Integrated Amp Reviews

I tested the Musical Fidelity M8xi with my Audio Precision SYS2722 system (see the January 2008 "As We See It"). I first preconditioned the amplifier by following the CEA's recommendation of operating it at one-eighth the specified power into 8 ohms for 30 minutes. At the end of that time, the top panel was very warm, at 109.6°F (43.1°C) and the heatsinks were too hot to touch, at 162.1°F (72.3°C). I then started performing my usual thermal stress test by running the amplifier at one-third power into 8 ohms for an hour. However, I cut the test short after 20 minutes, as the top panel's temperature was already 118.1°F (47.8°C) and that of the heatsinks 172.7°F (78.2°C) and I was concerned that I would damage the amplifier. Despite the massive heatsinks along its sides, the Musical Fidelity amplifier doesn't have quite enough thermal capacity for its power rating.

I looked first at the Musical Fidelity's performance via its line inputs. While the balanced inputs offered a maximum gain at the speaker outputs of 36.6dB into 8 ohms, the unbalanced inputs offered a significantly higher 43.8dB. Usually, a single-ended input offers 6dB lower gain than a balanced input. The maximum gain at the variable preamplifier outputs was 12.76dB, balanced in to balanced out, and 13.6dB, unbalanced in to unbalanced out. At the fixed-level, unbalanced line output, the gain was 0.4dB for the single-ended inputs and 7.24dB for the balanced inputs. The amplifier preserved absolute polarity at all sets of outputs for both unbalanced and balanced input signals. The volume control operates in accurate 0.5dB steps.

The single-ended line input impedance met the specification, at 25k ohms at 20Hz and 1kHz, though it dropped slightly to 21k ohms at 20kHz. The balanced input impedance was lower than specified, at 17.8k ohms across the audioband. The preamplifier output impedance was an appropriately low 48 ohms, single-ended, and 96 ohms, balanced, though the latter rose to 189 ohms in the low bass.

The Musical Fidelity's output impedance at the loudspeaker terminals was a little higher than usual for a solid-state design, presumably due to the bridged output stage. It measured 0.27 ohms at 20Hz and 1kHz, rising to 0.33 ohm at 20kHz. (The figures include the series impedance of 6' of spaced-pair loudspeaker cable.) The modulation of the amplifier's frequency response, due to the Ohm's law interaction between this source impedance and the impedance of my standard simulated loudspeaker, was ±0.2dB (fig.1, gray trace). The response into an 8 ohm resistive load (fig.1, blue and red traces) was down by 3dB just above 100kHz, which correlates with the Musical Fidelity's accurate reproduction of a 10kHz squarewave into that load (fig.2). Note that the very low bass in fig.1 rolls off a little, reaching –0.5dB at 20Hz and –1.5dB at 10Hz. Fig.1 was taken with the volume control set to its maximum; neither the frequency response nor the excellent channel matching changed at lower settings of the volume control.

Channel separation (not shown) was good, at >80dB in both directions below 4kHz and still 67dB at the top of the audioband. The Musical Fidelity's unweighted signal/noise ratio, taken with the unbalanced line inputs shorted to ground and the volume control at its maximum setting of "100.0," was a little disappointing at 55.9dB ref. 2.83V into 8 ohms (average of both channels). This ratio improved to 64.3dB when the measurement bandwidth was restricted to the audioband and to 67dB when A-weighted. The level of the background noise, which included spuriae at 60Hz and its even- and odd-order harmonics, depended on the setting of the volume control. This can be seen in fig.3, which shows the low-frequency spectrum of the amplifier's output while it drove 1kHz at 2.83V into 8 ohms with the volume control set to "100.0" (cyan and magenta traces) and to "80.0" (blue and red traces). Reducing the volume control setting by 20dB lowers the levels of the noise-floor components by around 13dB.

With both channels driven, the M8xi didn't quite meet its specified maximum power into 8 ohms of 550Wpc (27.4dBW), clipping (defined as 1% THD+noise) at 500Wpc (27dBW, fig.4). However, I don't hold the wall AC voltage constant for the tests; it measured 120V with the amplifier idle but dropped to 116.2V when the amplifier was clipping into 8 ohms. The Musical Fidelity clipped at 650Wpc into 4 ohms (25.1dBW, fig.5). (The wall voltage dropped to 113.5V with the amplifier delivering 650Wpc into this load.) Even with the volume control set to "90," the THD+N reading was dominated by noise below a few tens of watts, so I plotted how the THD+N percentage changed with frequency at 20V, which is equivalent to 50W into 8 ohms, 100W into 4 ohms, and 200W into 2 ohms. The distortion was very low into 8 ohms (fig.6, blue and red traces). However, it was higher into 4 ohms (cyan and magenta traces) and even more so into 2 ohms (gray trace), though still below 0.1% across the audioband.

The distortion was predominantly the subjectively benign third harmonic (fig.7). The volume control was set to its maximum for this measurement, and even with 32 captures averaged, the distortion waveform (bottom trace) was still overlaid with noise. Higher-order harmonics are present at lower levels, and the left channel had some second harmonic present (fig.8, blue trace). When the M8xi drove an equal mix of 19 and 20kHz tones at 30W into 8 ohms (fig.9), high-order intermodulation products all lay below –76dB (0.06%), and the second-order difference product at 1kHz can't be seen above the noise floor, even though I had the volume control set to "80.0" for this measurement to reduce the level of the noise.

The Musical Fidelity M8xi has optical and coaxial S/PDIF digital inputs, as well as a USB port. The coaxial inputs locked to data sampled up to 192kHz, the optical inputs to sample rates up to 96kHz. Apple's USB Prober utility identified the Musical Fidelity as "MF M8xi" from "Project." The USB port operated in the optimal isochronous asynchronous mode, and Apple's AudioMIDI utility revealed that the M8xi accepted 16- and 24-bit integer data sampled at all rates from 44.1kHz to 192kHz. The USB and S/PDIF inputs all preserved absolute polarity.

With the volume control set to "100.0," a 1kHz digital signal at –20dBFS resulted in an output level of 31.05V, which is equivalent to 120.5W into 8 ohms. As my measured clipping power into 8 ohms was 500W, which is equivalent to a voltage of 63.25V, it appears that the M8xi's digital inputs have around 14dB too much gain. The –20dBFS digital signal results in levels of 1.908V at the balanced preamplifier output, 953.6mV at the unbalanced preamplifier output, and 191.1mV at the fixed-level single-ended line output. To avoid damaging the Musical Fidelity's power amplifier stage with high-level digital signals, I performed all the measurements of the digital inputs' performance at the fixed-level outputs with the volume control set to "0.0."

The Musical Fidelity's impulse response with 44.1kHz data (fig.10) indicates that the reconstruction filter is a conventional linear-phase type, with time-symmetrical ringing on either side of the single sample at 0dBFS. With 44.1kHz-sampled white noise (fig.11, red and magenta traces), the M8xi's response rolled off sharply above 20kHz, reaching full stop-band suppression just above half the sample rate (vertical green line). An aliased image at 25kHz of a full-scale tone at 19.1kHz (blue and cyan traces) can't therefore be seen, though the distortion harmonics of the 19.1kHz tone are visible above the ultrasonic noise floor. The third harmonic is the highest in level at just above –60dB (0.1%). Peculiarly, the 19.1kHz tone and its third and fifth harmonics are accompanied by low-level, ±2kHz sidebands. The M8xi's digital-input frequency response was flat in the audioband with then a sharp rolloff just below half of each sample rate (fig.12).

Channel separation via the digital inputs was >100dB below 1kHz, but crosstalk rose linearly to –75dB at the top of the audioband, due to capacitive coupling between the channels. When I increased the bit depth from 16 to 24 with a dithered 1kHz tone at –90dBFS (fig.13), the noise floor components dropped by around 15dB, which implies that the M8xi offers between 18 and 19 bits' worth of resolution. However, a large number of low-level, power-supply–related spuriae were present in the noise floor. With undithered data representing a tone at exactly –90.31dBFS (fig.14, the three DC voltage levels described by the data were well-resolved. With undithered 24-bit data, the result was a somewhat noisy sinewave (fig.15).

I tested the M8xi for its rejection of word-clock jitter via both its TosLink and USB inputs. Though all the odd-order harmonics of the 16-bit J-Test signal's LSB-level, low-frequency squarewave were at the correct levels (fig.16, sloping green line), the spectral spike that represents the high-level tone at exactly one-quarter the sample rate was significantly broadened at its base. This suggests that the Musical Fidelity's digital inputs are affected by random, low-frequency jitter. With 24-bit J-Test data (fig.17), no jitter-related sidebands were present, but the spectral broadening could still be seen.

Musical Fidelity's M8xi offers high power and low distortion. Though its noise floor with the volume control set to its maximum was higher in level than I was expecting, the amplifier's high gain, especially with digital data and single-ended analog input signals, means that the volume control will never be used anywhere near its maximum.—John Atkinson