Krell Resolution 1 loudspeaker Measurements
The size and weight of the Resolution 1 worked against my being able to raise it high into the air for the acoustic measurements. I therefore had to window the time-domain data more aggressively than usual to eliminate the interfering effects of a reflection from the floor between the speaker and the microphone. As a result, the measured response will have less midrange resolution than usual, though this does not interfere with the reliability of the measurements.
The big Krell's voltage sensitivity came in slightly below the specified 90dB/W/m, at 88.7dB/2.83V/m. This will be due in part to the speaker's subdued high treble (see later). This is still usefully higher than average, but working against this is a truly brutal impedance character. As can be seen in fig.1, the impedance remains at or below 3 ohms for two-thirds of the audioband, relieved only by the reflex peaks in the low bass and a region between 4 and 6 ohms in the upper midrange/low treble.
Compounding the drive difficulty of this low impedance is an electrical phase angle that varies considerably. So not only is the minimum impedance a current-hungry 1.8 ohms at 58Hz, this is combined with a phase angle of –25 degrees. And a little lower in frequency, even though the impedance has risen to a more manageable 3.8 ohms at 43Hz, the phase angle is now –61 degrees! This behavior should not present any problems to Krell's own amplifiers, which have traditionally offered high current delivery, but some tube amplifiers are going to be gasping for breath when hit with high levels of bass information.
The traces in fig.1 are free from the glitches that would indicate the presence of panel resonances. I could find only a few low-level modes when I investigated the cabinet's vibrational behavior with a simple plastic-tape accelerometer. Fig.2, for example, is a cumulative spectral-decay plot calculated from the accelerometer's output when it was fastened to the front baffle between the tweeter and the woofers. Only the mode at 762Hz rises to the point where it might have subjective consequences, but this is high enough in frequency that the delayed energy drops rapidly, working against its audibility. Despite the large panel areas, the enclosure's sides are well-braced and -damped.
Fig.1 Krell Resolution 1, electrical impedance (solid) and phase (dashed). (2 ohms/vertical div.)
Fig.2 Krell Resolution 1, cumulative spectral-decay plot calculated from the output of an accelerometer fastened to the cabinet's front baffle between the tweeter and the woofers (MLS driving voltage to speaker, 7.55V; measurement bandwidth, 2kHz).
The saddle at 21.5Hz in the impedance magnitude trace suggests that this is the tuning frequency of the twin ports, which in turn implies excellent bass extension. The output of the ports (fig.3, black trace) peaks at around 20Hz, though it doesn't start rolling off until 50Hz. The output of the two woofers (red trace) has its minimum-motion point—ie, where the back pressure from the port resonance holds the cones still—at 25Hz, a little higher than suggested by the impedance measurement. The two woofers peak between 40 and 120Hz, handing over above that frequency to the lower-midrange driver (green), which in turn hands over to the midrange unit (blue) at around 450Hz. The acoustic crossover slopes all appear to be second-order.
Fig.3 Krell Resolution 1, nearfield responses of the midrange unit (blue), upper woofer (green), lower woofers (red), and ports (black), with the complex sum of the nearfield responses (magenta), weighted in the ratio of the square roots of the radiating areas.
The magenta trace in this graph shows the complex sum of these individual outputs, weighted according to the square root of the radiating areas. It is pretty flat, other than a rise in the region covered by the twin woofers. However, to a large extent this rise will be due to the nearfield measurement technique; the Krell's reflex alignment is usefully overdamped, which will give the best combination of extension and control in a typical room.
This summed LF output is repeated on the left-hand side of fig.4, spliced at 350Hz to the farfield response, which is averaged across a 30 degrees horizontal window centered on the 41"-high midrange axis (the listening axis Krell recommends). The midbass boost is evident, but as this is most probably a measurement artifact, I must assume that Mikey's thinking the Resolution 1 sounded a bit "rich" in the upper bass is due more to the woofers' restricted passband. (The more you limit a drive-unit in the frequency domain, the less well defined its output will be in the time domain.)
Fig.4 Krell Resolution 1, anechoic response with grille on the midrange 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, taking into account acoustic phase and distance from the nominal farfield point, plotted below 350Hz.
The Resolution 1 was extraordinarily flat in the midrange, but a gently sloped-down trend is apparent throughout the treble—noted by MF in his auditioning, which was performed with the grilles in place—broken by a series of peaks and dips. As shown in fig.5, both the subdued treble balance and the peaks/dips are due to the grille, which offers significant acoustic interference. (Although the grille superficially resembles the vertical strings used by Sonus Faber, the latter are thinner and are more widely spaced than Krell's.)
Fig.5 Krell Resolution 1, effect of the grille on the farfield response on the midrange axis (5dB/vertical div.).
The Resolution 1's lateral dispersion on the midrange axis (fig.6) is even, with no off-axis "hot spot" frequencies. Although it is hard to see from this graph, the off-axis responses are actually flatter than the on-axis response, the treble dips filling in and the peaks flattening out, meaning that the speaker's room sound will be smoothly balanced. However, the dispersion above 10kHz is quite limited, which will rob the speaker's balance of top-octave air. In the vertical plane (fig.7), the Krell's balance doesn't change significantly as long as the listener sits with his ears between the lower-midrange unit and the top of the lower woofers. However, a standing listener will perceive a suckout at the upper crossover frequency of 3.1kHz, as well as top octaves that are even lower in level than on the midrange axis.
Fig.6 Krell Resolution 1, lateral response family at 50", normalized to response on midrange 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.7 Krell Resolution 1, vertical response family at 50", normalized to response on midrange 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.8 shows the Resolution 1's step response on the midrange axis. The short, positive-going spike is the tweeter output, followed by the lazier rise, again positive-going, of the midrange unit. This leads into the negative-going output of the lower-midrange unit, which in turn is followed by the positive-going output of the woofers. Some reflections can be seen in this graph: the one just before 8ms is the reflection from the floor that resulted from my inability to hoist the Krell off the ground for the measurements; those marring the midrange unit's step are from the grille. This is shown graphically in fig.9, which overlays the Resolution 1's impulse response with and without the grille (red and blue traces, respectively). The grille reflections can easily be seen.
Fig.8 Krell Resolution 1, step response on midrange axis at 50" (5ms time window, 30kHz bandwidth).
Fig.9 Krell Resolution 1, impulse responses on midrange axis at 50", with (red) and without (blue) grille (5ms time window, 30kHz bandwidth).
As I mentioned earlier, these reflections don't affect the speaker's perceived balance, other than to subdue the tweeter's output in the top two octaves. However, they do mess up what would otherwise be a clean cumulative spectral-decay plot (fig.10).
Fig.10 Krell Resolution 1, cumulative spectral-decay plot at 50" (0.15ms risetime).
Extended low bass, a warm midbass, a smooth midrange, a subdued top end: Michael Fremer called it correctly, according to my measurements of the Krell Resolution 1.—John Atkinson