Audio Research Reference 75 power amplifier Measurements
I measured the Audio Research Reference 75 using Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see the January 2008 "As We See It" and www.ap.com). Before performing any measurements, I ran the amplifier for an hour; then, before performing any measurements, and using the front-panel meters, I checked that the bias was correctly set for each of the four KT120 output tubes, as instructed in the manual.
The Reference 75 has two output-transformer taps for each channel, marked 8 and 4 ohms, and a common ground terminal. The amplifier preserved absolute polarity from both taps (ie, was non-inverting), with the input XLRs wired with pin 2 hot. The voltage gain at 1kHz into 8 ohms was 24.6dB from the 8 ohm tap and 21.5dB from the 4 ohm tap, both lower than the norm. The input impedance was very high and close to the specified 300k ohms at low and middle frequencies, dropping to a still-high 205k ohms at 20kHz.
The output impedance was relatively low for an amplifier using a single pair of output tubes, at 0.74 ohm at 20Hz and 1kHz, and 0.95 ohm at 20kHz from the 4 ohm transformer tap. From the 8 ohm tap these impedances were 1.25 and 1.56 ohms, respectively. As a result, the variation in frequency response, due to the usual Ohm's Law interaction between these impedances and that of our standard simulated loudspeaker, was relatively small, at ±0.5dB from the 4 ohm tap (fig.1, gray trace) and ±0.8dB from the 8 ohm tap (fig.2, gray trace). These graphs also indicate that the Reference 75 has a wide bandwidth, with a flat response down to the very low bass. At the opposite end of the spectrum, however, a slight kink is visible in the traces two octaves above the audioband, particularly into higher impedances. These kinks indicate the presence of an ultrasonic resonance, most likely in the output transformer. This resonance is well damped, however, with just three cycles of ringing visible on the leading edges of a 10kHz squarewave (fig.3). A 1kHz squarewave (fig.4) was reproduced with a superbly square waveform from both taps.
Channel separation (not shown) was superb, at >100dB RL and >110dB LR below 3kHz, and still around 90dB at the top of the audioband. Noise levels were also low, the unweighted wideband signal/noise ratio from the 4 ohm tap (ref. 2.83V into 8 ohms, assessed with the input shorted) measuring 86dB in the left channel, 85.3dB in the right. With an A-weighting filter in-circuit, these ratios improved to 100.6dB left and 102.4dB right. The ratios from the 8 ohm tap were all a little lower, particularly in the left channel, but this is still a quiet amplifier.
The front-panel meters indicated an output of 1W when the measured power output was 1W into 8 ohms from the 4 ohm tap. Audio Research specifies the Reference 75's maximum power as 75Wpc (18.75dBW into 8 ohms) at typically 0.6% THD. Figs. 5, 6, and 7 show how the THD+noise percentage increases with power into 8, 4, and 2 ohms at 1kHz from the 4 ohm output transformer tap. We define clipping as when the THD reaches 1%; the Reference 75 doesn't quite meet its specified power at that level of distortion, delivering 61Wpc into 4 ohms (14.8dBW, fig.6), and lower powers into the other impedances: 43Wpc into 8 ohms (16.33dBW, fig.4), and 40W into 2 ohms with one channel driven (10dBW, fig.7). However, relaxing the definition of clipping to 3% THD, the Reference 75's 4 ohm tap gives 46.5Wpc into 8 ohms (16.67dBW), 70Wpc into 4 ohms (15.45dBW), and 94W into 2 ohms (13.7dBW). Slightly more power was available from the 8 ohm tap into 8 ohms (fig.8): 70Wpc at 1% THD (18.45dBW) and 78Wpc at 3% THD (18.9dBW).
In Figs. 58, the upward slopes of the traces above 1W or so reveal a fairly linear increase in distortion with increasing power, this a result of the small amount of negative feedback used. However, at low powers into loads matched to or greater than the nominal tap impedance, the Reference 75 offered respectably low levels of distortion at low and middle frequencies. The lowest distortion in absolute terms was obtained from the 8 ohm tap (fig.9), though the right channel (red and magenta traces) was not quite as linear as the left (blue and cyan). The distortion from the 8 and 4 ohm taps (fig.10) rose in the top two octaves and as the load impedance was reduced. However, there was no rise in THD at low frequencies, suggesting that the Reference 75's output transformers have an adequate amount of iron in their cores.
Fortunately, the distortion at low powers was dominated by the relatively innocuous third harmonic (fig.11), though the second and higher-order harmonics make an appearance as the power increases toward clipping and the load impedance drops (fig.12). The presence of sidebands at 120Hz intervals around the main signal at even moderate powers (fig.13) suggests that the power supply is working hard, though hum at 120Hz was respectably low in level. This graph also reveals that a little more low-frequency noise is present in the right channel than in the left. Finally, though the 1kHz difference product resulting from an equal and high-level mix of 19 and 20kHz tones lies at 54dB (0.2%), the higher-order products are relatively low in level for a tube amplifier (fig.14).
The Audio Research Reference 75 measures well for a classic tube amplifier design with a single pair of output tubes for each channel and a modest degree of loop negative feedback. Its output transformers are also of good quality, the only proviso being that the amplifier should not be used with loudspeakers whose impedance drops significantly below the nominal value of the output transformer tap.John Atkinson