Mark Levinson No.360 D/A converter Measurements

Sidebar 4: Measurements

The Mark Levinson No.360 didn't invert absolute polarity from either set of outputs (the XLRs are wired with pin 2 hot). The maximum level was 2V single-ended, 4V balanced, sourced from 7 and 13.4 ohms, respectively. The frequency response (fig.1, top traces) was identical from both sets of outputs, flat within the audioband with a very slight (-0.1dB) droop apparent at 20kHz. Playing back a de-emphasized sweep (fig.1, bottom traces), the rolloff at 20kHz was larger but still inconsequential. The processor successfully locked to all sample rates up to 96kHz; fig.2 shows the frequency response at 32kHz, 48kHz, and 96kHz data rates.

Fig.1 Mark Levinson No.360, unbalanced frequency response at 0dBFS (top) and response at -12dBFS with de-emphasis (bottom) at 44.1kHz sample rate. (Right channel dashed, 0.5dB/vertical div.)

Fig.2 Mark Levinson No.360, balanced frequency response at 0dBFS at 32kHz, 48kHz, and 96kHz sample rates (2dB/vertical div.).

Crosstalk (not shown) was buried beneath the No.360's noise floor over most of the audioband. Driving the unit with 16-bit code representing a dithered 1kHz tone at -90dBFS resulted in the top pair of traces in fig.3, derived by sweeping the Audio Precision System One's 1/3-octave filter across the audioband. There are no spuriae present in these traces, and the noise floor, due to the dither noise present, smoothly rises with frequency because of the logarithmic-frequency nature of this measurement technique. Increasing the data bit depth to 24 bits gave the lower pair of traces, with an almost 20dB reduction in higher frequency noise. This implies an almost 20-bit dynamic range for the No.360, which is excellent performance.

Fig.3 Mark Levinson No.360, spectrum of dithered 1kHz tone at -90dBFS, with noise and spuriae, 16-bit (top) and 24-bit (bottom) data. (Right channel dashed.)