) environment. The balance is fairly even over the audioband, but salient features are the slight energy excess in the middle of the midrange, the peak at the top of the Uni-Q midrange unit's passband, and the slightly elevated tweeter level. (The HF adjust jumper was in place for this measurement, as that is how I preferred it for my auditioning—but see below.)

Fig.3 KEF Reference 207, anechoic response on Uni-Q axis at 50", averaged across 30 degrees horizontal window and corrected for microphone response, with the complex sum of the nearfield responses plotted below 300Hz.

The KEF 207's horizontal dispersion (fig.4) was narrower than usual above 800Hz but remarkably even, presumably because the Uni-Q drive-unit arrangement ensures that the tweeter and midrange radiation patterns are matched through the crossover region. The top-octave dispersion is also wider than I usually see, the supertweeter's smaller-than-usual radiating diameter contributing to this. The radiation uniformity below 10kHz can also be seen in the plot of vertical dispersion (fig.5), though the top-octave notches above and below the Uni-Q axis are due, I assume, to interference between the two widely spaced tweeters.

Fig.4 KEF Reference 207, lateral response family at 50", normalized to response on Uni-Q 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.

Fig.5 KEF Reference 207, vertical response family at 50", normalized to response on Uni-Q axis, from back to front: differences in response 15 degrees-5 degrees above axis, reference response, differences in response 5 degrees-10 degrees below axis.

Fig.6 shows how the speaker's quasi-anechoic behavior adds up in my listening room. (I average 120 measurements, with each speaker driven individually, across an area centered on my ear position in my listening chair.) The 1/3-octave balance is remarkably flat. Though the slight excesses in the middle of the midrange and the mid-treble that were revealed in the earlier graphs can be seen in fig.6, the overall variation from 100Hz to 10kHz lies within ±1.5dB limits. It's no wonder I found the KEF's tonal balance so neutral. Again, this graph was taken with the HF adjust shorting plug in place. However, removing the plug dropped the level of the octave between 8kHz and 16kHz by about 0.5dB. Although the difference is small, the region covered is wide enough for it to be audible.

Fig.6 KEF Reference 207, spatially averaged, 1/3-octave, freefield response in JA's listening room.

The speaker's step response (fig.7) indicates that all the drive-units are connected with positive acoustic polarity, but that, even with the stepped-back placement of the tweeters, the speaker is not overall time-coherent. This will be due to the use of high-order crossover filters. Some small HF reflections are apparent about half a millisecond after the main arrival. I have no idea what these are due to, unless they are reflections of the Uni-Q tweeter's output from the midrange cone termination. However, though quite clean overall, the 207's cumulative spectral-decay plot (fig.8) indicates that the 5kHz peak seen in figs. 2 and 3 is associated with a ridge of delayed energy typical of some sort of resonance. Perhaps this mode correlates with the occasional hardness I noticed on choral recordings, but given the nature of the graph, I'm surprised that I was not bothered more by excessive stridency or a "steely" coloration in my auditioning. Perhaps this measured peak was exaggerated by the low ambient temperature on the day I examined the speaker's technical performance.

Fig.7 KEF Reference 207, step response on Uni-Q axis at 50" (5ms time window, 30kHz bandwidth).

Fig.8 KEF Reference 207, cumulative spectral-decay plot at 50" (0.15ms risetime).

Overall, this is excellent measured performance, especially regarding the speaker's in-room delivery.—John Atkinson

10 Timber Lane
Marlboro, NJ 07746
(723) 683-2356