Tube Power Amp Reviews

I performed full sets of measurements on both samples of this retro-styled amplifier, using Audio Precision's SYS2722 system (see www.ap.com and the January 2008 "As We See It"). As instructed in the manual, after inserting the tubes and running each amplifier for a while to make sure everything was at the optimal operating temperature, and with no signal applied, I rotated the rear-panel knob of each monoblock to minimize measured hum in the output. This was a straightforward operation and, as you can see from the signal/noise measurements later in this report, an effective one.

The output-transformer taps were left in the 8 ohm setting used by Art Dudley in his auditioning, and I began measuring the amplifiers after they'd been running at a low level into 8 ohms for an hour. There I ran into trouble—although I'd understood from AD that Reinhard Thöress required measurement with the analyzer inputs floating with respect to ground, the 300Bs were unstable when so measured, delivering no more than 50mW into 8 ohms. The fix was to connect the amplifiers' black output terminals to the analyzer's chassis ground—a peculiar solution, given that loudspeakers are themselves electrically isolated from ground. Nonetheless, from then on, measuring the Thöress 300Bs was straightforward.

As the manual warns, the 300B offers very low voltage gain: just 6.85dB into 8 ohms at 1kHz. It preserved absolute polarity (ie, was non-inverting), and the input impedance was high at low and middle frequencies, at 90k ohms, dropping slightly to 62k ohms at the top of the audioband.

The output impedance was relatively low for a single-ended design, at 1.9 ohms at 20Hz and 1kHz, rising to 2.4 ohms at 20kHz, which resulted in response variations of ±1.2dB with our standard simulated loudspeaker (fig.1, gray trace). Into resistive loads, the 300B offered a superbly flat response in the audioband, with –1dB limits into 16 ohms at 10Hz and 18kHz (blue trace). The high-frequency rolloff occurred at slightly lower frequencies as the impedance dropped (magenta and red traces), but what can also be seen in this graph are two major ultrasonic resonances, both of which increase in level as the load impedance increases. This kind of behavior, common with SET designs, is related to the output transformer, but it adds a degree of overshoot and ringing to the Thöress 300B's reproduction of a 1kHz squarewave (fig.2), and obscures the shape of a 10kHz squarewave (fig.3).

With the amplifier's rear-panel potentiometer adjusted for lowest hum, the 300B's unweighted, wideband S/N ratio (ref. 1W into 8 ohms) was 74dB; reducing the measurement bandwidth to the audioband improved this to 79.2dB, and the A-weighted ratio was an impressive 95.7dB. Fig.4 reveals that the highest-level hum component is the full-wave–rectified 120Hz component at –80dB (0.01%), the 60Hz component lying at a very low –110dB ref. 1W into 8 ohms.

As expected, this is a low-powered amplifier. Figs. 5 and 6 plot the percentage of THD+noise in the amplifier's output against power into 16 and 8 ohms, respectively. The THD+N is impressively low below 30mW, especially into 16 ohms, but rises linearly with output power until the onset of waveform clipping. The Thöress 300B is specified as delivering 8W into 16 ohms (12dBW), but fig.5 reveals that, with the 8 ohm connection used by AD, the amplifier reaches our normal definition of clipping (1% THD) at 6W (10.8dBW). Even with a relaxed definition of clipping of 3% THD, it delivers 7W into 16 ohms rather than 8W. Into 8 ohms (fig.6), however, while its 1% THD power is just 3.5W (5.45dBW), the 300B achieves a power output of 10W (10dBW) at 3% THD.

Because of the 300B's low maximum power, I measured how its THD+N varied with frequency at 2V rather than 2.83V (equivalent to 500mW into 8 ohms rather than 1W). The results are shown in fig.7. The distortion remains uniform with frequency until the top octave, where the measurement is affected by the ultrasonic resonances mentioned earlier. As expected, the THD+N rises with decreasing impedance, which strongly suggests that the speakers used with the Thöress monoblocks should have an impedance that doesn't drop significantly below the nominal value of the amplifier's output-transformer tap. However, into all three impedances in this graph, there is a spike of distortion in the midrange. I double-checked the measurement, and this behavior was consistent; I have no idea what it is due to, though it's fair to note that AD noticed no coloration that might have resulted from this behavior.

This is probably because the Thöress amplifier's distortion signature is heavily second-harmonic in nature (figs. 8 and 9), which is not unmusical (see "Listening," March 2015). Of course, the perceived nature of the harmonic distortion will also depend on the amount of intermodulation distortion generated by the same nonlinear transfer characteristic, and in this respect the Thöress 300B can be rated as only fair. An equal mix of 19 and 20kHz tones at a peak level of 500mW into 8 ohms generated a 1kHz difference component at –43dB (0.7%), though the higher-order products at 18 and 21kHz were more than 20dB lower in level (fig.10).

Overall, the Thöress 300B monoblock measures well for a SET amp. As long as it is not used with low-impedance speakers, and the speakers are sufficiently sensitive not to be bothered by its low maximum output power, it can be recommended.—John Atkinson