CD Player/Transport Reviews

I measured the Metronome CD8 S with my Audio Precision SYS2722 system (see www.ap.com and the January 2008 "As We See It"). As well as using test signals on CDs, I tested the Metronome by feeding its coaxial input S/PDIF data from the SYS2722, and its USB input data sourced from a battery-powered MacBook Pro running Pure Music 2.0. The S/PDIF input would accept data sampled at all rates from 44.1 to 192kHz. Apple's US Prober utility identified the Metronome as "Combo384 Amanero" from "Amanero Technologies," and confirmed that it operated in the asynchronous mode. Mac's AudioMIDI utility stated that the Metronome's USB input would accept 32-bit integer data at all sample rates from 32 to 384kHz. However, when I fed the CD8 S's USB data sampled at 384kHz, although the player's display correctly indicated "P 384," there was no audio output.

Tested with the Pierre Verany Digital Test CDs, the CD8 S demonstrated superb error correction, playing tracks with gaps in the data spiral of up to 3mm without skipping. The maximum output level at 1kHz was 6.06V from the balanced outputs and 3.06V from the unbalanced outputs, the latter 3.7dB greater than the CD standard's 2V. Both outputs preserved absolute polarity (ie, were non-inverting), the XLR jacks being wired with pin 2 hot. The unbalanced input impedance was very low, at 57 ohms at all audio frequencies. The balanced output impedance was appreciably higher, at 1196 ohms across the audio band. Channel separation (not shown) was excellent, at >105dB in both directions across the audioband.

Fig.1 shows the Metronome's impulse response with CD data: a conventional, time-symmetrical Finite Impulse Response (FIR). The red and magenta traces in fig.2, taken with white noise burned to a CD-R Audio disc, reveals that this filter rolls off sharply above half the sample rate (indicated by the green vertical line), the result being that the aliased tone at 25kHz that results from a full-scale signal at 19.1kHz (blue and cyan traces) is suppressed by 85dB.

The CD8 S's measured performance so far appeared to be as expected, but when I looked at its frequency response I ran into problems. The green and gray traces in fig.3, taken with tones from a test CD, show the same sharp rolloff seen in fig.2. The cyan and magenta traces in fig.3, taken with 96kHz-sampled data fed to the Metronome's S/PDIF input, feature a sharp rolloff above 43kHz. However, with 192kHz-sampled data fed to the S/PDIF input, I got the blue and red traces in fig.3: the response overlaps the 96kHz traces up to 48kHz, including the very sharp rolloff, but then returns to full level above 50kHz. Puzzled, I repeated the test with USB data and got the same result. There is something awry with how the CD8 S handles data sampled at rates greater than 96kHz.

Looking at this result for data sampled at 192kHz in greater detail, note that the horizontal scale in the graph is the frequency of the input signal encoded by the digital data. When I actually looked at the frequency of the analog signal being output by the Metronome for input frequencies above 48kHz, it was actually an aliased product. For example, when the input signal had a frequency of 60kHz, the output signal was actually 36kHz (96,000–60,000). So what you see above 48kHz with the blue and red traces in fig.3 are frequencies mirrored above the 96kHz Nyquist frequency (half the 192kHz sample rate), not actual audio. Real music does not have high-level content above 48kHz, so this behavior might look worse than it sounds. But again it suggests that the CD8 S does not correctly handle high–sample-rate data.

I asked Art Dudley if he had heard any difference in sonic character when playing 96kHz data vs CD-derived data; he hadn't, responding that "they were similarly pacey, and not the least bit harsh."

With 44.1kHz-sampled data fed to the S/PDIF input, an increase in the bit depth from 16 to 24 dropped the noise floor by more than 20dB (fig.4), suggesting that the CD8 S's DAC section offers resolution approaching 20 bits, which is excellent. But note the appearance in fig.4 of odd-order harmonics with 24-bit data (blue and red traces), which suggests that the LSB is being truncated somewhere in the signal-processing circuitry. I got an identical result with 44.1kHz data fed to the Metronome's USB input, though there was now a low-level spurious tone at 5.7kHz present (fig.5).

Then I found more anomalous behavior. Repeating the test used to produce fig.4 with 24-bit data sampled at 48kHz fed to the S/PDIF input, I got the result shown in fig.6. The odd-order harmonics are still present, but the spectrum of the 1kHz tone at –90dBFS is now overlaid with a large number of low-level spikes. I repeated the test with data sampled at 96kHz and 176.4kHz and got similar results to that shown in fig.6, though with data sampled at 88.2kHz, the spectrum was as clean as it had been in fig.4. It appears that while the Metronome performs well on this test with data sampled at 44.1kHz and 88.2kHz, it has problems with 48kHz-family data (48, 96, 192kHz) as well as with 176.4kHz data. A puzzle—unless the CD8 S uses a sample-rate converter for data sampled at frequencies other than 44.1kHz.

Nevertheless, with 16/44.1 data representing an undithered 1kHz tone at exactly –90.31dBFS, the output waveform was essentially perfect (fig.7), with the three DC voltage levels clearly defined. With undithered 24-bit data, the result was an excellent sinewave (fig.8).

The Metronome player offered low levels of harmonic distortion. Even into 600 ohms (fig.9), the second harmonic, the highest in level, lay at –77dB (0.014%). However, a regular series of low-level paired spuriae can be seen in this graph, which was taken with 24/44.1 data. Repeating the test with 24/96 data (fig.10) increased the level of the second harmonic to –70dB (0.03%), which is probably not significant. However, many more spuriae can be seen, again suggesting that the CD8 S does not handle 48kHz-family data as well as it does 44.1kHz-family data.

The CD8 S performed well on the demanding high-frequency intermodulation test, with all the distortion products at extremely low levels (fig.11). But note the spectral spreading at the bases of the two primary tones. This is due to sidebands at the power-supply–related frequencies of ±120, ±240, and ±360Hz, etc. And when I tested the CD8 S's rejection of word-clock jitter with 16-bit J-Test data burned to a CD-R Audio disc, the spectrum was spoiled by similar sidebands, at much higher levels than I have found with other players (fig.12). The odd-order harmonics of the LSB-level low-frequency squarewave are all slightly higher than they should be, the desired levels indicated in this graph by a green line. Repeating this test with S/PDIF data gave a very similar result, while USB J-Test data resulted in the production of much worse jitter-related sidebands in the left channel (fig.13, blue trace).

To check that there was not some kind of incompatibility with the Audio Precision test system, I recorded the Metronome's output on a battery-powered digital recorder while it played the J-Test CD-R. Performing FFT analysis on the resulting WAV file gave a result similar to that shown in fig.10. When I then looked closely at the 44.1kHz-sampled 19.1kHz data used to generate the blue and cyan traces in fig.2, it appeared that the lowest amount of supply-related sidebands could be seen with USB data. CD data were slightly worse, and S/PDIF data worse still.

One thing I did note was that the Metronome's data receiver could work with sample rates very different from the "legal" frequencies. For example, when I fed the S/PDIF input with 1kHz data sampled at 200kHz, the player's display still indicated "192" and the output was a 1kHz sinewave. When I fed the Metronome 1kHz data sampled at rates ranging from 42kHz to 47kHz, this display indicated "44:1" and the output was a consistent 1kHz sinewave. This tolerance for the input-data sample frequency is generally not felt to be a good thing, because it means that the receiver circuitry will have compromised word-clock jitter rejection.

The most likely cause of the supply-related sidebands in figs. 12 and 13 is inadequate rejection of voltage-rail ripple on the DAC chip's reference-voltage pin. I'd had the Metronome plugged into the wall with a conventional power cord fitted with a three-pin, grounded plug for all the testing. But when I repeated the testing of the player's jitter rejection, lifting the ground with a cheater plug, there was no significant change in the measured performance.

The Metronome CD8 S is a beautiful-looking audio component. But I was concerned about its idiosyncratic measured performance, especially with data sampled at rates other than 44.1kHz and 88.2kHz. It could be argued that as the result of the questionable behavior occurs at low levels, it might not have a major effect on sound quality. And it is fair to note that Art Dudley did most of his auditioning of the CD8 S with data sampled at 44.1kHz, where the player performs at its best. Nevertheless, its measured performance suggests that the Metronome player is sub-optimally engineered.—John Atkinson