Mark Levinson No.326S line preamplifier Measurements
In full balanced mode, the Mark Levinson No.326S had a maximum voltage gain of 0.1dB, 6.1dB, 12.2dB, or 18.1dB, depending on how the input gain was set. The unbalanced figures were half this. The input impedance was 100k ohms at low and midrange frequencies for both balanced and unbalanced inputs, dropping slightly but inconsequentially at 20kHz to 86k ohms balanced and 67k ohms unbalanced. The unbalanced output impedance was a usefully low 29 ohms across the audioband, the balanced value only a little higher at 38 ohms.
The No.326S's frequency response was flat to well above the audioband and was just 0.75dB down at 200kHz (fig.1). This response was identical at all volume-control and gain settings for both balanced and unbalanced operation, and the channels were superbly well matched. Channel separation was also superb, at better than 120dB in both directions below 10kHz (fig.2). The preamp's signal/noise ratio depended on the gain setting. At 0dB gain, with the input shorted, the unweighted, wideband figure ref. 1V was an excellent 93.5dB, this improving further to 111dB when A-weighted. Switching in 6dB of line-stage gain reduced these ratios by less than 1dB, and not until I set the gain to 18dB did the ratios degrade by 9dB or so. This is one quiet preamplifier!
Fig.1 Mark Levinson No.326S, Volume control at max, gain = 0dB, balanced frequency response at 1V into (from bottom to top): 600 ohms, 100k ohms (0.5dB/vertical div.).
Fig.2 Mark Levinson No.326S, balanced channel separation (R–L dashed, 10dB/vertical div.).
The No.326S clipped (1% THD) at 16.4V, balanced drive, into 100k ohms (fig.3), and at 5.6V single-ended into the same load (fig.4), with very low distortion up to those points. The clipping point into 600 ohms was not much lower, but as fig.3 reveals, the THD starts to rise above 2.5V. However, this increased THD begins to become significant only above 10V output, much higher than the levels where the preamp will be working into power amplifiers with typical sensitivities. Even at 4V output into 600 ohms, the THD+noise percentage remained below 0.002% in the audioband (fig.5)!
Fig.3 Mark Levinson No.326S, 6dB gain, distortion (%)vs balanced 1kHz output voltage into (from bottom to top): 100k ohms, 600 ohms.
Fig.4 Mark Levinson No.326S, 6dB gain, distortion (%)vs unbalanced 1kHz output voltage into (from bottom to top): 100k ohms, 600 ohms.
Fig.5 Mark Levinson No.326S, 6dB gain, balanced THD+N (%)vs frequency at 4V into (from bottom to top): 100k ohms, 600 ohms.
The spectrum of that distortion for single-ended drive was primarily second-harmonic (fig.6), though this was at a very low level, –103dB (0.0007%), with the higher harmonics at the residual level of my signal generator;s output. With balanced drive at the same 1V level into 100k ohms, the harmonics were all at the background noise level; increasing the level to 5V and reducing the load to 600 ohms brought up the third harmonic, but still to just –90dB (0.003%).
Fig.6 Mark Levinson No.326S, unbalanced spectrum of 1kHz sinewave, DC–1kHz, at 1V into 8k ohms (linear frequency scale).
The Levinson's performance on the high-level, high-frequency intermodulation test was excellent into high impedances, particularly with balanced drive, where the 1kHz difference component lay at the –100dB mark (not shown). But with unbalanced drive into lower impedances, the difference component increased to –80dB (0.01%), with some higher-order components now visible (fig.7). These spuriae will not be audible, but they do suggest that the No.326S will generally be happier with higher-impedance loads.
Fig.7 Mark Levinson No.326S, unbalanced HF intermodulation spectrum, DC–24kHz, 19+20kHz at 1V into 8k ohms (linear frequency scale).
Overall, the No.326S offers measured performance that is beyond reproach, evident of some keen application of audio engineering skill.— John Atkinson