Harbeth HL-P3 loudspeaker Measurements
Fig.1 shows the Harbeth's impedance magnitude and phase. Apart from dropping to 4 ohms in the upper bass, the value stays reasonably high, implying not too demanding an amplifier load. The sealed box is tuned to 68Hz, a little lower than the LS3/5a; the speaker is fractionally less sensitive. Note the wrinkles in the amplitude trace at 340Hz and 33kHz: the former indicates a cabinet resonance at that frequency; the latter is the tweeter's primary resonance, well up out of harm's way.
Fig.1 Harbeth HL-P3, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
The individual responses of the two drive-units are shown in fig.2. The woofer's output gently peaks between 100Hz and 200Hz in pretty much the same fashion as that of the LS3/5a, rolling out at a gentle 12dB/octave below that region, reaching –6dB at 56Hz compared with the LS3/5a's 60Hz. Flat in the rest of its passband, the woofer rolls off steeply above 3kHz with an acoustic slope of 24dB/octave. The tweeter also comes in with a steep acoustic slope, and is basically flat on-axis. The overall response on the tweeter axis (fig.3) is superbly flat through the midrange and treble.
Fig.2 Harbeth HL-P3, acoustic crossover on tweeter axis at 50", corrected for microphone response, with nearfield woofer response plotted below 300Hz.
Fig.3 Harbeth HL-P3, anechoic response on tweeter axis at 50", averaged across 30° horizontal window and corrected for microphone response, with nearfield woofer response plotted below 300Hz.
Fig.4 shows that the HL-P3 is much better behaved off-axis than the LS3/5A, its treble output being evenly well-maintained until 30° to the speaker's side. Beyond that the top octave increasingly drops down, as does the low treble, but the smooth manner of the off-axis differences—recessing the woofer obviously works—will mean that the speaker will not change its character too much when it is moved from an overdamped room to one that is more live. Vertically (fig.5), the balance doesn't change significantly as long as the listener is somewhere between the top and bottom of the front baffle. Sit so you can see the top of the cabinet, however, and a large suckout appears in the crossover region. Don't use too short a stand with this Harbeth. Other than the usual bumps and dips in the lower midrange and bass due to residual room effects, the spatially averaged in-room response (fig.6) is beautifully smooth and even.
Fig.4 Harbeth HL-P3, horizontal response family at 50", normalized to response on tweeter axis, from back to front: differences in response 90°–5° off-axis; reference response; differences in response 5°–90° off-axis.
Fig.5 Harbeth HL-P3, vertical response family at 50", normalized to response on tweeter axis, from back to front: differences in response 15°–5° above tweeter axis; reference response; differences in response 5°–10° below tweeter axis.
Fig.6 Harbeth HL-P3, spatially averaged 1/3-octave response in JA's listening room.
In the time domain, the HL-P3's impulse and step responses are shown in figs.7 and 8, respectively. The tweeter is connected with negative polarity, the woofer with positive, and the two outputs are, of course, not coincident in time. The waterfall plot calculated from the impulse response (fig.9) has only a degree of treble hash between 7kHz and 12kHz spoiling what would otherwise be a superbly clean decay. A slight bunching of the contour lines can be seen at the top of the woofer's passband, but this is probably inconsequential; looking at the waterfall plot of the woofer alone (not shown) revealed nothing of consequence.
Fig.7 Harbeth HL-P3, impulse response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Fig.8 Harbeth HL-P3, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
Fig.9 Harbeth HL-P3, cumulative spectral-decay plot at 50" (0.15ms risetime).
Remember the impedance wrinkle at 340Hz? That this is associated with a cabinet resonance is shown by fig.10, the waterfall plot calculated from the output of a small PVDF accelerometer fastened to the center of the side wall. A strong mode at 340Hz is noticeable, with one lower in level at 120Hz. Both modes are detectable on the cabinet top and back, and there is a strong similarity with the behavior of the LS3/5a cabinet, despite the differences in construction and material. The 120Hz resonance was much less noticeable with a stethoscope, however. It could be that the HL-P3's better sense of pace is partly due to the better handling of the lower-frequency cabinet mode.—John Atkinson
Fig.10 Harbeth HL-P3, cumulative spectral-decay plot of accelerometer output fastened to center of cabinet sidewall (MLS driving voltage, 7.55V, 2kHz bandwidth).