Mark Levinson No.30.6 Reference D/A processor:
Measurements part 3
Distortion was very low, there being just a trace (0.001%) of third harmonic apparent with a full-level 50Hz tone (not shown). Similarly, the Levinson's performance on the punishing high-frequency intermodulation test was excellent (fig.6).

Fig.6 Mark Levinson No.30.6, HF intermodulation spectrum, DC-22kHz, 19+20kHz at 0dBFS, 100k ohm load (linear frequency scale, 20dB/vertical div.).

In fig.7, I have tried not only to reveal the harmonics present in the Levinson's output, but also to show how close it comes to state-of-the-art resolution. The blue trace in this graph is a computer-generated 24-bit/1kHz tone analyzed in the digital domain; you can see that the FFT bins representing its noise and distortion are between -170dBFS and -180dBFS. The green trace is the spectrum of the Audio Precision System One's analog output fed to a dCS 904 24-bit professional A/D converter at an input level equivalent to -1dBFS. You can see that the dCS introduces a regularly decreasing series of low-order harmonics, starting with the second at -100dBFS (0.001%).

Fig.7 Mark Levinson No.30.6, spectrum, DC-22kHz, 1kHz at -1dBFS: 24-bit computer-generated data (blue); System One analog output digitized with dCS 904 ADC (green); Levinson output digitized with dCS 904 (red). (Linear frequency scale, 20dB/vertical div.)

The red trace is the computer-generated 24-bit tone converted to analog by the No.30.6 and redigitized by the dCS, again at -1dBFS. The noise floor from this process is comparable with that of the dCS A/D alone and about the 20-bit level. But it can be seen that while the second and third harmonics introduced by the No.30.6 lie below the level of those from the Audio Precision/dCS combination, the No.30.6 does introduce some higher-order harmonics. Even so, other than the fifth harmonic (5kHz, -100dBFS), these are all very low in level.

Finally, using the Miller Analyzer to look for the presence of jitter-related sidebands in the No.30.6's analog output, and using our standard PS Audio Lambda transport and an ST-optical datalink, gave a very low jitter figure of 153 picoseconds. Though not quite as good as the best I have measured—again see the Sony review in this issue—this is still excellent. The spectrum of the Levinson's jitter is shown in fig.8. The sidebands marked with red numbers are data-related and are all low; the two main jitter-related sidebands, indicated with purple "1" and "3" markers, lie at ±15.6Hz and ±395Hz.

Fig.8 Mark Levinson No.30.6, high-resolution jitter spectrum of analog output signal (11.025kHz at -6dBFS with LSB toggled at 229Hz). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

The Mark Levinson No.30.6's measured performance is about as good as it can get. No wonder I liked its sound so much.—John Atkinson