Mark Levinson No.360 D/A converter Measurements part 2

However, note that at lower frequencies, the increase in word length has unmasked a couple of power-supply components, at 120Hz and 240Hz. I tried different grounding arrangements to eliminate these, without success—but at around -130dBFS, their presence will have zero subjective effect. Changing the data to represent a -1LSB DC offset gave a smooth, low noise floor (not shown), again with some content noticeable at the power-supply frequencies but without any idle tones present.

This very low noise floor contributed to excellent linearity performance (fig.4). With 16-bit input data (top traces), any level error remained below 1dB down to -110dBFS, while increasing the word length to 24 bits revealed a very-low-level negative error that remained slight, even at -130dBFS. (The data were dithered for this measurement.) The excellent linearity and low noise resulted in reproduction of a textbook-perfect undithered 1kHz tone at -90.31dBFS (fig.5, 16-bit data; fig.6, 24-bit data).

Fig.4 Mark Levinson No.360, departure from linearity, 16-bit data (top), 24-bit data (bottom). (Right channel dashed, 2dB/vertical div.)

Fig.5 Mark Levinson No.360, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit data.

Fig.6 Mark Levinson No.360, waveform of undithered 1kHz sinewave at -90.31dBFS, 24-bit data.