Cary 303/200 CD player Measurements
As supplied for review, the Cary player had a maximum output level of 5.94V from its balanced XLR jacks, 3V RMS from the unbalanced RCAs. The latter is 3.5dB higher than the CD standard's 2V, a very audible difference in A/B comparisons. (All the measurements were taken with the volume control at its maximum.) Neither set of outputs inverted absolute polarity, and the source impedance was usefully low, at 99 ohms unbalanced and 198 ohms balanced. (Both figures remained the same across the audioband.) The CD303/200's error correction was superb, the player coping with gaps in the data spiral up to 2mm in length with only occasional glitches.
The CD303's S/PDIF input successfully locked to datastreams with sample rates ranging from 32kHz to 96kHz. The frequency response for CD playback was flat within the audioband for both sets of outputs and didn't change when the upsampling was switched in. (Interestingly, when upsampling was switched in, the resultant 96kHz-sampled data was available at the CD303's digital output, with all 24 bits active.) The flat response is shown in fig.1. Also shown in this graph is the response with pre-emphasized CD data. The rising response means that, in common with a distressingly increasing number of modern players, the Cary doesn't switch in the appropriate de-emphasis. As a reasonably large number of older CDs are pre-emphasized, these will sound thin and screechy on the CD303. Channel separation (not shown) was excellent, at better than 110dB below 7kHz (R-L) and 20kHz (L-R).
Fig.1 Cary CD303/200, CD frequency response at -12dBFS into 100k ohms, with de-emphasis (bottom) and without (top), the latter with Standard filter (top) and Optional filter (middle). (Right channel dashed, 0.5dB/vertical div.)
Fig.2 shows 1/3-octave spectral analyses of the Cary's output while it decoded dithered 16- and 24-bit data representing a 1kHz tone at -90dBFS, both upsampled by the Cary to 24/96. The increase in bit depth dropped the noise floor by 11dB in the treble, implying D/A performance around the 18-bit level, which is good. There are no significant distortion harmonics apparent, though a small amount of 120Hz power-supply hum is evident in the left channel. This is not high enough in level to be audible, however.
Fig.2 Cary CD303/200, 1/3-octave spectrum of dithered 1kHz tone at -90dBFS, with noise and spuriae, 16-bit CD data (top) and 24-bit external data (bottom). (Right channel dashed.)
Of possibly more interest was the appearance of spectral components at 200Hz and 400Hz, the former high enough in level to break through the 16-bit noise floor. I thought at first that these were measurement spuriae, but as the respective levels of the noise in the left and right channels swapped over when I swapped leads, I must assume it is real. It can also be seen in the spectrum of the Cary's output while it decoded 16-bit "digital black," with the third harmonic at 600Hz and the fourth at 800Hz both visible above the noise floor (fig.3).
Fig.3 Cary CD303/200, 1/3-octave spectrum of digital black, with noise and spuriae, 16-bit CD data (right channel dashed).
Fig.4 shows the CD303/200's linearity error plotted with dithered CD data. It is negligible down to -110dBFS, which is excellent, confirming that the Cary's BB DAC chips have intrinsically better resolution than the CD medium's 16 bits. This can also be seen in figs.5 and 6, which show the CD303's reproduction of undithered data representing a 1kHz tone at exactly -90.31dBFS with 16- and 24-bit word lengths, respectively. The three voltage levels describing this signal with 16-bit data can be clearly resolved in fig.5, while the increase in bit depth results in a good if slightly noisy sinewave shape (fig.6).
Fig.4 Cary CD303/200, left-channel departure from linearity, 16-bit CD data (2dB/vertical div.).
Fig.5 Cary CD303/200, waveform of undithered 1kHz sinewave at -90.31dBFS, 24-bit external data.
Fig.6 Cary CD303/200, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit CD data.
The Cary offered low levels of harmonic distortion. Fig.7 shows an FFT-derived spectral analysis of the player's single-ended output while it decoded CD data representing a 1kHz tone at 0dBFS. The load was 8k ohms. The THD (actual sum of the harmonics, disregarding noise) was a very low 0.0025%, with the consonant second harmonic the highest in level at -93dB, followed by the fourth at -99dB. However, although the third harmonic is vanishingly low in level, higher-order harmonics up to the 11th appear in this graph. These are all very low in level, so it would be foolhardy to correlate this behavior with BD's occasionally finding the CD303/200's upper midrange to sound edgy. But their existence is a little alarming in absolute terms. The good news was that intermodulation distortion was almost nonexistent at 0.0009% (fig.8).
Fig.7 Cary CD303/200, spectrum of 1kHz sinewave, DC-1kHz, at 0dBFS into 8k ohms, CD data (linear frequency scale).
Fig.8 Cary CD303/200, HF intermodulation spectrum, DC-25kHz, 19+20kHz at 0dBFS into 8k ohms, CD data (linear frequency scale).
The Cary's rejection of wordclock jitter differed according to whether the external data input or internal CD playback was used, and whether or not the upsampling was switched in. The lowest jitter level, measured with the Miller analyzer, was a good 256 picoseconds peak-peak, this obtained for CD playback without upsampling. Switching in the upsampling increased the jitter level to 312ps, which is still low in absolute terms. The spectrum of that jitter is shown in fig.9. While data-related sidebands (red numeric markers) are low in level, two sets of sidebands, at ±15Hz (purple "1") and ±1870Hz (purple "12"), contribute most to the measured total. Sidebands at the power-supply-related frequency of 120Hz (purple "2") contribute 53ps to the total, while a pair of sidebands also exists at ±200Hz (purple "3"), the frequency of the spuriae seen in figs.2 and 3.
Fig.9 Cary CD303/200, 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. (Grayed-out trace is for external data sourced from a PC with an RMS Digi96/8 Pro soundcard and a TosLink connection.)
Changing to external data supplied via a TosLink connection dropped the measured jitter level slightly, to 306ps, which suggests good jitter rejection on the digital input, and also made the central peak less broad.
Overall, this is pretty good measured performance, though I was concerned by the lack of de-emphasis and the presence of high-order harmonics in the Cary's output.—John Atkinson