CD Player/Transport Reviews

I looked at the Primare CD32's measured performance using my Audio Precision SYS2722 system (see www.ap.com and the January 2008 "As We See It"). For some tests, I also used my vintage Audio Precision System One Dual Domain. Unless otherwise mentioned, the measured results refer to the player used with its upsampling turned off and were taken from the balanced outputs. I tested the Primare's behavior only with test signals on CD; the CD32 didn't recognize uncompressed files stored on a flash drive plugged into its rear-panel USB port, only MP3s.

Error correction was superb, the CD32 not suffering from glitches in its output with the Pierre Verany Digital Test CD until the gaps in the data spiral reached 2.5mm in length. (The CD standard specifies only that a player cope with gaps of up to 0.2mm.) The maximum output level at 1kHz was 4.16V from the balanced jacks, 2.07V from the single-ended jacks, and both sets of outputs preserved absolute polarity (ie, were non-inverting). (The XLR jacks are wired with pin 2 hot.) The balanced output impedance was a low 94 ohms at all audio frequencies. Peculiarly, it was higher from the single-ended jacks, at 370 ohms at 20Hz and 1kHz, and 362 ohms at 20kHz—but this is still low in absolute terms.

The CD32's impulse response (fig.1) indicates that the digital reconstruction filter, a Burr-Brown DF1706, is a conventional linear-phase type, with symmetrical ringing before and after the single-sample impulse mapping the FIR filter coefficients. This impulse response was not affected by upsampling to 48 or 96kHz, and neither was the filter's ultrasonic rolloff (fig.2, red trace). This graph was taken with upsampling to 96kHz turned on; the ultrasonic noise floor was a little cleaner than with no upsampling, or with upsampling to 48kHz. Though the rolloff begins just above 20kHz, the output is suppressed by 20dB at the Nyquist frequency of 22.05kHz (half the sample rate), shown by the vertical green line in fig.2. Nevertheless, the image at 25kHz of a full-scale tone at 19.1kHz (fig.2, blue trace) is suppressed by 115dB. Distortion harmonics of this tone can be seen; the second, at 38.2kHz, is the highest in level, at –78dB (0.012%).

Looking at the audioband frequency response in more detail (fig.3, blue and red traces), the output is down by just 0.3dB at 20kHz, but some small ripples can be seen in the top two octaves. With pre-emphasized data (cyan and magenta traces, offset by 1dB for clarity), the CD32 applies the correct de-emphasis, with just a tiny (–0.05dB) plateau visible in the mid-treble. Channel separation (not shown) was superb, at >120dB in both directions between 200Hz and 4kHz, and still 115dB at the top of the audioband.

A problem in measuring a CD player that lacks a digital input that can handle data with a bit depth greater than 16 is that it becomes impossible to test the player's intrinsic performance. This is shown by figs.4 and 5: respectively, a 1/3-octave spectral analysis and a narrowband FFT spectral analysis of the player's output as it decoded data representing a dithered 1kHz tone at –90dBFS. In both graphs, the noise floor above 300Hz is actually the spectrum of the dither used to encode the signal. The CD's own noise is below that floor, except at low frequencies, where small peaks are evident only in the left channel (fig.4, solid trace; fig.5, blue trace), at the power-supply–related frequencies of 120 and 240Hz. No distortion harmonics can be seen, however.

Linearity error (not shown) was <1dB to –110dBFS, which is superb, and the CD32 correctly decoded undithered data representing a tone at exactly –90.31dBFS (fig.6). The waveform is symmetrical, DC offset is negligible, and the three DC voltage levels describing the data are well defined.

Even into the punishing 600 ohm load, harmonic distortion was relatively low (fig.7), though a series of low-level, high-order harmonics is evident. The second harmonic was higher in the left channel (blue trace) than in the right (red), at –78dB (0.012%) vs –91dB (0.0029%). The left channel also had more difference product evident when tested with an equal mix of 19 and 20kHz tones (fig.8). Finally, playing CD data representing the Miller-Dunn J-Test signal revealed no spectral spreading of the spike that represents the high-level tone at 11.025kHz (fig.9). The odd-order harmonics of the LSB-level, low-frequency squarewave are all close to the correct levels (indicated by the green line), and almost no other sidebands are visible.

The Primare CD32 offers excellent measured performance, but I keep coming back to that superb error correction. After all, a CD player's most fundamental task is to get the data off the disc, no matter how damaged that disc might be.—John Atkinson