Audio Research Reference 110 power amplifier Measurements
Before performing any measurements on the Audio Research Reference 110 amplifier, I checked the bias of each pair of tubes, something made easy by the convenient probe sockets adjacent to the tubes on each of the circuit boards. All bias readings were within 1mV of the target 63mV, and the "Check" readings were all in the center of the specified range.
The specified input impedance is very high, at 300k ohms, and my estimate was in this region. (Measuring high input impedances is not very precise with our usual voltage-drop method, because of the very small voltage changes involved.) The balanced inputs, each wired with pin 2 hot, preserve absolute polarity; ie, are non-inverting. The voltage gain from all transformer taps was lower than average; in addition, the channels didn't match very well, the left channel consistently offering a higher level than the right. With a 1kHz tone, the 4 ohm tap offered 21.2dB gain into 8 ohms, left, and 20.7dB, right; the 8 ohm tap, 23.9dB and 23.3dB, respectively; the 16 ohm tap, 26.4dB and 25.7dB, respectively.
As expected, the Reference 110's output impedance was significantly higher than that of a typical solid-state amplifier, and increased both at very high frequencies and with the nominal value of the output-transformer tap. The 4 ohm tap offered the lowest impedance, at 0.5 ohm at low and midrange frequencies, rising to 0.8 ohm at 20kHz. The 8 ohm tap's impedance was 0.87 ohm at low frequencies, 1.45 ohms at 20kHz; the 16 ohm tap, 1.43 ohms at low frequencies, 1.85 ohms at 20kHz. As a result, there will be a significant modification of the amplifier's frequency response due to the Ohm's Law interaction between its source impedance and the impedance modulus of the loudspeaker. With our standard simulated loudspeaker, the response variations from the 8 ohm tap, for example, reached ±0.9dB (fig.1, top trace at 2kHz), which will be audible. These variations were a little greater from the 16 ohm tap, a little smaller from the 4 ohm tap.
Fig.1 Audio Research Reference 110, 8 ohm tap, frequency response at 2.83V into (from top to bottom at 2kHz): simulated loudspeaker load, 8, 4, 2 ohms (1dB/vertical div., right channel dashed).
The channel mismatch can also be seen in this graph, with the right-channel response into 8 ohms (top dotted trace) almost overlying that of the left channel into 4 ohms (bottom solid trace). But note that though the response does droop a little in the treble, presumably due to the increasing source impedance, the amplifier's small-signal bandwidth is actually quite wide, with the left channel's output not reaching –3dB until 95kHz. As a result, a 10kHz squarewave was reproduced with short risetimes and a good square shape (fig.2). This graph was taken with the output-transformer tap nominally matched to the load; in this case, the 8 ohm tap driving 8 ohms. An overshoot develops when the load impedance is higher than the nominal tap value (fig.3), but this is both slight in degree and free from ringing.
Fig.2 Audio Research Reference 110, 8 ohm tap, small-signal 10kHz squarewave into 8 ohms.
Fig.3 Audio Research Reference 110, 4 ohm tap, small-signal 10kHz squarewave into 8 ohms.
Given the Reference 110's dual-mono construction, with each channel's audio circuitry carried on its own board, I was surprised to find that the channel separation at 1kHz was 85dB (L–R) and 96dB (R–L), decreasing by another 10dB at 50kHz. This is still very good, however. The wideband, unweighted signal/noise ratio (ref. 2.83V into 8 ohms, 8 ohm tap) was good rather than great at 79.5dB. Switching an A-weighting filter into circuit improved the figure to 97.8dB, suggesting that it is noise at the frequency extremes that is affecting the unweighted result.
Figs. 4, 5, and 6 plot how the THD+noise percentage in the left channel changes with increasing power into loads ranging from 4 to 16 ohms, from the 4, 8, and 16 ohm output-transformer taps, respectively. General points to note are: 1) the distortion is very low from all taps at levels of 1W or below; 2) the distortion doubles with each halving of the load impedance (ie, with each doubling of the output current); 3) the distortion rises linearly with increasing output power, suggesting a low overall level of loop negative feedback; 4) the amplifier gives its maximum power when the load impedance ranges from equal to half the nominal transformer tap value; and 5) under those conditions the amplifier meets its rated power of 120Wpc. The highest clipping powers measured (1% THD) were 130W, 16 ohm tap into 16 ohms (24.15dBW); 120W, 8 ohm tap into 8 ohms (20.8dBW); and 125W, 4 ohm tap into 2 ohms (15dBW).
Fig.4 Audio Research Reference 110, 4 ohm tap, distortion (%) vs 1kHz continuous output power into (from bottom to top at 10W): 16, 8, 4, 2 ohms.
Fig.5 Audio Research Reference 110, 8 ohm tap, distortion (%) vs 1kHz continuous output power into (from bottom to top at 10W): 16, 8, 4, 2 ohms.
Fig.6 Audio Research Reference 110, 16 ohm tap, distortion (%) vs 1kHz continuous output power into (from bottom to top at 10W): 16, 8, 4, 2 ohms.
These graphs indicate that the distortion products start to rise out of the noise floor at a level of 2.83V, so I plotted how the THD+N percentage changes with frequency at that level. The results for the 4, 8, and 16 ohm taps are shown in figs. 7, 8, and 9, respectively. General points to note are: 1) again, increases in output current from each tap result in corresponding increases in THD; 2) the distortion is very low in level when the load impedance is equal to or twice the output tap; 3) the distortion rises both at infrasonic and ultrasonic frequencies; and 4) the right channel is as good as the left at midrange frequencies from both the 4 and 8 ohm taps, but worse at high and low frequencies, and is also worse overall from the 16 ohm tap.
Fig.7 Audio Research Reference 110, 4 ohm tap, THD+N (%) vs frequency at 2.83V into (from bottom to top): 16, 8, 4, 2 ohms (right channel dashed).
Fig.8 Audio Research Reference 110, 8 ohm tap, THD+N (%) vs frequency at 2.83V into (from bottom to top): 16, 8, 4, 2 ohms (right channel dashed).
Fig.9 Audio Research Reference 110, 16 ohm tap, THD+N (%) vs frequency at 2.83V into (from bottom to top): 16, 8, 4, 2 ohms (right channel dashed).
The spectral content of the distortion was predominantly the subjectively innocuous second harmonic at low output currents and at midrange frequencies (fig.10), joined by the third harmonic at higher currents and lower frequencies (fig.11). The decrease in circuit linearity at high frequencies seen in figs. 7–9 and at high powers seen in figs. 4–6 results in somewhat disappointing performance on the demanding high-frequency intermodulation test, when the amplifier under test is asked to drive an equal mix of 19 and 20kHz tones at a level close to visible clipping on an oscilloscope screen. With the 4 ohm tap driving 8 ohms, the 1kHz difference component lay at 0.2% (–54dB), which will probably be okay subjectively (fig.12).
Fig.10 Audio Research Reference 110, 4 ohm tap, 1kHz waveform at 1.5W into 8 ohms (top), 0.039% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).
Fig.11 Audio Research Reference 110, 8 ohm tap, spectrum of 50Hz sinewave, DC–1kHz, at 70W into 8 ohms (linear frequency scale).
Fig.12 Audio Research Reference 110, 4 ohm tap, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 55W peak into 8 ohms (linear frequency scale).
As I have come to expect from Audio Research products, the Reference 110's measured performance is respectable, especially considering the low level of loop negative feedback, though I was bothered by the level mismatch between the channels and the differences in linearity.—John Atkinson