Tube Preamp Reviews

I measured Balanced Audio Technology's Rex using Audio Precision's SYS2722 top-of-the-line test set-up. I used both pairs of current-source tubes —the 6C45 that MF liked and the 6H30 that he didn't —but a problem developed with the left channel soon after the preamplifier landed on my test bench. When the Rex was first powered up, the gains and responses in the two channels were very closely matched, but after five minutes or so the level of the left channel gradually dropped until it was about 20dB lower than the right, with the signal being very distorted. I turned the Rex off, took off the two chassis covers, and reseated all the tubes. The ribbon cable in the control chassis that connects the left-channel input/output board to the main signal-handling board was a little loose, so I reconnected that and turned the preamp back on. All was well for five minutes, but then the left-channel gain again dropped by 20dB. Obviously, something was seriously wrong with that channel, but as the right channel seemed to be working fine and has its own power supply, I continued the testing with that channel alone.

The maximum voltage gain was slightly higher than the published specification, at 18.7dB, but the volume control operated in 0.5dB steps, as specified. The unity-gain setting of the volume control was an indicated "106" out of 140. The preamp preserved absolute polarity (ie, was non-inverting) with the front-panel LED off. The input impedance was very high, at 470k ohms or more at low and midrange frequencies, dropping to 163k ohms at 20kHz. (These figures are approximate because of the large margin of error when measuring AC impedances this high.) The Rex will therefore not load down source components at all. However, its output impedance was a little higher than specified, at 1000 ohms at 20kHz, 415 ohms at 1kHz, and 4.8k ohms at 20Hz. The last is due to the necessarily finite size of the output coupling capacitors. As MF noted, it should not be a significant issue in listening to music as long as the power amplifier has an input impedance greater than about 50k ohms. (BAT's own power amplifiers have input impedances well above 100k ohms.)

Fig.1 shows the Rex's frequency response at 1V into 200k ohms with its volume control at "140." The output is down 2dB at the top of the audioband, with a well-controlled rolloff above that frequency. This measurement was taken with the 6H30 current sources before the left channel went bad; you can see that the channels are very well matched in both level and response, and there was no difference in behavior when I switched to the 6C45 current-source tubes (not shown).

Fig.1 Balanced Audio Technology Rex, 6H30 current-source tubes, frequency response at 1V into 200k ohms with volume control set to maximum (2dB/vertical div.; left channel, blue; right channel, red).

However, it's very unlikely that the Rex will be used with its volume control at the maximum setting, so I repeated the test with it set to unity gain, again at 1V output level. The red trace in fig.2 shows the right channel's behavior into the high 200k ohms loading. While still flat within the audioband, the response now starts to rise a little above 50kHz, reaching +2.8dB at 200kHz. Reducing the load impedance to 600 ohms produced the blue trace in fig.2. The ultrasonic output is now –5.8dB at 200kHz, implying that using a power amplifier with an input impedance of 30 –50k ohms will give a basically flat ultrasonic response from the Rex with its volume control set to unity gain. In any case, this behavior will probably have no subjective consequences. At the other end of the spectrum, the Rex's output into high impedances is flat down to the 10Hz limit of the graph (red trace). But into the unrealistically low impedance of 600 ohms (blue), the bass does roll off early, due to the rise in output impedance.

Fig.2 Balanced Audio Technology Rex, 6C45 current-source tubes, frequency response at 1V into 200k ohms (red) and 600 ohms (blue) with volume control set to unity gain (2dB/vertical div., right channel dashed).

Channel separation (not shown) was superb at >110dB below 2kHz. The signal/noise ratio, measured in the worst case —with the input shorted but the volume control at "140" —was okay at 66dB (wideband, unweighted ref. 1V output). This improved to a good 83.6dB when A-weighted.

The influence of the noise floor can be seen as the downward slope of the traces below 1V or so in fig.3, which plots the percentage of harmonic distortion and noise in the Rex's right-channel output against output voltage into 200k ohms (right) and 600 ohms (left). Even into 600 ohms, the minimum THD+noise percentage is a very respectable 0.013%, considering the Rex's zero-loop-feedback design. This minimum occurs at 800mV, which is when the distortion components start to emerge from the noise floor into this load. Into 200k ohms, the minimum THD+N occurs at 2.5V, which is around the maximum level required to drive a power amplifier to clipping. Actual clipping —which was relatively "soft" according to the oscilloscope I was monitoring during the testing —doesn't occur until very high output levels: the 1% THD point was 35V into 200k ohms, 12V into 600 ohms. But more important, the shape of the curves in this graph indicates a sensible gain architecture for the Rex, the distortion remaining buried in the background noise at all practical volume levels.

Fig.3 Balanced Audio Technology Rex, 6H30 current-source tubes, distortion (%) vs 1kHz output level into (from bottom to top at 10V): 100k, 600 ohms.

Fig.4 plots the right channel's THD+N percentage against frequency at a level (4V) where I was sure I was measuring actual distortion rather than noise. It is commendably low at all frequencies into 200k ohms (blue trace), and while it rises into the unrealistically demanding 600 ohm load (red), this output voltage level is actually above the point of inflection seen in fig.3. More important, the spectrum of the distortion (fig.5) is almost entirely the innocuous second and third harmonics. The left channel (blue) is slightly worse than the right (red); this graph was taken before the left channel went down for the second time. It was also taken with the 6H30 current-source tubes; there was no appreciable difference in the level of low-order harmonics when I changed to the 6C45 current-source tubes (fig.6). However, the fifth and seventh harmonics now make an appearance with this tube, though not to a level that will be anywhere near audible. Reducing the load impedance to 600 ohms made these higher-order harmonics disappear, though the third harmonic rises to match the level of the second (fig.7), explaining the increase in the THD percentage seen in fig.4.

Fig.4 Balanced Audio Technology Rex, 6C45 current-source tubes, distortion (%) vs frequency at 4V into: 200k ohms (blue), 600 ohms (red).

Fig.5 Balanced Audio Technology Rex, 6H30 current-source tubes, spectrum of 1kHz sinewave, DC –10kHz, at 4V into 200k ohms (linear frequency scale; left channel, blue; right channel, red).

Fig.6 Balanced Audio Technology Rex, 6C45 current-source tubes, spectrum of 1kHz sinewave, DC –10kHz, at 4V into 200k ohms (linear frequency scale).

Fig.7 Balanced Audio Technology Rex, 6C45 current-source tubes, spectrum of 1kHz sinewave, DC –10kHz, at 4V into 600 ohms (linear frequency scale).

Finally, the Rex produced low levels of intermodulation distortion into 200k ohms (fig.8). Reducing the load to 600k ohms brought up the 1kHz, 18kHz, and 21kHz intermodulation products only to a still fairly low –70dB (not shown).

Fig.8 Balanced Audio Technology Rex, 6C45 current-source tubes, HF intermodulation spectrum, DC –24kHz, 19+20kHz at 3V peak into 200k ohms (linear frequency scale).

Although the failure of the left channel was disappointing and may have resulted from the less-than-white-gloves handling that UPS gives expensive audio gear, the Balanced Audio Technology Rex is a well-engineered piece of gear. However, there was no clue in its measured performance as to why Mikey preferred its sound with the 6C45 current-source tubes. For now, that will have to remain a mystery.—John Atkinson