The small blips at 200Hz and 2kHz in fig.15 are due to the Audio Precision and should be ignored. The small peak at 120Hz is real, however, and can also be seen in the spectrum of the player's output while it decodes digital silence (fig.16; again, ignore the blips at 2kHz). Why should the right channel have power-supply noise present on SACD playback (albeit at a very low level) but not on CD playback? I have no idea, but it could have been introduced when Sony's provisional test SACD was mastered.

Fig.16 Sony SCD-1, spectrum of digital silence with noise and spuriae, DSD data. (1/3-octave analysis, "standard" output setting bottom at 80kHz.)

That test SACD didn't have the analytical jitter signal on it, but it did have an 11.025kHz tone at 0dBFS. Padding down the player's analog output to match the 2V RMS maximum input level of the National Instruments DSP card used to run the Miller Audio Research jitter test gave the high-resolution spectrum shown in fig.17. Obviously, no data-related sidebands are present, but the noise floor is, again, superbly low. However, it is no lower than I found from CD. This is presumably due to the player's analog stage rather than to the SACD medium. Again, sidebands can be seen at ±15.6Hz (purple markers "1"), but now a pair of sidebands exists at the power-supply frequency of ±120Hz (purple markers "3"). Again, I suspect that this might be disc-related rather than due to the player. The blue "16" marker is not jitter-related, but instead is a spurious single tone with an absolute frequency of 10.627kHz. However, it's so low in level that it won't matter subjectively.

Fig.17 Sony SCD-1, high-resolution spectrum of analog output signal (11.025kHz at 0dBFS, DSD data). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

Finally, when I first auditioned the SCD-1 for myself, I compared the CD release of Nature's Realm, Water Lily Acoustics' Philadelphia Orchestra recording, with the SACD version (See "As We See It"), not realizing that the SACD was a hybrid with a CD layer. The audible difference between the CD and the SACD was enormous—so much so that I suspected something other than DSD for the difference.

It turns out that the Water Lily CD must have been mastered from the analog tape at a different time using a different analog machine, as there was a large spectral difference between the CD and the SACD's DSD layer. Fig.18 shows a 1/3-octave spectral analysis of the first 3:30 of Liszt's Les Préludes, made with an Audio Control SA-3050A 1/3-octave analyzer set to its "peak hold" function. The blue trace is the DSD layer on the SACD, the red trace the Water Lily CD. Paradoxically, the CD sounded brighter than the SACD but actually has significantly less peak treble energy. It also has less peak low-frequency energy, and it did sound leaner.

Fig.18 Water Lily Acoustics, Liszt's Les Préludes, first 3:30, 1/3-octave peak spectral analysis from CD (red), SACD DSD layer (blue), and SACD CD layer (green). (10dB/vertical div.)

The CD layer on the SACD is shown in green in this graph: it is overlaid exactly by the blue DSD trace almost the entire time, which is why you can't see it. When it is visible, the difference is within the analyzer's 1dB margin of error—strong evidence that its peak spectrum is identical. Nevertheless, I did prefer this SACD's DSD layer to its CD layer.

The moral? When you compare SACD with CD, you may not be comparing like with like.—John Atkinson