Tube Power Amp Reviews

To measure one of the Lamm ML2.2 amplifiers (serial no.B10254), I used Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see www.ap.com and the January 2008 "As We See It"). Before performing any measurements, I ran the amplifier with a 1kHz tone at 1W into 8 ohms for an hour, then turned off the signal generator and adjusted the output-stage plate voltage and current, following the detailed instructions in the manual and using the recommended Fluke 87 meter. Lamm recommends that the wall voltage be 120V for this procedure; unfortunately, the voltage in my quiet suburban neighborhood remained higher than that, at 124.4V. After adjustment, the plate voltage for the measurements was 174.4V DC and the plate current voltage was 0.305V, both figures within the recommended ranges.

The signal from all three output transformer taps was non-inverting. The input impedance was 42k ohms at 20Hz, dropping slightly to 37k ohms at 1kHz, then again to 22k ohms at 20kHz. The voltage gain depended on the tap. It was 20dB from the 4 ohm tap, 22.7dB from the 8 ohm tap, and 25.1dB from the 16 ohm tap, all of these lower than usual for a power amplifier. The output impedance also depended on the transformer tap used. From the 4 ohm tap, the impedance was a low 0.38 ohm at low and middle frequencies, rising to 0.5 ohm at 20kHz. As expected it was higher from the 8 ohm tap, at 0.7 ohm at 20Hz and 1kHz, and 0.8 ohm at 20kHz, and higher still from the 16 ohm tap: 1.35 ohms at all audio frequencies.

These figures are all commendably low for a tube design with a single-ended output stage. As a result, the variation in the ML2.2's frequency response due to the Ohm's Law interaction between the source impedance and the manner in which the impedance of our standard simulated loudspeaker changes with frequency was relatively small. Even from the 16 ohm tap (fig.1, gray trace), the variation was only ±0.9dB, while from the 4 ohm tap (fig.2, gray trace) it was just ±0.3dB. The graphs also reveal that the ML2.2 has a very wide small-signal bandwidth, with the output into 16 ohms from all output taps being flat to below 20Hz and to the top of the audioband, and –3dB above 100kHz. As a result, the ML2.2's reproduction of a 10kHz squarewave (fig.3) featured very short risetimes, though a small amount of overshoot and ultrasonic ringing can be seen on the leading edges.

I measure an amplifier's signal/noise ratio (SNR) with the input shorted. The ML2.2's SNR varied with the output tap chosen, in general decreasing by the amount of extra voltage gain available. The wideband, unweighted ratio, ref. 2.83V into 8 ohms, was 77.3dB from the 16 ohm tap, 79.5dB from the 8 ohm tap, and 82.1dB from the 4 ohm tap. These ratios all improved by 9dB when A-weighted, and, as can be seen in fig.4, the noise primarily comprises components at 60Hz and its even and odd harmonics, the last due to magnetic interference from the AC transformer. Overall, however, these spuriae are all relatively low in level for a tube design.

Figs. 5, 6, and 7 were all taken from the 8 ohm output transformer tap, and show how the THD+noise percentage in the amplifier's output changes with output power into, respectively, half the nominal tap value, when the load is the same as the tap value, and into twice the tap value. The curves from the other two taps were very similar, so I haven't shown them. The ML2.2 more than meets its specified output power of 18W into a matched load (12.55dBW) at 3% THD, offering 20W into 8 ohms from its 8 ohm tap (13dBW, fig.6), and around 15W into 4 ohms (9dBW, fig.5) and 16 ohms (14.8dBW, fig.7). (In part, this extra power will be due to the fact that I was measuring with a wall AC voltage of 124.4V instead of 120V.) The 1% THD powers are much less, of course, due to the fact that the single-ended output-stage topology and a small degree of global negative feedback lead to a steadily rising amount of distortion as the power increases.

Fig.8, taken from the 4 ohm tap, shows that the THD doesn't change significantly with increasing frequency, though it does increase alarmingly as the load impedance halves. (The 8 and 16 ohm taps, figs.9 & 10, respectively, gave similar pictures, but the additional gain from these taps is accompanied by increasingly high levels of distortion overall.) That THD doesn't increase significantly at very low frequencies is a testament to the quality of the output transformer. Fortunately, the distortion is almost pure second harmonic (fig.11), which tends to be subjectively benign. Even at low frequencies, where the output transformer's core might start to saturate, the third harmonic is very much lower than the second (fig.12), which again suggests that the ML2.2's output transformer is quite special.

The ML2.2 offers fairly low levels of second-harmonic distortion at powers below a few hundred milliwatts, which will be the power levels asked for by AD's sensitive loudspeakers. Whether or not second-harmonic distortion is subjectively innocuous will depend on whether or not the amplifier also develops high levels of intermodulation distortion. As its circuit's nonlinearity doesn't increase as the frequency increases, the ML2.2 scores relatively well here. Fig.13 shows the spectrum of the amplifier's output as it reproduced from its 4 ohm tap an equal mix of 19 and 20kHz tones with a peak output of 1W into 8 ohms. While the 1kHz second-order or difference component lies at –54dB (0.2%), any higher-order intermodulation components all lie below –80dB (0.01%). Even when the power was increased to just below clipping into 8 ohms (fig.14), the higher-order spuriae still lie at –70dB (0.03%), though the difference component has risen to –44dB (0.6%).

Whenever I measure one of Vladimir Lamm's amplifiers, I am always impressed by the quality of the engineering. Yes, the ML2.2 has a bent transfer function, which means that it produces higher-than-usual levels of second-harmonic distortion—but this is not accompanied by high levels of high-order intermodulation. And you have that low output impedance and very wide bandwidth!—John Atkinson