Linn Karik/Numerik CD player 1994 Measurements
The Karik had a maximum output level of 1.96V when playing a full-scale, 1kHz sinewave. Its output impedance measured just under 100 ohms at any audio frequency. DC-offset levels were below the voltmeter's sensitivity, implying that any DC was less than 500µV. The Karik doesn't invert absolute polarity; a positive-going impulse on a CD produces a positive-going impulse at the Karik's analog output. Frequency response and de-emphasis error (fig.1) were both excellent. Unlike most players, the Karik had no rolloff at 20kHz. The de-emphasis circuit had a negligible (0.1dB) negative error in the treble.
Fig.1 Linn Karik, frequency response (top) and de-emphasis error (bottom) (right channel dashed, 0.5dB/vertical div.).
Channel separation, shown in fig.2, was excellent, measuring 105dB at 1kHz and greater than 80dB at 16kHz. Fig.3, a spectral analysis of the Karik's output when decoding a -90dB dithered sinewave, reveals that the Numerik has a slightly higher noise-floor than other processors (compare fig.3 to fig.10, the Numerik's performance on this test). There's also a small peak of energy at 3kHz—something I've seen in other processors and players using the CS4328 DAC.
Fig.2 Linn Karik, crosstalk (right-left dashed, 10dB/vertical div.).
Fig.3 Linn Karik, spectrum of dithered 1kHz tone at -90.31dBFS, with noise and spuriae (1/3-octave analysis, right channel dashed).
The Karik's linearity (fig.4) was excellent, maintaining good performance to below -100dB, where the noise floor intrudes on the measurement. If you look at the Karik's reproduction of a -90dB undithered sinewave (fig.5), you can see a fairly high level of audioband noise overlaying the waveform. Fig.6 is the Karik's noise-modulation performance—a test that plots the device's noise floor as a function of frequency at five different digital input levels (-60dBFS to -100dBFS). The tighter the trace groupings, the better. The Karik shows some deviation in the traces below 2kHz, but good performance above that frequency. Note the noise floor's overall shape, with an increase in energy centered about 3kHz.
Fig.4 Linn Karik, right-channel departure from linearity (2dB/vertical div.).
Fig.5 Linn Karik, waveform of undithered 1kHz sinewave at -90.31dBFS.
Fig.6 Linn Karik, noise modulation, -60 to -100dBFS (10dB/vertical div.).
The Karik's intermodulation spectrum, taken when driving the player with a full-scale mix of 19kHz and 20kHz sinewaves, is shown in fig.7. The 1kHz difference component is moderately high in level (though still below -90dB), as are the other intermodulation products. Finally, the Karik's tracking ability was above average, with the player able to play track 35 on the Pierre Verany Test CD without a glitch. Average performance is track 33 or 34.
Fig.7 Linn Karik, HF intermodulation spectrum, DC-22kHz, 19+20kHz at 0dBFS (linear frequency scale, 20dB/vertical div.).
I didn't measure the Karik's jitter, because the DAC is hidden underneath a soldered-shut shield. Removing the shield may have allowed noise to intrude, resulting in an unreliable measurement.
The Numerik had a maximum output level of 1.995V, conforming to the CD standard of 2V output. Output impedance was a low 60 ohms across the band. DC levels were 0.7mV and 1.4mV at the left and right output jacks, respectively. The Numerik doesn't invert absolute polarity. Interestingly, the Numerik locked to 44.1kHz and 48kHz, but not to 32kHz; however, this point is moot—almost no 32kHz sources exist today. Unless otherwise noted, all Numerik measurements were taken from the unit with the 1702 DACs and the switching power supply.