Vienna Acoustics Mozart loudspeaker Measurements
All the measurements were performed without any foam in the ports. Given the Mozart's high specified sensitivity of 90dB/W/m, I was puzzled to arrive at a low calculated figure of just over 83dB at 1m for 2.83V drive. However, the B-weighted figure we use at Stereophile will be affected by frequency-response anomalies in the upper midrange/low treble. If, for example, there was a big suckout in a speaker's crossover region, this would adversely affect the calculated sensitivity. We shall see. The speaker's impedance (fig.1) is moderately demanding in that it dips to below 4 ohms between 100Hz and 400Hz, the lower-midrange region where music has a lot of energy. Be sure to drive the Mozart with a good solid-state amplifier, or a tube design with a beefy 4-ohm output tap.
Fig.1 Vienna Acoustics Mozart, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
The saddle in the magnitude trace at 43Hz indicates the tuning of the twin ports, implying reasonable LF extension. The impedance traces are free from resonance-caused wrinkles, implying a solid, optimally braced enclosure design. Looking at the cabinet panels' vibrational behavior with a simple PVDF-tape accelerometer revealed that this was indeed the case. Even the juiciest mode I could find—on the cabinet sidewall about 10" from the top (fig.2)—was very high in frequency, meaning that it would be unlikely to add coloration.
Fig.2 Vienna Acoustics Mozart, cumulative spectral-decay plot of accelerometer output fastened to side of enclosure near the top. (MLS driving voltage to speaker, 7.55V; measurement bandwidth, 2kHz.)
Fig.3 shows, from right to left, the individual responses of the tweeter, the two woofers (which are identical), and the complex sum of the two port outputs. The latter are very similar below 200Hz, and present a classic reflex bandpass response centered on the overall tuning frequency of 43Hz—which is, not coincidentally, the frequency at which the woofer cone is prevented from moving. But note the two high-level peaks in the port outputs, at 600Hz from one port and 750Hz from the other. Presumably due to some kind of pipe resonance, these could be heard when I was standing behind the speaker. From in front of the speaker, however, their effect was effectively masked.
Fig.3 Vienna Acoustics Mozart, acoustic crossover on tweeter axis at 50", corrected for microphone response, with nearfield woofer and port responses plotted below 300Hz and 800Hz, respectively.
The woofer can be seen from fig.3 to be reasonably flat in its passband, with higher-frequency cone modes well-suppressed by the crossover. The tweeter, however, seems to come in a little too high in frequency, at least regarding its on-axis response.
So how do these individual drive-unit outputs add up? The answer—not that well—is shown in fig.4. The ports and woofers sum to give a maximally flat response down to a musically useful 38Hz, –6dB—though, as RD found, this alignment is sensitive to room problems in the port output region. More important, the woofer and tweeter outputs effectively cancel in the region where they overlap, giving rise to an 18dB-deep notch at the crossover frequency of 2.8kHz. This explains the low measured sensitivity. To make sure that I was not simply seeing a fault with the first sample (serial number 4792), I measured the second sample (4326) under identical circumstances. Both measurements can be seen in fig.5. The speakers are superbly well-matched—but both feature the crossover suckout!
Fig.4 Vienna Acoustics Mozart, anechoic response on tweeter axis at 50", averaged across 30 degrees horizontal window and corrected for microphone response, with the complex sum of the nearfield woofer and port responses plotted below 300Hz.
Fig.5 Vienna Acoustics Mozart, anechoic response on tweeter axis at 50", corrected for microphone response, of samples 4792 (top) and 4326 (bottom, offset by 5dB for clarity).
The obvious culprit to blame when you see an on-axis response like this is a miswired drive-unit. Accordingly, I flipped the electrical polarity of the tweeter and remeasured. The result is shown in fig.6: The trace with the notch is with the nominally "correct" phasing; the top trace is with the tweeter wired in inverted electrical polarity. You can see that now the suckout fills in, other than for a smaller notch in the bottom octave of the tweeter's passband (this expected from the individual response plots of fig.3). If it wasn't for the fact that the second sample was identical, I would have suspected a construction fault.
Fig.6 Arcam Alpha 10, phono input, THD+noise vs frequency (from top to bottom at 1kHz): MC, MM (right channel dashed).
Another way of thinking about this phenomenon is to remember that if the drive-units are electrically connected to produce a crossover notch on the tweeter axis, there will be an axis somewhere where the outputs do add in-phase. Fig.7 shows the Mozart's responses on-axis ranging from 15 degrees below the tweeter to 15 degrees above. It should be obvious from this graph that the flattest output in the crossover region is obtained on the lowest axis.
Fig.7 Vienna Acoustics Mozart, vertical response family at 50", from back to front: differences in response 15 degrees-5 degrees above axis; on-axis response; differences in response 5 degrees-15 degrees below axis.
Remember that RD reported that Sumiko's John Hunter tilted the speakers back as far as possible. Here is the reason: Other than inverting the tweeter's electrical polarity, that's the only way to get a reasonably neutral mid-treble balance from the Mozart. The result is shown in fig.8. Note how similar it is to the top trace in fig.6. There is still a small lack of on-axis energy in the bottom of the HF unit's region, but the speaker's horizontal dispersion (not shown) reveals that there is some horizontal "flare" in this region which will—to an extent governed by room size and furnishings—compensate.
Fig.8 Vienna Acoustics Mozart, anechoic response on optimum axis at 50", corrected for microphone response, with complex sum of the nearfield woofer and port responses plotted below 300Hz.
The mismatch between the tweeter and woofer polarities on the HF unit's axis can also be seen in the step response (fig.9). Usually, where the crossover's electrical phase response mandates inverting the tweeter polarity, the return of the tweeter's step output to the time axis coincides with the positive-going but slower start of the woofer's step. Here you can see that that is not the case.
Fig.9 Vienna Acoustics Mozart, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Finally, fig.10 shows the Mozart's cumulative spectral-decay or waterfall plot, calculated on the optimal axis used to derive fig.8. It is commendably clean and goes a long way toward explaining the Mozart's good sound quality once its intrinsic on-axis problems had been compensated for.
Fig.10 Vienna Acoustics Mozart, cumulative spectral-decay plot on optimum axis at 50" (0.15ms risetime).
So, Bob, you're not deaf! But it's depressing how heroic setup procedures had to be used to compensate for what appears to be a suboptimal implementation of the Mozart's crossover. The Mozarts appear to do many things well; but if they were my speakers, I'd use the bi-wiring terminals to invert the tweeter's electrical polarity. In my own auditioning, this really made the Mozarts sing.—John Atkinson