Hales System Two loudspeaker Measurements
Sidebar 3: Measurements
The Listening Environment Diagnostic Recording (LEDR) on the Chesky Test CD produced an impressive feeling of the image moving above the loudspeaker. The "over" test was similarly good, with a continuous "rainbow"-shaped arc traced between the loudspeakers. Driving the System Two with a sinewave oscillator and feeling the cabinet for resonances revealed the enclosure's rigidity. Very minor resonances were felt and heard at 65Hz, 200Hz, with the strongest at 500Hz. The System Two was noticeably less inert than the Two Signature, but far more solid than most loudspeakers.
The System Two has an impedance magnitude and phase curve (fig.1) nearly identical to the Signature's. Throughout most of the low-frequency region, the impedance drops to 3 ohms, averages about 5 ohms through the midrange, and reaches 6.5 ohms at 20kHz. This is a challenging load for an amplifier; I recommend one with the ability to drive current into low impedances. It's no coincidence that the Krell KSA-250 and Threshold S/550e, both rock-solid powerhouses, coax the best bass performance from the System Two.
Fig.1 Hales System Two, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
The impulse response, measured 48" away on the tweeter axis, is shown in fig.2. Note the metal-dome tweeter's ultrasonic ringing. Unusually, several tiny reflections, shown as bumps and dips in the straight line following the tweeter ringing, are visible between the 4.5 and 5.5ms markers. (The large reflection at 7.5ms is that from the listening room ceiling.) This is indicative of very early reflections, perhaps from the 12 hex bolts that secure the woofers to the baffle. After performing a Fast Fourier Transform on this time-domain plot to get a frequency-domain plot, we will see the effects of these reflections. The step response is shown in fig.3.
Fig.2 Hales System Two, impulse response on tweeter axis at 48" (5ms time window, 30kHz bandwidth).
Fig.3 Hales System Two, step response on tweeter axis at 48" (5ms time window, 30kHz bandwidth).
With the System Two on the floor without stands, the listening axis is much higher than the tweeter axis. In fact, my ears, at 36" above the floor, were a full 9" above the tweeter axis. I noted during the auditioning that the sound without stands lacked immediacy and was somewhat threadbare. This impression is confirmed by fig.4, the FFT-derived frequency response measured on the upper woofer axismy listening axis without the stands. There is a slight but very broad dip in the amplitude between 1kHz and 4kHzexactly where the ear is most sensitive to amplitude changes. This dip will certainly be audible, both because of the ear's sensitivity and the great amount of musical information that lies in this important band.
Fig.4 Hales System Two, anechoic response on upper-woofer axis at 48".
Measured on the tweeter axis (the listening axis with the stands), the frequency-response curve looks quite different. This is seen on the right-hand side of fig.5, the FFT-derived frequency response spatially averaged over a 30° lateral window. The low-treble dip is gone and the overall response is much flatter. When I first auditioned the System Twos without stands, I thought they sounded noticeably inferior to the Signatures. The stands, however, brought the System Twos' presentation much closer to that of their big brothers.
Fig.5 Hales System Two, anechoic response on listening axis at 48", averaged across a 30° lateral window, with nearfield woofer response plotted below 300Hz.
Remember those very early reflections seen as tiny bumps in the impulse response? Looking at their effect in the frequency domain, it appears as though they are creating some minor comb-filtering in the treble. These are not severe dips, but tiny ripples caused by constructive and destructive interference between the direct sound and the slightly delayed sound reflected from the woofer mounting bolts. I doubt these ripples are audible because of their very low amplitude: the comb filtering generated by the listening room's sidewalls would be far more severe (although at a much lower frequency).
Looking at the System Two's individual response curves directly on the horizontal axis, at 7.5° and 15° off axis (15° plot shown in fig.6), the slight excess of treble energy diminishes greatly even a few degrees off-axis. In other words, the loudspeaker is flatter in the treble when it isn't pointed directly at the listener. This was suggested during the auditioning when I noted that the HF balance was smoother with the System Twos toed-in slightly so they pointed past the listener.
Fig.6 Hales System Two, anechoic response 15° to the side of the tweeter axis at 48".
Another correlation between listening impressions and measurements is the top-octave rolloff seen in fig.6, the response 15° off-axis. I observed that the System Two didn't seem to have a feeling of infinite HF extension. Although I detected no differences in instrumental timbres, I felt less sense of "air riding over the presentation"exactly what one would expect with an extreme HF rolloff off-axis caused by the tweeter's increasingly narrow dispersion as frequency rises. In addition, reduced high-frequency reflections from the listening room's sidewalls will also contribute to the lessened feeling of air and extension. This is why the Type C/IV incorporates a rear-firing tweeter, compensating for the front tweeter's narrower dispersion at high frequencies.
To the left of fig.5 is shown a nearfield woofer measurement. A microphone is placed directly in front of the woofer (nearly touching the dustcap), and a swept signal from the Audio Precision System One drives the loudspeaker via a power amplifier. The rolloff slope below 50Hz, typical of sealed-enclosure design, correlates with my impression of the System Two's lack of LF extension.
Finally, the System Two's cumulative spectral-decay or "waterfall" plot is shown in fig.7. This is the anechoic frequency response at discrete time slices, revealing the loudspeaker's decay characteristics. The decay is very clean, but with some hash around 5kHz, the same frequency as the peak in the initial response (shown by the cursor position).Robert Harley
Fig.7 Hales System Two, cumulative spectral-decay plot on listening axis at 48" (0.15ms risetime).