Mark Levinson No.37 CD transport & No.36S D/A converter Measurements part 2

Fig.6 shows the results of the Levinson decoding a 1kHz, undithered 16-bit sinewave at $90.31dBFS. This is an excellent result. Though there is some noise (as there always is at such low levels), the ideal stairstep response is clearly visible. Most CD players and D/A converters do far worse on this test.

Fig.6 Mark Levinson No.36S, waveform of undithered 1kHz sinewave at -90.31dBFS (16-bit data).

The noise modulation as a function of signal level vs frequency is plotted in fig.7. The more tightly clustered the results here—plotted for five low-level signals from -60dBFS to -100dBFS—the better. While the clustering here is not the best we have seen, the overall result is at such an extremely low level where the curves diverge that the audible significance of better clustering might be seriously disputed.

Fig.7 Mark Levinson No.36S, noise modulation, -60dBFS to -100dBFS (10dB/vertical div.).

Feeding a full-scale combined 19kHz plus 20kHz signal into the Levinson and performing a Fast Fourier Transform analysis of the output results in the intermodulation plot shown in fig.8. The artifacts are very low, with the highest—at 18 and 21kHz—over 80dB down (less than 0.01%).

Fig.8 Mark Levinson No.36S, HF intermodulation spectrum, DC-22kHz, 19+20kHz at 0dBFS (linear frequency scale, 20dB/vertical div.).

Robert Harley checked out the jitter on the No.36S using our Meitner LIM Detector with the PS Audio Lambda transport and cable we use for all of our jitter measurements. Fig.9 is the No.36S's jitter spectrum responding to a 1kHz, 0dBFS sinewave; fig.10 is the spectrum when processing all zeros (digital silence). The jitter components in fig.9 are visible at multiples of the signal frequency, but are low in level. The only significant component in the "digital silence" jitter in fig.10 is at 7.35kHz—the subcode rate jitter, which appears in all similar jitter measurements we have taken and is not a flaw in the converter. The RMS jitter in these two figures was so low that RH had difficulty separating it from the inherent jitter in the measuring devices. He estimated the RMS jitter at 12 picoseconds (fig.9) and 8ps (fig.10).

Fig.9 Mark Levinson No.36S, word-clock jitter spectrum, DC-20kHz, when processing 1kHz sinewave at 0dBFS; PS Audio Lambda transport (linear frequency scale, 10dB/vertical div., 0dB=1ns).

Fig.10 Mark Levinson No.36S, word-clock jitter spectrum, DC-20kHz, when processing digital silence; PS Audio Lambda transport (linear frequency scale, 10dB/vertical div., 0dB=1ns).

Fig.11 shows the result of the most difficult test, with an input of 1kHz at -90dBFS. While there are more jitter components visible here, the RMS jitter only increased to 60ps. We have never measured jitter any lower than this. In my judgment, we are here well below the level at which jitter might be audibly significant in any way. Though it's unlikely we will ever be able to eliminate jitter entirely, the No.36S comes about as close as we might hope to come to jitterless operation at any even remotely affordable cost.

Fig.11 Mark Levinson No.36S, word-clock jitter spectrum, DC-20kHz, when processing 1kHz sinewave at -90dBFS; PS Audio Lambda transport (linear frequency scale, 10dB/vertical div., 0dB=1ns).

The sort of measurements produced by the No.36S are hard to say anything very significant about: They're just too good to criticize.—Thomas J. Norton

COMPANY INFO
Mark Levinson
P.O. Box 781
Middletown, CT 06457
(860) 346-0896
ARTICLE CONTENTS

X