Unity Audio Signature 3 Measurements
Other than impedance, all acoustic measurements were made with the DRA Labs MLSSA system and a calibrated B&K/DPA 4006 microphone. To minimize reflections from the test setup, the measuring microphone is flush-mounted inside the end of a long tube. Reflections of the speaker's sound from the mike-stand hardware will be sufficiently delayed not to affect the measurement.
The Unity Audio's sensitivity was about average at a calculated 86dB/2.83V/m. As stated by Kathy Grost, the Unity Signature 3 does have a flat impedance with frequency, as can be seen from its impedance plot (fig.1). This was for one of the second samples; the first sample was fundamentally similar. The phase angle is generally moderate, but with a magnitude at all times below 6 ohms and averaging under 4 ohms through the midrange, the Siggie 3 is best left to beefy solid-state amps and respectable tube amps with a goodly amount of current available from their 4 ohm transformer taps. The impedance peak at 80Hz indicates the sealed-box tuning frequencyas MK found, this is basically a floorstanding minimonitor! Note the small wrinkle in both magnitude and phase traces at 210Hzsome kind of strong resonance is present at this frequency.
I wanted to investigate the Signature 3's bass performance in some detail, as the company's literature is confusingly worded: "Sound...is an energy mass," it states, continuing that "extensive physics research isolated a way to use two identical drivers, operating in parallel, each using the rear energy of the other to cancel its own rear, out-of-phase 'sound'...this works independent of internal volume." (Unity's italics.) Unity terms this concept "Inverse Force Vector Coupling"; they go on to claim that this "revolutionary phase-locked driver loading...produces detail, efficiency, and dynamic bass punch previously unheard of in compact loudspeakers." All well and good, but fig.1 and MK's auditioning indicate that the Signature 3 behaves as a traditional, small sealed-box design without much low-frequency extension.
The lower trace in fig.2 shows the nearfield response of one of the woofers (the other is virtually identical). It rolls off with a classic 12dB/octave slope below 100Hz: sealed-box behavior. Above that frequency, there is a notch in its output at 200Hzthe frequency of the wrinkle in the impedance plotbut with then a flat response through the rest of its passband. Contrary to what might be expected from the passage quoted above, both woofers are wired in-phaseie, both cones move outward from the cabinet when the speaker is fed a positive electrical pulse. Unity's designer seems to have believed that because the two woofers are mounted on the opposite faces of the cabinet, this means their outputs are in opposite polarity, internally as well as externally. This is simply wrong: the drive-units may be mounted facing in opposite directions, but if they are connected with the same electrical polarityas they aretheir acoustic outputs are also in-phase. The two woofers are mounted in a small sealed box, and that's exactly how they behave, electrically and acoustically.
The only modification of the classic sealed-box response is the notch at 200Hz. I suspect that this is basically the result of mounting two woofers at the ends of a pipe of rectangular cross-section. As the length of that pipe appears to be exactly a half-wavelength at 200Hz, the rear wave from one woofer arrives at the other 180° out-of-phase at that frequency, canceling its output. As is shown in fig.2.
And look what happens when the nearfield outputs of the two same-polarity woofers are summed, taking into account the phase response of each and the path-length difference calculated at a listening distance of 5'. The resultant response is shown as the upper trace in fig.2. Both woofers have an internal interference notch at 200Hz, hence their summed output also does. Below that frequency, their outputs sum in-phase to give an approximate boost in level of 6dB compared with the single woofer. Their outputs also sum in-phase above 450Hz or so, again resulting in a 6dB level boost. But in between these two frequency regions, the path-length difference results in a cancellation between the two woofer outputs. The result is a large suckout centered on 325Hz, a frequency right in the middle of the midrange.
Note that this is a theoretical calculation. It's possible that in a real room, the two woofer outputs will not cancel in so neat an manner. Nevertheless, in my own auditioning of the Signature 3s, both in Muse Kastanovich's room in Boulder, CO and in the Stereophile listening room in Santa Fe, I was bothered by a midrange hollowness to the speaker's tonal balance. And though MK liked the speaker overall, he did comment on some midrange coloration.
So how did the Signature 3 measure in the far field? Fig.3 shows the quasi-anechoic response of the first sample with the microphone on its tweeter axis at a distance of 50". Though this is a low 33.5" from the ground, the slight tiltback due to the base plate will make it a typical listening axis. On the left of this graph is the complex sum of the two nearfield woofer outputs; this is spliced at 300Hz to the far-field response averaged across a 30° horizontal window. Points to note are: the notch at 200Hz due to the internal interference between the two woofers; the expected lack of energy in the lower midrange, due to the external interference between the two woofers (which appears, therefore, to be real, not just theoretical); a lack of energy in the lower treble due to poor integration between the woofers and tweeter, at least on this axis; and a tweeter level that averages around 6dB too high. (MK did find the speaker to have plenty of top-octave air.) The second sample, measured under identical conditions (fig.4), was considerably flatter on the tweeter axis, though there was still an excess of top-octave energy.
So how come MK was not bothered by the speaker's measured problems? I suspect that the answer to that question lies in the Signature 3's complicated interaction with the room. Fig.5 shows the speaker's response measured 180° behind. The rear woofer's output, broken by the expected interference dips, actually extends all the way into the low treble. When the speaker is positioned relatively close to the wall behind it, it is possible that the delayed, reflected energy from the rear woofer reinforces the overall sound in a beneficial manner.
In addition, some of the on-axis dips fill in to the speaker's sides, as shown in fig.6, which will mean that the room reverberant field will be more evenly balanced than the Signature's direct sound. Remember also that MK chose to sit beneath the tweeter axis. As shown by fig.7, this also will result in better integration between the woofers and tweeter and a more even mid-treblethough if you sit too low, a big suckout appears in the crossover region.
In their literature, Unity Audio makes much of their crossover. They term this a "Balanced Class-A Design," due to both positive and negative legs featuring identical circuitry compared with a conventional crossover. The latter would place the network in series with just one of the drive-unit terminals, the other being a direct ground return. Their crossover topology, they claim, "eliminates the phase error at the drivers, resulting in huge advances in our loudspeaker's total phase characteristics." (Their italics.)
So how does the Signature perform in the time domain? Fig.8 shows its step response on the tweeter axis. The design is not time-coherent on this axis, the sharp spike of tweeter energy leading the slower spike of woofer energy by about 0.2ms on this axis. Note also the uneven decay of the step shape, this due, I imagine, to the delayed arrival of the sound from the rear woofer. The associated graph of excess phasethe difference between the speaker's actual phase response and what it would have were it a true minimum-phase designis shown in fig.9. Without a first-order crossover and either a staggered front baffle or time delay in the tweeter feedsee fig.7 in the Thiel CS1.5 review in this issue for the kind of phase coherence that can be achieved by this strategythe Signature 3's large negative phase error above 1kHz is exactly what you would expect from a conventional nontime-coherent design.
Another way to move the tweeter's output back in time is to tilt the speaker back or to sit lower, as MK indeed did. Fig.10 shows the Unity's step response approximately 15° below the tweeter axis. The tweeter output now does align in time with the front woofer's, but the price to pay for this coherency, even with the tiltback provided by the Unity's bass plate, is a listening height that will be impractical unless you sit a long way away from the speaker. And the amplitude response will not be appreciably different from that on the tweeter axis, as revealed by the lowest trace in fig.7.
Finally, fig.11 is the Siggie 3's waterfall plot. The initial decay is clean, but a resonant ridge appears in the tweeter's output just above 10kHz, while the midrange decay is enigmatic.
Overall, these measurements suggest that while it might be possible to get a tonally balanced presentation from the Unity Audio Signature 3s with careful setup and choice of listening distance and height, the speaker is not inherently neutral. And consistent factors in its presentation will be an exaggerated top octave and a restricted LF, which will make a recommendation very much a matter of taste.John Atkinson