CD Player/Transport Reviews

The SD-22's frequency response (fig.1) is flat over most of the range, though its definite high-frequency rolloff beginning about 6kHz would likely translate into some softness in the upper octaves. The ripples in the top two audio octaves are typical of the Philips digital filter used. In fig.2, the SD-22's de-emphasized response tracks the HF rolloff of the player as a whole. The squarewave response in fig.3 is typical of the players in this group—a good rise-time combined with the moderate ripple indicating a linear-phase digital filter, though more overshoot than the other players. The crosstalk in fig.4 (only left to right is shown; right to left was virtually identical) was higher than that from the other players, though still unlikely to be audibly significant.

Fig.5 shows the SD-22's output when decoding a –90.31, dithered 1kHz sinewave. Some low-level non-linearity is present, along with some (likely) power-supply–related noise at 60Hz and 180Hz—though these are well below –100dB. The bump in the trace at 7.5kHz is suspicious, however, perhaps indicating some idling tones. Linearity is shown in figs.6 (left channel) and 7 (right). The left-channel response is fair, within 2dB down to –110dB, but the disappointing right channel shows a quite high deviation below –90dB. Both channels show unusual nonlinearity for a 1-bit player, most of which pass this test with flying colors. (I found this result so surprising that I ran it three times on three different days—with the same outcome each time.) The fade-to-noise with dither track from the CBS CD-1 test disc did not produce a particularly clean result, with several whistles (below the signal level) popping up intermittently as the signal faded to noise. The SD-22's noise spectrum (fig.8) shows the rising noise level above 20kHz typical of the other 1-bit machines, the minor power-supply noise artifacts at 60 and 180Hz, and little else of significance apart from the idling tone at 7.5kHz.

The SD-22's waveform of a 1kHz, –90.31dB undithered tone is presented in fig.9. The player's high noise floor makes this graph's interpretation a bit difficult, with little evidence of the desired stairstep response. The Philips digital filter has less stop-band rejection than most of the Japanese filter chips, as can be seen from the 19+20kHz (0dB) IM spectrum (fig.10). The cursor shows the 24.1kHz aliasing product to be 55dB down, however, which is still good.

The Sonographe inverted polarity at its main outputs, with 0V DC offset from both channels. As noted by CG in his review, its output was significantly lower than those of the other players here, producing 0.732V (left channel) and 0.733V (right) in response to a 1kHz tone at 0dB, these 8.7dB below the standard CD playback level of 2V RMS. Its output impedance measured 624 ohms (left channel) and 621 ohms (right).

Not unexpectedly, the measured results did not correlate well with the listening tests. In particular, measured low-level linearity, as it has a number of times in the past, did not follow the subjective results with any consistency. One of the top-rated players in the subjective tests was the Sonographe and its disappointing performance on this test showed a distinct difference between channels.—Thomas J. Norton