Sidebar 2: Measurements
Like other PS Audio processors, the Digital Link II has a high output voltage—3.6V when decoding a full-scale, 1kHz sinewave. Output impedance was extraordinarily low, measuring 2.1 ohms at 20Hz, increasing only slightly to 2.6 ohms at 20kHz. This combination of high output voltage and low output impedance makes the Digital Link II ideal for use with passive level controls.
The Digital Link II's frequency response (fig.1) showed a gradual but rather early HF rolloff, being down 0.2dB at 10kHz and 0.6dB at 20kHz. This was very similar to the Sigma's rolloff, and correlated to the listening impressions of less treble detail. De-emphasis error (fig.1, bottom trace) was significant: 0.5dB positive error at 5kHz, increasing to +0.75dB at 10kHz, and decreasing slightly to +0.5dB at 20kHz. This will add a slight tizziness to the sound of pre-emphasized CDs. These kinds of de-emphasis errors are typically caused by component (resistor, and particularly capacitor) tolerances.
Fig.1 PS Audio Digital Link II, frequency response (right channel dashed, 0.5dB/vertical div.).
Channel separation (fig.2) was rather poor, measuring 70dB at 1kHz, decreasing to 47dB (R–L, worst case) at 20kHz. Note that this graph has been rescaled to show the rather high levels of crosstalk. A spectral analysis of the Digital Link II's output when decoding digital silence (not shown) revealed no converter misbehavior, but a rather high level of power-supply–related noise at 60Hz, 120Hz, and 180Hz. This noise can be seen in fig.3, a similar spectral analysis of the unit's reproduction of a dithered –90dB, 1kHz sinewave. Note that most of the power-supply noise is in the left channel. There is also an unusual peak at 2kHz in both channels, most likely caused by a converter error shifting some of the energy up an octave. (The 2kHz peak was not present when the Digital Link II was decoding digital silence.)
Fig.2 PS Audio Digital Link II, channel separation (right–left dashed, 10dB/vertical div.).
Fig.3 PS Audio Digital Link II, spectrum of dithered 1kHz tone at –90.31dBFS, with noise and spuriae (1/3-octave analysis, right channel dashed).
Linearity (fig.4) was good considering the fact that PS Audio doesn't trim the DACs' MSB value, but not exemplary in absolute terms. The Digital Link II's noise-modulation plot is shown in fig.5. The overall noise level and tightness of the groupings indicate good behavior, but there is a significant shift in the noise floor's spectral balance above 10kHz as a function of signal level. Note how the traces diverge above 10kHz. The Digital Link II's reproduction of an undithered 1kHz, –90dB sinewave is shown in fig.6.
Fig.4 PS Audio Digital Link II, departure from linearity (right channel dashed, 2dB/vertical div.).
Fig.5 PS Audio Digital Link II, noise modulation, –60 to –100dBFS (5dB/vertical div.).
Fig.6 PS Audio Digital Link II, waveform of undithered 1kHz sinewave at –90.31dBFS.
Looking at the Digital Link II's FFT-derived intermodulation spectrum when decoding a full-scale mix of 19kHz and 20kHz (fig.7), we can see a moderately high level of the 1kHz difference component (–90dB) and a fairly low level of sidebands. Overall, this is good performance. The unit's reproduction of a full-scale, 1kHz squarewave is shown in fig.8. There is slightly more overshoot on the leading edge than is seen from most processors using NPC filters. Finally, The Digital Link II doesn't invert absolute polarity, and there was a low level of DC at the outputs—between 200µV and 300µV.
Fig.7 PS Audio Digital Link II, HF intermodulation spectrum, DC–22kHz, 19+20kHz at 0dBFS (linear frequency scale, 20dB/vertical div.).
Fig.8 PS Audio Digital Link II, 1kHz squarewave at 0dBFS (2ms time window).
The Digital Link II's bench performance was only fair; the de-emphasis error was significant, low-level linearity was less than ideal, and the level of crosstalk was higher than that usually measured.—Robert Harley
