measurements from ja
The Dvorak's calculated sensitivity was a little higher than the specification, at 90dB/W/m (B-weighted). The main panel's electrical impedance (fig.1) showed the specified minimum of 3.1 ohms at 2.1kHz; though the magnitude is mainly much higher than this, note the high phase angle in the low treble, which means that the Dvorak will be quite a hard load for an amplifier to drive. In addition, the large variation in magnitude means that the speaker's frequency response will vary significantly if an amplifier with a high output impedance is used. In particular, the entire treble region will be shelved down -- a tube amplifier will definitely make the Dvorak sound sweeter, as long as it is not fazed by the awkward loading between 1kHz and 3kHz.
As there is no cabinet to speak of , the low-frequency peak in fig.1 is not due to the usual box resonance. Rather, it is due to the free-air resonance of the mid/woofers, which appears to lie at 31Hz. The little wrinkle in the traces at 26kHz is due to the metal-dome tweeter's "oil-can" resonance.
Fig.2 shows the dipole subwoofer's impedance magnitude and phase. The peak below 20Hz is due to the driver-units' free-air resonance; the rise in magnitude in the midrange is due to the drivers' voice-coil inductance. But note that there are two strong wrinkles in the traces, one at 1500Hz and one around 180Hz. These will be due to resonances of some kind. While the higher-frequency one is innocuous, the lower-frequency one is less than an octave above the crossover to the main panel -- risky business, unless the low-pass crossover filter is very steep-sloped.
The shaped output responses of the Dvorak's electronic crossover are shown in fig.3. The input level was 100mV and the subwoofer level control was set to its maximum position. The subwoofer drive signals, shown to the left of the graph, reveal a combination of cut about 100Hz and boost below to compensate for the dipole rolloff. The subwoofer drive is down by almost 20dB at the resonance frequency, which should result in minimal excitation of the problem. The lower subwoofer curve shows the effect of the "Video" switch: it cut the boost with eventually a 6dB/octave rate to avoid overloading the drive-units with subsonic explosions.
The main-panel drive signals are shown to the right of fig.3. The signal is increasingly boosted below 300Hz, again to compensate for the speaker's dipole rolloff. With the high-pass crossover switched in (bottom curve), the boost is reduced below 100Hz.
Electrically, the crossover performed well. Its input impedance (at 1kHz) measured just under 20k ohms, while its output impedance from the main outputs was 235 ohms (panel outputs at 1kHz) and 225 ohms (subwoofer outputs at 100Hz). Its insertion loss of 0.4dB will not be significant, and its distortion and noise were both low. Fig.4 shows its overload points: The top trace shows that the panel output at 1kHz clips at an output of 62.V, equivalent to an input voltage of just over 7V, which is well above the maximum output level of any combination of source components and preamplifier with which the Dvorak will be used. The bottom trace is subwoofer performance, assessed at 100Hz with the level control full up: The lower distortion level is due to the output's low-pass function, any additional harmonics being rolled-off; the output clipping point (1% THD+N) is 4V, equivalent to an input of 11V, this again well in the safety region. Though the low-frequency boosts applied to both sets of output signals will reduce the clipping margin, this should not be a practical problem.
As Siegfried Linkwitz points out in the accompanying interview, assessing the low-frequency performance of dipoles is not easy, due to the usual nearfield techniques failing to allow for the dipole cancellation. For interest's sake, however, fig.5 shows the nearfield output of the Dvorak subwoofer, measured without the crossover/equalizer (top trace) and with the crossover set to normal operation (middle trace) and to "Video" (bottom trace). Without EQ, the subwoofer's natural output extends to above 1kHz. Despite the impedance wrinkle at 180Hz, the highest acoustic output is obtained at 270Hz. Note, however, that this peak is knocked down by 40dB once the EQ is switched in! It should not have any audible effect. Note also that the overall LF boost appears to be very mild, at 3dB or less (the difference between the levels below the 20Hz region with and without EQ). With the "Video" button pressed in, the entire 30-50Hz region is reduced in level by about 5dB, reaching -12dB at 20Hz, significantly reducing the excursion demands on the drive-units.
The individual responses of the main panel's drive-units are shown in fig.6. (Again, note that the woofer's nearfield traces are not representative in that they don't allow for the dipole cancellation.) The acoustic crossover is to spec at 2kHz, with steep, approximately 24dB/octave slopes. The passbands of the mid/woofer and tweeter look impressively flat. However, there does appear to be a little too much overlap between the drive-units to give a perfectly flat summed response through the crossover region. In my experience, the steeper the crossover slopes, the closer the tolerance necessary of the parts used if the actual crossover is not to depart from the target performance.
Note the deep, narrow notch at 24kHz. This is at its most extreme exactly on-axis and is due, I imagine, to the "phase plate" that covers the tweeter dome. The middle trace on the left of fig.6 shows the nearfield response of one of the mid/woofers, while the top trace is the same woofer's output modified by the electronic crossover. Remember, however, that these measurements do not show the rolloff due to the dipole cancellation.
The Dvorak's main-panel overall response, measured on its tweeter axis at a distance of 50" and averaged across a 30o horizontal window, is shown in fig.7. The top audio octave appears to be rolled-off in this graph, partly due to the tweeter being quite directional in this region. While the midrange and treble regions are otherwise impressively flat, the 1.5-3kHz crossover region is plateau'd up by 2dB. Whether or not this is due to the drive-unit overlap in this region, I could hear it in my own auditioning as a narrow band of brightness. Note also that Shannon did find the Dvorak to sound a little on the analytical side, which is not unexpected given this kind of on-axis balance (though the speaker's off-axis behavior is also relevant here).
On the left of fig.7 is the same equalized nearfield woofer response as shown in fig.6, plus the same measurement made with the high-pass crossover filter switched-in. To get an idea of the Dvorak panel's true bass extension, I used noise with a 1kHz bandwidth and measured the speaker's in-room response at the same 50" distance used to obtain the traces to the right of figs.6 and 7. Plotted on a half-octave-smoothed basis to eliminate some of the major room effects, this is shown as the bottom trace in the graph. The dipole cancellation does appear to cancel almost all the nearfield bass boost. However, there is still a slight energy excess apparent in the lower midrange. I assume that as SD didn't remark on any coloration in this region, its effect is benign, particularly as it falls in a spectral region where room effects dominate the perceived balance.
Vertically, while the tweeter's 41.5" distance from the floor is a little elevated for typical seated ear heights, the Dvorak's balance remained remarkable even across a wide range of listening heights. As long as the listener sits with his or her ears between the bottom of the bottom mid/woofer (35" from the floor) and the top of the top mid/woofer (51"), there should be no significant changes in balance. Only a standing listener close to the panel will hear a response with a lack of energy in the crossover region.
Horizontally (fig.8), the Dvorak's output falls relatively evenly with increasing off-axis angle, as expected. In the bottom octave of the tweeter's bandpass, where the dispersion is at its widest, the output of 90o has fallen significantly more than at 45o off-axis. However, the null at the speaker sides is less deep than I had anticipated, something that could be heard as well as measured. The front and rear waves do not see sufficiently the same acoustic environments to cancel completely at 90o. According to Siegfried Linkwitz, they cancel at around 110o, and the overall total power output follows the classic dipole characteristic.
In the time domain, the Dvorak panel's step response (fig.9) indicates that the tweeter and mid/woofers are connected with opposite acoustic polarity. (The tweeter's output is the small, initially negative-going spike of energy at 3.75ms; the woofers' output is the larger, positive-going triangle of energy immediately after the 4ms mark. As Linkwitz states in his interview, whether such a lack of time coherence is significant or not is not known. However, my experience is that if everything else in a speaker design is right, time coherence adds that last, essential element of image focus.)
Finally, the Dvorak's cumulative spectral-decay plot (fig.10) indicates an initially clean decay, but with some low-level hash present in the mid-treble. This may be resonant behavior; alternatively, and in my view more probably, this is due to low-level early reflections of the sound from the speaker's structure. Note that the ultrasonic tweeter resonance is down in level in fig.10, which was taken on-axis. The resonance is more noticeable -- at least to a microphone -- off-axis.
--John Atkinsonsummary thoughts
After a year of careful listening, my take on the Audio Artistry Dvorak is that it successfully incorporates many of the positive qualities of large panel dipoles and moving-coil designs while minimizing most of their respective drawbacks. In contrast to so many speakers that are simply variations on common themes, the Audio Artistry Dvorak offers a refreshing perspective. With its marked lack of room, cabinet, and air-cavity colorations, its smooth and consistent power response, its natural top-to-bottom tonal balance, and its unforced yet dynamic expressiveness, I felt that I was hearing all that this combination of components could offer. Its wonderful balance of attributes answers the question posed at the beginning of this review -- "Does this speaker really possess a unique and compelling communicative skill?" -- with a resounding "Yes!" The Dvorak is a genuine music-lover's speaker of the first order.
While the Dvorak scores highly on the typical audiophile sonic checklist, its ability to tie the whole experience together and directly convey the emotional undercurrent of good music made conducting such an analytical assessment seem almost frivolous. Yes, a very few of the many outstanding speakers I have heard at length have a slight performance edge in a few specific areas, but I wouldn't trade the Dvoraks for any of them for sheer musical enjoyment or long-term reference in a normal-sized room. This includes a number of elite models costing nearly three times the Dvorak's $5995 price! Given unlimited resources and a cavernous room -- well, since I live in an apartment, that's a moot point.
Those in the market for a full-range loudspeaker for use in rooms under 600ft2 in area should make The Audio Artistry Dvorak a top priority on their audition lists, regardless of budget. Those planning on spending much more may be pleasantly surprised, and those with tighter budgets are likely to find themselves justifying "creative financing" alternatives in order to take these babies home. Happy listening!
--Shannon Dickson
Stereophile
-- Vol. 19 No. 4