Linn Classik CD receiver Measurements: CD Section part 2

To generate fig.4, the CD section decoded data representing a dithered 500Hz tone that faded down to zero from -60dBFS. The level of the analog output was measured with a narrow-bandpass filter centered on 500Hz. An increasingly positive amplitude error develops below -75dBFS and plateaus at +6dB at -95dBFS, meaning that tones are twice the level they should be. Noise starts to dominate the measured level below about -100dBFS, and obscures the waveform of an undithered 1kHz tone at -90.31dBFS (fig.5).

Fig.4 Linn Classik, left-channel departure from linearity, 16-bit data (2dB/vertical div.).

Fig.5 Linn Classik, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit data.

Given the very low level of the signal in these four tests, it might be possible to lower the noise floor by increasing the volume. However, I didn't think this appropriate, as the first high-level tone to come down the pike would clip the Linn's power-amplifier section.

The Classik's preamp section offered very low distortion. Fig.6 shows the spectrum of a full-scale 50Hz tone with the volume control at "80." The third and fifth harmonics are close to the -100dB level, which is about the bottom of the Audio Precision System One's resolving power. The punishing mix of 19kHz and 20kHz tones, each at -6dBFS, which results in a waveform that just crests at the 0dB mark, gave a spectrum (not shown) relatively free from intermodulation products, the 1kHz difference tone lying about 83dB down. Decreasing the load impedance for these two measurements from 100k ohms to a very demanding 600 ohms introduced no significant increase in harmonic or intermodulation distortion.

Fig.6 Linn Classik, spectrum of 50Hz sinewave, DC-1kHz, at 0dBFS into 100k ohms (linear frequency scale).

Finally, I examined the Classik's jitter performance using the Miller Audio Research Jitter Analyzer. This drives the unit being tested with a high-level 11.025kHz tone (Fs/4), over which is overlaid the LSB toggling on and off at 229Hz. The analyzer performs a narrow-band spectral analysis on the player's analog output and searches the noise floor for symmetrical sideband pairs, which will be due to jitter. The Classik's result is shown in fig.7. As suspected, the analog noise floor is about 9dB higher than the very best CD replay. And while data-related jitter (indicated with red numeric markers) is low, there are power-supply-related sidebands at ±60Hz and ±180Hz (brown "1" and "2" markers) and a strong pair of sidebands at ±255Hz (purple "4"). The actual jitter level is low, at 506 picoseconds peak-peak. Still, many spurious noise spikes (blue markers) compromise the CD section's overall dynamic range.—John Atkinson

Fig.7 Linn Classik, high-resolution jitter spectrum of analog output signal at Preamplifier Out jacks (11.025kHz at -6dBFS with LSB toggled at 229Hz). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.