Audio Research Reference Two preamplifier Measurements

Sidebar 3: Measurements

I measured the performance of the Reference Two in both full single-ended and full balanced modes. Maximum voltage gains were 15.2dB balanced in to balanced out, and 13.8dB single-ended in to single-ended out. With a single-ended input, the maximum gain available from the balanced outputs was just under 20dB. The preamp was non-inverting from both sets of outputs. The unity-gain settings of the volume control were 2:00 balanced, 2:30 unbalanced.

The preamp's input impedance varied from 52.5k ohms (unbalanced at 20kHz) through 66.5k ohms (unbalanced, bass through upper midrange) and 163k ohms (balanced at 20kHz), to 235k ohms (balanced, bass through midrange). The output impedance was basically 440 ohms balanced and 270 ohms unbalanced, though these figures did rise at the bottom of the audioband to 625 and 750 ohms, respectively.

As a result, the bass rolled off early into the very low 600-ohm load (fig.1, bottom trace), though into the more typical 100k ohm load (top trace) the Reference Two offered excellent extension at the low-frequency extreme. The high frequencies were slightly curtailed into both loads, at -1.5dB at 20kHz. As MF noted, "the Ref 2's top end was ever so slightly soft."

Fig.1 Audio Research Reference Two, balanced frequency response at 1V output with volume control at maximum gain into 100k ohms (top below 100Hz) and 600 ohms (bottom below 20kHz). (0.5dB/vertical div.)

Channel separation (fig.2) was excellent, but the ubiquitous 6dB/octave rise in crosstalk with frequency due to capacitive coupling can be seen in this graph. (Double the frequency and you get double the crosstalk.) Channel separation was slightly worse in the balanced mode, presumably because then you have double the number of signal-carrying paths.

Fig.2 Audio Research Reference Two, channel separation with volume control at its maximum, from top to bottom: R-L, L-R, balanced; R-L, L-R, unbalanced (10dB/vertical div.).

Distortion at 1V output into 100k ohms was very low in either mode (fig.3), though the unbalanced THD was a little higher overall, perhaps due to the contribution of noise to this result. Figs.4 and 5 show the spectrum of a 50Hz sinewave at 2V in unbalanced and balanced modes, respectively. Now it is the balanced result (fig.4) that has slightly higher THD, with the second and third harmonics around -70dB (0.03%). Note also the presence of an AC supply-related component at 180Hz (-76dB, or 0.015%), presumably due to magnetic coupling. The only harmonic of significance in the unbalanced spectrum (fig.5) is the second, at -74dB. While the 180Hz component is gone, a 120Hz electrical hum can now be seen at -90dB (0.003%).

Fig.3 Audio Research Reference Two, distortion (%) vs frequency at 1V into 100k ohms, balanced (bottom) and unbalanced (top).

Fig.4 Audio Research Reference Two, unbalanced spectrum of 50Hz sinewave, DC-1kHz, at 1V into 100k ohms (linear frequency scale).

Fig.5 Audio Research Reference Two, balanced spectrum of 50Hz sinewave, DC-1kHz, at 1V into 100k ohms (linear frequency scale).

Audio Research
3900 Annapolis Lane North
Plymouth, MN 55447-5447
(763) 577-9700