Music Hall Marimba loudspeaker Measurements

Sidebar 3: Measurements

Both Sam Tellig ("Sam's Space," December 2012) and Stephen Mejias ("The Entry Level," June 2013) enthused over the price/performance ratio offered by Music Hall's Marimba bookshelf loudspeaker. Sam concluded that the Marimba was "an astonishing little speaker, as long as you don't expect it to do much." Stephen, too, was impressed by the Marimbas' startling imaging capabilities, and found the bass to be impressively tight, tuneful, and fast. With this consensus from the oldest and youngest of Stereophile's writers, I decided to run the Marimba through my speaker test regime.

I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the Marimba's frequency response in the farfield and an Earthworks QTC-40 for the nearfield responses. My estimate of the Marimba's sensitivity on its tweeter axis was a fraction of a dB higher than the specified 87dB/W/m. The impedance and electrical phase are shown in fig.1. Although the nominal impedance is specified as 6 ohms, the impedance magnitude drops below 6 ohms only in the lower midrange, reaches a minimum value of 5.4 ohms at 250Hz, and remains above 8 ohms for most of the audioband. Although the phase angle is highly capacitive at 120Hz and 3.5kHz, the impedance at those frequencies is high; the Marimba will therefore be an easy load for the partnering amplifier to drive.


Fig.1 Music Hall Marimba, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).

Two small discontinuities are visible in the impedance traces, one between 800 and 900Hz and the other just above 200Hz, implying the presence of cabinet vibrational modes at those frequencies. I investigated the cabinet's behavior with a piezoelectric plastic-tape accelerometer. Fig.2 shows a cumulative spectral-decay plot calculated from the accelerometer's output when it was fastened to the center of the sidewall. A resonant mode can be seen at 840Hz, but it is too high in frequency and too low in level to have any audible consequences. There is some liveliness visible between 200 and 300Hz, but it is commendably low in level.


Fig.2 Music Hall Marimba, cumulative spectral-decay plot calculated from output of accelerometer fastened to center of side panel (MLS driving voltage to speaker, 7.55V; measurement bandwidth, 2kHz).

The saddle centered on 58Hz in the impedance magnitude (fig.1, solid trace) suggests that this is the tuning frequency of the rear-facing port. This is confirmed by the fact that the minimum-motion notch in the woofer's output (fig.3, blue trace) occurs at the same frequency. (At this frequency, the woofer is held motionless by the back pressure from the port resonance.) The port's output (red trace) peaks as expected between 40 and 100Hz, but a couple of resonances are present in the midrange, the strongest of which occurs at the same 840Hz as the panel mode noted in fig.2, and coincides with the small peak in the on-axis farfield response (black trace). I could hear this port mode as a slight whistle with pink noise if I stood behind the speaker, but it was inaudible from the listening position in front of the speaker.


Fig.3 Music Hall Marimba, anechoic response on HF axis at 50", averaged across 30° horizontal window and corrected for microphone response, with nearfield responses of woofer (blue) and port (red) respectively plotted below 300Hz and 900Hz, and complex sum of nearfield responses plotted below 300Hz (black).

Below 300Hz, the black trace in fig.3 shows the sum of the nearfield woofer and port outputs, taking into account both acoustic phase and the different distances of the radiators from a nominal farfield microphone position. The apparent rise in response in the upper bass is entirely an artifact of the nearfield measurement technique; the speaker's output is down by 6dB at the port tuning frequency of 58Hz, which is relatively good extension for so small a speaker. Higher in frequency in fig.3, the small peak between 800 and 1200Hz might be expected to add a slight nasality, but neither ST nor SM commented on any such coloration. This graph was taken with the clumsy grille removed; adding the grille accentuates the 1kHz peak by a dB or so, but reduces the output in the tweeter's passband by up to 1.5dB. Without the grille, the response trend appears to be basically flat, but in this graph the tweeter seems to be balanced 2–3dB too high in level. Whether or not that will be perceived as excessive top-octave energy will depend also on the Marimba's dispersion in this region.

Fig.4 shows the Marimba's lateral dispersion, normalized to the speaker's response on the tweeter axis, which therefore appears as a straight line. The speaker does become increasingly directional above 7kHz, which, in a typical room, will compensate for the excess of energy in the on-axis output. The contour lines in this graph are otherwise even, something that correlates with the precise stereo imaging noted by Sam and Stephen. The Marimba's vertical dispersion is shown in fig.5, again normalized to the tweeter-axis response. A notch appears in the speaker's output in the crossover region more than 10° above and below the tweeter axis. The Marimbas should be used with stands tall enough to place the listener's ears close to the tweeter height, if the speaker is not to sound a little hollow.


Fig.4 Music Hall Marimba, lateral response family at 50", normalized to response on HF axis, from back to front: differences in response 90–5° off axis, reference response, differences in response 5–90° off axis.


Fig.5 Music Hall Marimba, vertical response family at 50", normalized to response on HF axis, from back to front: differences in response 45–5° above axis, reference response, differences in response 5–45° below axis.

Turning to the time domain, the Marimba's step response (fig.6) reveals that both drive-units are connected with the same, positive polarity; the smooth integration between the decay of the tweeter's step with the start of the woofers indicates optimal crossover design. The low-frequency undulation visible in the decay of the woofer step gives rise to a ridge of delayed energy in the Marimba's cumulative spectral-decay plot (fig.7) that coincides with the response peak in the upper midrange. Otherwise, this waterfall plot is superbly clean, especially considering the speaker's price.


Fig.6 Music Hall Marimba, step response on HF axis at 50" (5ms time window, 30kHz bandwidth).


Fig.7 Music Hall Marimba, cumulative spectral-decay plot on HF axis at 50" (0.15ms risetime).

I was impressed with the Marimba's measured performance, and by the fact that Music Hall's specifications don't exaggerate its sensitivity, but are actually conservative regarding the speaker's impedance. An honest loudspeaker offering honest performance at a very competitive price.—John Atkinson

Music Hall
108 Station Road
Great Neck, NY 11023
(516) 487-3663

LS35A's picture

Speaking of Epos and Roy Hall is anyone EVER going to review the new Elan speakers from Epos?  

They just don't seem to be getting any traction at all, which is a shame.  I trust the Epos brand and on paper they look great.