Tube Preamp Reviews

I measured the Balanced Audio Technology Rex II's electrical performance with my Audio Precision SYS2722 system (see the January 2008 "As We See It") with the 6C45 AC shunt-regulator tubes. BAT warns in the Rex II's manual not to stack the two chassis; the supplied umbilical cables were just long enough for me to put the power chassis on the floor and the control chassis on my test bench. One thing that should also be noted by prospective purchasers is that both of the Rex II's chassis need plenty of ventilation. After a few hours' running, the top panel of the control chassis was hot, measuring 113.2°F (45.1°C).

All the testing was done in balanced mode, ie, balanced input to balanced output. The maximum gain (with the volume control set to "140") was 20.3dB, with the unity gain setting "103." The preamp preserved absolute polarity, ie, was non-inverting, with the front-panel pushbutton set to Normal, and the input impedance was very high, at >500k ohms at low and middle frequencies and still 360k ohms at the top of the audio band. The output impedance, specified as 200 ohms, measured around 700 ohms at 20Hz and 1kHz at full volume, rising to 1770 ohms at 20kHz. With the volume control at "103," the impedance was 700 ohms across the band.

With the volume control set to its maximum, the ultrasonic frequency responses into a high-impedance load differed between the channels. The blue trace in fig.1 shows that the left channel starts to roll off above 10kHz, reaching –1.6dB at 20kHz and –6dB at 58kHz. By contrast, the right channel (red trace) was down by just 0.5dB at 20kHz and –6dB at 90kHz. Into 600 ohms (cyan and magenta traces), the increasing output impedance at high frequencies means that both channels roll off prematurely. Users are not going to operate the Rex II with its volume control set to its maximum, so I repeated the high-impedance measurement with the preamp set to unity gain. The result is shown in fig.2, plotted with greater vertical resolution than fig.1. The audio-band response is now perfectly flat from 40Hz to 30kHz with excellent matching between the channels. However, a small rise in the preamp's ultrasonic output can be seen centered on 110kHz.

Channel separation at 1kHz was good, at 82dB R–L and 90dB L–R, though it worsened to 72dB, R–L, and 77dB, L–R, at 20kHz. The unweighted, wideband S/N ratio, ref.1V and measured with the input shorted to ground but the volume control set to "140," was 58.7dB (average of both channels). This ratio improved to 74.9dB when the measurement was restricted to the audio band, and 78.7dB when A-weighted. Unusually, the S/N ratio didn't improve significantly at lower settings of the volume control. Fig.3 shows that there are some low-level AC supply spuriae present, but that the primary noise contributor is random.

Plotting the THD+N percentage against output voltage into 100k ohms (fig.4) reveals that the distortion starts to rise above the noise floor between 2V and 3V, suggesting a sensibly arranged gain architecture. However, that distortion is low, at 0.02%, and rises gently with voltage. We define clipping as when there is 1% THD+N present, and into this high impedance, the Rex II doesn't clip at 1kHz until the output is an extraordinary 70V! Even into 600 ohms, the Rex II doesn't clip until 10V (not shown).

Fig.5 shows how the THD+N varies with frequency at a level, 3V, where I could be sure I was looking at actual distortion rather than noise. The left channel (blue trace) has slightly lower distortion that the right channel (red) between 100Hz and 1kHz, but both channels feature increasing distortion below 100Hz, presumably due to the onset of core saturation in the output transformers. The cyan and magenta traces in this graph confirm that this preamplifier should not be used into pathologically low impedances. It will work well with power amplifiers having an input impedance of 10k ohms or greater.

At low frequencies and a moderate output level into 100k ohms, the distortion signature is primarily the third harmonic, then the second harmonic (fig.6). Increasing the signal frequency to 1kHz (fig.7) virtually eliminates the third harmonic and reduces the level of the second harmonic by a factor of 10, from –73dB to –94dB (0.002%). Intermodulation distortion (fig.8) is also very low.

Overall, the measured performance of the BAT Rex II is excellent, though dominated by the characteristics of the T-Rex output transformers.—John Atkinson