Tube Power Amp Reviews

I measured the Audio Research Reference 150 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 installed a new set of tubes, both the 6H30 small-signal tubes and the four matched pairs of KT120 output tubes. The instructions in the manual for setting the bias current for each pair were clear: With zero input signal, each main tube was set to a voltage drop of 65mV DC, measured at its test points, with the slave tube running between 57 and 73mV at its test points. The only tricky part of this operation was that the test points were close to the tubes' hot glass envelopes, which resulted in occasional mild curses as I brushed a tube with my fingers.

The Reference 150 has three output-transformer taps, marked 16, 8, and 4 ohms, and a common ground terminal. The amplifier preserved absolute polarity from all three taps (ie, was non-inverting), with the input XLRs wired with pin 2 hot. The voltage gain into 8 ohms was 26.5dB from the 16 ohm tap, 23.9dB from the 8 ohm tap, and 21.3dB from the 4 ohm tap. The input impedance was very high at all frequencies, at approximately the specified 300k ohms. (With such a high impedance, the voltage-drop measurement technique I use has too large a margin of error for me to be any more precise.)

As expected, the Ref150's output impedance varied according to the transformer tap selected. The 16 ohm tap measured 1.4 ohms at low and middle frequencies, rising to 1.9 ohms at the top of the audioband. The figures for the 8 ohm tap were 1 and 1.4 ohms; for the 4 ohm tap, they were 0.55 and 0.87 ohm. All three taps offer quite a low source impedance for a transformer-coupled design; as a result, the modulation of the amplifier's frequency response, due to the Ohm's Law action between that impedance and that of our standard simulated loudspeaker, was relatively mild. From the 8 ohm tap (fig.1, gray trace), it was ±0.8dB; the 4 ohm tap offered ±0.4dB, the 16 ohm tap ±1dB. Fig.1 indicates that the Ref150 has a wide bandwidth, particularly into loads higher than the nominal tap value, which correlates with a well-defined 10kHz squarewave (fig.2). While this graph reveals a small but critically damped overshoot on the leading edges of the waveform, presumably due to an ultrasonic transformer resonance, no ringing is visible. Channel separation (not shown) was superb, at >100dB at 1kHz.

The Ref150 was fairly quiet, though the left channel was noisier than the right. The unweighted, wideband signal/noise ratios, ref. 2.83V into 8 ohms and taken from the 8 ohm output with the input shorted, were 70.1dB left and 82.1dB right. When A-weighted, these figures improved to 88.5 and 99.2dB, respectively. Fig.3 shows a spectral analysis of the amplifier's low-frequency noise floor, taken while it drove a 1kHz tone at 1W into 8 ohms from the 4 ohm tap. The left channel has a higher level than the right of AC-supply–related spuriae, primarily at 60, 120, 180, and 300Hz, which is presumably why its S/N ratios were not as good. What really puzzled me in this graph was the appearance in both channels of a spectral component at 100Hz, which is related neither to the signal frequency nor to the AC line frequency. (It's the difference between the signal frequency and the 15th harmonic of the AC frequency.) I checked and double-checked this measurement, with no change in the result.

All taps behaved similarly when it came to the maximum output power. Into a load twice the nominal tap value, the Ref150 clipped (defined as 1% THD) at 90W (19.6dBW, fig.4). Into the nominal tap value, it clipped at the specified 150W (21.75dBW, fig.5), but with a higher level of distortion. Into half the tap value, the amplifier clipped at 80W (13dBW, fig.6), but with even higher distortion at lower powers. It is important, therefore, to use the transformer tap that best matches your preferred loudspeaker.

Fig.7 plots the THD+noise percentage from the 4 ohm tap into loads ranging from 2 to 16 ohms at a level, 4V, where the actual distortion rises above the noise floor. The THD increases at low frequencies due to the onset of saturation in the output transformer, and in the top octave due to the decreasing open-loop gain margin in this region reducing the effectiveness of the negative feedback. But over most of the audioband, and when the load impedance is very much higher than the transformer-tap value, the Ref150 offers low distortion. The picture from the 8 and 16 ohm taps was similar (not shown), but with higher levels of distortion overall.

At low power, the content of the distortion is predominantly second-harmonic in nature (fig.8), the second harmonic being subjectively innocuous both because of its proximity to the fundamental tone and because adding a tone an octave higher than the fundamental is musically natural. At higher powers and at low frequencies, a regular series of harmonics appears (fig.9), this connected with the Ref150's relatively small amount of overall negative feedback. This, and the circuit's decreasing linearity at high frequencies, means that the Audio Research amplifier does only moderately well on the demanding high-frequency intermodulation test (fig.10), with the difference product at 1kHz lying at –54dB (0.2%).

A successful tube-amplifier design must walk a narrow path between competing aspects of performance. Audio Research's Reference 150 successfully treads that path, with the caveat that it performs at its best with loudspeakers that have impedances equal to or higher than the nominal output-tap impedance.—John Atkinson