Bowers & Wilkins CM5 loudspeaker Measurements
Sidebar 3: Measurements
I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the Bowers & Wilkins CM5's frequency response in the farfield, and an Earthworks QTC-40 for the nearfield and spatially averaged room responses.
My estimate of the CM5's voltage sensitivity was 87.5dB(B)/2.83V/m, which is within experimental error of the specified 88dB. The CM5 is specified as having a nominal impedance of 8 ohms with a minimum value of 3.7 ohms; fig.1 confirms the specification, and reveals that the minimum impedance occurs at the top of the audioband, where there will be little musical energy. Though the electrical phase angle varies widely, the impedance magnitude tends to be high at the most extreme phase angles, ameliorating the negative effect of the phase angle. The saddle centered at 48Hz in the impedance magnitude (solid trace) suggests that this is the tuning frequency of the flared port on the rear of the enclosure. Plugging this with the supplied foam cylinder resulted in an impedance plot typical of a sealed enclosure (fig.2), with a tuning frequency of 76Hz. When the CM5 must be placed close to the wall behind it, it will work best with its port plugged.
There is a small wrinkle at 30kHz in both impedance traces; this will be due to the aluminum-dome tweeter's fundamental diaphragm resonance. Another discontinuity is visible between 900Hz and 1kHz, behavior I expected to correlate with an enclosure resonance of some kind. However, while investigating the cabinet's panels with a simple plastic-tape accelerometer did uncover a strong vibrational resonance at 414Hz that could be detected on all surfaces (fig.3), there was nothing significant higher in frequency.
Looking at the nearfield response of the port (fig.4, red trace), a peak can be seen just below 1kHz, but this is well down in level. Other than that, the port is well behaved, with a textbook bandpass response centered on 48Hz, the frequency of the corresponding minimum-motion notch in the woofer's output (green trace). The woofer's higher-frequency response is fairly flat before crossing over to the tweeter (blue trace) at the specified 4kHz. However, it appears from fig.4 that the CM5 uses a slow-slope crossover, perhaps 6dB/octave. I have extended the horizontal scale of this graph to 40kHz so you can see the effect on the response of the tweeter's "oil-can" resonance. There is a 20dB-high peak at 30kHz. Fortunately, this resonance lies 57kHz higher in frequency than is usual for an aluminum-dome tweeter, and it won't be excited with CD playback. In playing back LPs with a moving-coil cartridge, however, the resonance will be set in motion by ticks. It will also be excited by high-resolution digital playback, with an unpredictable effect on sound quality within the audioband.
Fig.5 shows how these individual responses sum in the farfield on the tweeter axis, averaged across a 30° horizontal window. There is only a trace of the usual nearfield upper-bass response hump, suggesting that the CM5 has a somewhat overdamped low-frequency tuning. While BJR did comment on "a slight thickening in the upper bass that called attention to that range," I suspect that he was actually responding to that strong cabinet resonance at 414Hz rather than to a problem with the woofer. Higher in frequency, there is a slight lack of energy just below the crossover point; all things being equal, this would lend the B&W a rather laid-back quality. However, it would also make the speaker kind to overcooked rock recordings.
The CM5's horizontal dispersion on the tweeter axis (fig.6) reveals a well-controlled radiation pattern; the apparent off-axis flare shown by the cursor position at 6.4kHz is actually due to the small suckout centered at this frequency in the on-axis response filling in to the speaker's sides. In the vertical plane (fig.7), the use of a low-order crossover with a lot of overlap between the drive-units leads to significant modifications as the listener moves above and below the tweeter axis. However, this graph suggests that the flattest treble balance will be heard when the listener's ears are slightly above the tweeter, which in turn suggests that low stands will work better than high ones.
Fig.8 shows the spatially averaged, 1/6-octave response in JA's listening room. The CM5 excites the lowest-frequency room mode, resulting in good low-frequency extension. The B&W is a little laid-back in the low treble but the extra energy above 5kHz produced in-room by the CM5 lent it a light balance overall. The B&W's high-amplitude tweeter resonance just below 30kHz makes its presence known in this graph, but this is well above what anyone can hear.
In the time domain, the step response (fig.9) indicates that both drive-units are connected with positive acoustic polarity and that, as is usual with a flat-baffle design, the tweeter output leads that of the woofer. The cumulative spectral-decay plot (fig.10) has its floor suppressed by the 20dB height of that ultrasonic tweeter resonance. However, it still reveals a superbly clean decay at almost all frequencies, correlating with BJR's feeling that the CM5 had a commendable purity to its sonic character.
It is difficult for a conventional speaker with a flat baffle to use a low-order crossover without there being compromises in dispersion and response. However, the measured performance of Bowers & Wilkins' CM5 shows little sign of such compromises. I am not surprised BJR liked this speaker as much as he did.John Atkinson