Sidebar 2: Measurements
The R1 fared very well on both the CBS CD-1 and Pierre Verany PV.788031/788032 test discs.
The CBS has several series of 1kHz tones of diminishing level, down to –90dB and below, with and without dither added. Nearly every player I have tried them on has produced severe harmonics and/or buzzing from the undithered –80 and –90 tracks, and most were incapable of producing a recognizable pitch from the –100dB dithered signal; it became undifferentiated hiss. With the R1, the undithered –80 track had perceptible even-harmonic content and the –90 track had noticeably more, but there was virtually no perceptible buzzing, indicating God-knows what. The dithered –100dB track was (as always) very hissy, but the 1kHz tone was still clearly audible. (Note that –100dB is 4dB below the theoretical dynamic-range floor of a 16-bit system; proof that dither does what it is claimed to do.)
Measured departure from linearity at very low levels was the best John Atkinson (who did the testing) has seen: within the experimental error limits, down to –80.77dB (ie, perfect), and between 0.5dB and 2dB of compression at –90.31dB (ie, effectively perfect). The undithered waveform at this level had an almost perfect three-step shape (fig.2). It consists of a progression of the codewords representing –1, 0, and +1), though some asymmetry was visible in that the positive-going step was smaller than the negative-going.
The error-correction tracks on the Verany disc have "blemishes" ranging from 0.05 to 4.0mm across. The CD standard calls for the ability to fully correct a data loss of up to 0.2mm across, but the Cross-Interleaved Reed-Solomon Code used for CD error correction allows, theoretically, for full correction of dropouts of up to 2.4mm across. The R1 did not quite achieve the ideal; at 2mm, there were occasional bursts of 1rpm clicks, and at 2.4mm, the clicks occurred once for every disc revolution. At 2.5mm, there was intermittent muting—indicating the limit of the player's interpolation capability. The results were the same with the closer-spaced tracks which followed: clicks were occasionally audible at 2mm, and muting set in at 2.4mm.
The successive-dropout tests consist of two signal interruptions of the same size, separated by an equivalent space. The test indicates the rapidity with which a player's error-correction system can "reset" itself after having performed a correction. The R1 had no trouble with this test, responding exactly as it had to the single-dropout tests.—J. Gordon Holt
Fig.3 shows that the Sony DAS-R1 features exactly 2dB of compression apparent at –90.31dB—a little worse than I had previously assessed using a 'scope, but still excellent. Though the 1kHz peak is well-defined, peaks can be seen at its second, third, and possibly fourth harmonic. LF noise, however, was at least as good as the Marantz CD94 in absolute terms, and more than 6dB better with respect to the level of the fundamental. That the HF noise is a function of the recording and not the players can be seen by the correlation of the former's absolute level with that of the 1kHz tone.
Fig.2 Sony DAS-R1, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data.
The Pierre Verany disc contains tests that no other CD test discs have: tracks to test for a player's ability to cope with normal and abnormal variations in signal-track speed and spacing, ill-defined pit-to-land transitions, and the triple-threat challenge of two dropouts occurring in quick succession. Some of these gave the R1 a real workout, and the fact that it failed some of the most demanding ones (ones that no other player would probably track, either) allowed me to quantify some of its limitations. Details:
According to Verany's instructions (which are another hilarious example of what I call Finglish—foreign English, sort of like generic Japlish), it is unlikely that any player could do as well on these as the Sony did. But then Sony CD players have long had a reputation for unequalled tracking and error-correction capabilities. And it probably helped that Sony used this same test disc for evaluating the performance of their R1.
On the first P-V tests, 23 tracks cover the variations in pitch (spacing) and linear velocity of the data tracks which are permitted by the CD specifications "red book." The R1 sailed through all of them without a glitch. The next four tracks contain simulated dropouts, similar in test function to those of the Philips 5A test disc but differing in that, while the Philips disc has the blemishes applied to the disc surface, the P-V used instead interruptions of the laser cutting beam to produce essentially the same interruptions in the recorded pit patterns. The effect is not quite the same, though: while interruptions at the disc surface are subject to defocusing by the optical system, and are thus "seen" as having indefinite boundaries, the laser-simulated "blemishes" are sharply focused and have "hard" boundaries. Nonetheless, the P-V disc serves well enough as a standardized test for comparing one player with another.
Fig.3 Sony DAS-R1, spectrum with noise and spuriae of dithered 1kHz tone at –120dBFS with 16-bit data (10dB/vertical div.).
As the code representing this tone has had dither applied when the CD was cut, it should reproduce as a pure, if noisy, sinewave. Any distortion components present, therefore, can be laid at the door of the individual player's decoder and electronics. As this graphical representation of a player's low-level performance is more informative than a straight statement of by how many dB it compresses or expands the level at –90.31dB, from now on we shall make this measurement (an average of an arbitrarily decided number of sample spectra) a standard feature of Stereophile's CD player and decoder reviews.—John Atkinson
