Arcam Alpha MCD 6-disc CD changer Measurements
The Arcam Alpha MCD's output impedance measured a low 123 ohms over most of the audio band, increasing to 600 ohms below 20Hz. This is still low enough that there should no problems of lightweight bass with the preamplifiers with which the player will typically be used. The player's analog output is absolute polarity-correct, and the maximum output level at 1kHz was 2.5V, almost 2dB above the standard 2V RMS. As BD noted, the left and right outputs were reversed on this sample.
The MCD's error correction was only fair, the player tracking up through track 31 on the Pierre Verany Test CD without any problems. Dropouts started to occur on track 32, however, which has 1.25mm gaps in the data spiral.
The top trace in fig.1 shows the MCD's frequency response at full level: the bass is flat down to 20Hz, reaching –0.5dB at 10Hz, while the treble has a subjectively inconsequential droop to –0.25dB at the upper band edge. The bottom pair of traces shows the response with de-emphasis switched in; the error—a maximum of –0.3dB at 16kHz—will make those few CDs that are pre-emphasized sound very slightly laid-back. (This might be a small error, but the bandwidth covered is large, meaning that "the area under the curve"—the missing treble energy—is also large, hence audible.) The channel separation (not shown) was around 80dB in both directions, with the right-to-left bleedthrough slightly lower than the left-to-right.
Fig.1 Arcam Alpha MCD, frequency response at 0dBFS (top) and de-emphasis response (bottom) (right channel dashed, 0.5dB/vertical div.).
Fig.2 shows a spectral analysis of the player's output while decoding data representing a dithered 1kHz tone at –90dBFS. The noise floor is higher than usual, and there are hints of some second- and third-harmonic distortion apparent. Note also the peak at the power-supply frequency of 120Hz. I tried various grounding schemes between the Arcam and the Audio Precision test setup, but this was the best I could do. However, this hum is still low enough in level that it's unlikely to be audible (except, perhaps, with very sensitive speakers).
Fig.2 Arcam Alpha MCD, spectrum of dithered 1kHz tone at –90.31dBFS, with noise and spuriae (16-bit data, right channel dashed).
Repeating this measurement, but changing the signal to "digital black" and extending the bandwidth to 200kHz, gave the spectrum shown as the lower pair of traces in fig.3. The high-frequency noise drops dramatically compared with fig.2, implying that the DAC actually mutes its output when it detects this special signal. I confirmed this by changing the signal to one representing a minuscule DC offset of –1LSB, which lifted the DAC muting. Now I got the upper pair of traces in fig.3, which reveal the massive amount of noiseshaping needed to achieve CD resolution from a 1-bit DAC. The noise boost takes place well above the audio band, but I do wonder how this might affect marginally stable amplifiers that might be used with the Arcam.
Fig.3 Arcam Alpha MCD, spectrum of digital silence (bottom above 2kHz) and of data representing a –1LSB DC signal (top above 2kHz), with noise and spuriae. (16-bit data, 1/3-octave analysis, right channel dashed)
Fig.4 shows the right channel's linearity error, plotted against absolute amplitude. There is a slight amount of negative error apparent at –90dBFS, this correlating with the distortion shown in fig.2, but the relatively high noise level results in apparent positive error below that level. The left channel was a bit better in this respect. The noise can also be seen in fig.5, which shows the waveform of an undithered 1kHz tone at –90.31dBFS. At this level, the waveform should be reproduced as a toggling among three distinct levels, with some overshoot apparent at each bit transition. This is obscured by the Arcam's noise floor, however.
Fig.4 Arcam Alpha MCD, right-channel departure from linearity (2dB/vertical div.).
Fig.5 Arcam Alpha MCD, waveform of undithered 1kHz sinewave at –90.31dBFS (16-bit data).
With respect to high-level linearity, the Alpha MCD offers very low levels of distortion. The spectrum of a full-level 61Hz tone can be seen in fig.6—the only harmonics poking their noses above the –100dBFS line are the second, third, and seventh. Similarly, the MCD produced very low levels of intermodulation products when playing back an equal mix of 19kHz and 20kHz tones, each at –6dBFS (fig.7).
Fig.6 Arcam Alpha MCD, spectrum, DC–1kHz, 61Hz at 0dBFS (linear frequency scale, 20dB/vertical div.).
Fig.7 Arcam Alpha MCD, HF intermodulation spectrum, DC–22kHz, 19+20kHz at 0dBFS (linear frequency scale, 20dB/vertical div.).
I assessed the MCD's jitter performance using the Miller Audio Research Analyzer, which is based on a National Instruments data-acquisition card in a host PC. The CD player being tested plays a special CD-R containing an 11kHz tone with a peak level of –6dBFS, overlaid with a low-level squarewave that toggles the LSB on and off at a rate of 229Hz. Fig.8 shows a narrow-band spectral analysis of the MCD's analog output centered on the 1kHz component, with symmetrical pairs of spectral components due to the player's jitter.
Fig.8 Arcam Alpha MCD, high-resolution jitter spectrum of analog output signal (11kHz at –10dBFS with LSB toggled at 229Hz). Center frequency of trace, 11kHz; frequency range, ±3.5kHz. (Grayed-out trace is the Meridian 508-24.)
The absolute clock error was reasonable, at +65ppm, though the absolute jitter level was quite high: 1451 picoseconds. Almost all of the measured jitter (1289ps) is due to the 120Hz-spaced sideband, marked with a dark blue "3." There are also some 60Hz and 180Hz sidebands (brown markers "2" and "4," at 99.6ps and 73.5ps, respectively). The second-highest jitter components (purple "9") are spaced 360Hz to either side of the fundamental; I have no idea what these are due to. Data-related jitter, indicated with red markers, is very low, however.
Predicting the subjective effect of a jitter pattern such as this is hard. However, the 60Hz sidebands might add a thickening of the bass, and might also contribute to the softening of transients noted by Brian Damkroger in his auditioning.—John Atkinson