Mark Levinson No.32 Reference preamplifier Measurements
Unless otherwise noted, the measurements presented are for balanced operation in the standard gain settings (+6dB balanced, +12dB unbalanced), the level control set to maximum.
The output impedance of the Mark Levinson No.32 at its line output measured 13.9 ohms in the left channel, 14.2 ohms in the right (and 10.3 ohms left, unbalanced). The results for the balanced output impedance are just slightly higher than specified, but in any event are so low that they are still outstanding. The line-level input impedance measured a usefully high 103k ohms (96k ohms unbalanced), with essentially no difference between channels. The setting of the level control had no significant effect on input or output impedance, which is how it should be.
The output impedance at the tape output is 100 ohms with either a 50 ohm or a 600 ohm source impedance, which indicates that the tape outputs are buffered. DC offset at the No.32's outputs measured 0.3mV in the left channel, 1.6mV in the right (0.9mV and 0.6mV, respectively, unbalanced). The preamp is noninverting from its line inputs to its main outputs in unbalanced mode; in balanced mode, pin 2 is positive.
Line-stage gain measures 5.97dB in the balanced mode at the tested +6dB gain setting. Interestingly, however, I measured gain settings 6dB lower than specified for the unbalanced configuration. That is, at the standard gain setting of +12dB (the setting in which I took all of the unbalanced measurements), the gain is 6.1dB. The gain is 0dB at the +6dB setting, and 12.1dB at the +18dB setting. This nominal gain is that of the balanced output. Having half the voltage swing, the unbalanced output will offer half that "nominal gain." This should not result in any practical problems.
Signal/noise (ref. 1V, to the nearest dB) measured a low 107.5dB unweighted over a 22Hz-22kHz bandwidth, 93.5dB unweighted from 10Hz to 500kHz, and 109.5dB A-weighted. (The unbalanced figures were 107.5dB, 93.5dB, and 108.5dB, respectively.) I took these readings three times (and checked to make certain that the No.32 was actually turned on!) because I did not believe them. They are correct.
The measurements for crosstalk (fig.2) and THD+noise vs frequency (fig.3) were taken with an input of 1V (resulting in an output of approximately 2V) to minimize the effect of noise. The remaining curves were run with an input of 100mV.
Fig.1 Mark Levinson No.32, frequency response into 100k ohms. (0.5dB/vertical div.)
Fig.2 Mark Levinson No.32, channel separation. (10dB/vertical div.)
Fig.3 Mark Levinson No.32, THD+noise (%) vs frequency at 2V into 100k ohms.
At this point, with most products, we would discuss each curve. With the No.32, however, the results indicated in figs.1-5 speak for themselves. They are so outstanding, in all respects, that discussing them individually would be a waste of good ink. To the best of my recollection, we have never before seen a preamp that sailed through our measurements with such high marks.
Fig.4 Mark Levinson No.32, spectrum of 50Hz sinewave, DC-1kHz, at 10V into 100k ohms (linear frequency scale).
Fig.5 Mark Levinson No.32, distortion (%) vs output voltage into 100k ohms from unbalanced (top) and balanced (bottom) outputs.
The No.32 is one of those products that make you wonder, after you've measured them, why you bothered. Its test-bench performance is as close to perfect as the state of the art currently allows.—Thomas J. Norton