A Case of the Jitters Good enough for hi-rez audio playback

Test Results 3: Good enough for hi-rez audio playback

Simaudio Moon Evolution Andromeda Reference CD player
$12,500, reviewed by Brian Damkroger in January 2008.
Playing CDs (the Simaudio also has a useful input for external data), the Andromeda offered low levels of word-clock jitter, at 286ps peak–peak. The spectrum of the player's output is shown in fig.17. The highest-level sideband pair is data-related, lying at 11.025kHz, ±229Hz, though most of the higher-frequency sidebands lie at the residual level. Evident is some slight spreading of the central peak due to random low-frequency time-base fluctuations, but the noise floor overall lies close to the Audio Precision's resolution limit.

Fig.17 Simaudio Moon Evolution Andromeda Reference, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).

Meridian 808.2 CD player–preamplifier
$14,995 (review forthcoming).
Though the Meridian's left channel has a very low noise floor (fig.18, blue trace) similar to that of the Simaudio Andromeda (fig.17), the right channel (red trace) is a little higher in level. However, both are 12dB or more below the CD's 16-bit quantizing floor (cyan and magenta traces). The data-related jitter sidebands lie at the residual level of the test signal, though some spectral spreading can be seen at the base of the central tone. Jitter level was a low 250ps peak–peak.

Fig.18 Meridian 808.2, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red); cyan and magenta traces show noise floor of perfect CD data.

Bryston BCD-1 CD player
$2695 (review forthcoming).
Considering it's just a CD player, the Bryston BCD-1 proved to be one of the highest-resolution disc players I have measured in the past year. The central peak in fig.19, representing the 11.025kHz tone, is narrow and spectrally pure, while the noise floor of both channels is close to the Audio Precision's resolution limit. The BCD-1's analog output offered around 150ps peak–peak of jitter—which is basically at the resolution limit of the Miller analyzer—and the evenly spaced low-level spikes in fig.19 are almost all residual harmonics of the 229Hz squarewave, not jitter-induced sidebands. However, the spuriae immediately to the left of the central tone are all a little higher in level than they should be, while those immediately to the right are all a little lower in level. I have no idea why this should be.

Fig.19 Bryston BCD-1, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).

Boulder 1021 CD player
$25,000 (review forthcoming).
The Boulder actually reads CDs at several times the normal rate, playing back the music from RAM. The noisefloor is well below the 16-bit level, not far above the AP's resolution limit, and while the data-related sidebands are all at the residual level, some spreading of the central tone is evident, due to the presence of random low-frequency timing uncertainty.

Fig.20 Boulder 1021, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).

Ayre C-5xe universal player
$5950, reviewed by Wes Phillips in July 2005.
This universal player's noise floor lies close to –148dBFS, which is both well below the quantization floor of a 16-bit PCM music signal (fig.20, cyan and magenta traces) and close to the resolution limit of the Audio Precision analyzer. The central peak in this graph, representing the 11.205kHz tone, shows very little spectral spreading at the base of the peak, which would result from low-frequency random noise affecting the accuracy of the player's master clock oscillator—and while a regular picket fence of spectral lines can be seen at the base of the graph, these are the residual odd-order harmonics of the Fs/192 squarewave. The C-5xe is not emphasizing these components. In fact, the only sidebands visible in fig.21 that are due to the Ayre player are the very-low-level sidebands at 11.025kHz, ±120Hz, and 11.025kHz, ±240Hz. These frequencies indicate that the sidebands are power-supply related, but so low in level that they will be of only academic interest. In terms of jitter rejection and resolution, the Ayre C-5xe offers close to state-of-the-art performance.

Fig.21 Ayre C-5xe, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red); cyan and magenta traces show noise floor of perfect CD data.

Conclusion
There is no consensus about what levels of jitter in a digital product's output are acceptable—the audibility will depend on both level and spectrum. Some authorities also insist that the ear will tolerate relatively high levels of jitter, up to a few nanoseconds, though that has not been the experience of this magazine's writers. But as an indicator of a product's ultimate resolution, these measurements of jitter and noise floor are illuminating. The McIntosh, PrimaLuna, and Krell are all disappointing in different ways, while the Simaudio, Meridian, Bryston, Boulder, and Ayre measurements in particular indicate the presence of some serious audio engineering talent on the staffs of these companies.

X