Audio Valve Conductor preamplifier Measurements
I primarily used Stereophile's loaner sample of the top-of-the-line Audio Precision SYS2722 system (see the January 2008 "As We See It" and www.ap.com) to examine the Audio Valve Conductor preamplifier's measured performance.
Using the preamplifier's impossible-to-read display and following the instructions in the manual, I set up the first pair of line-level inputs for "XLR" operation and the second for "cinch" (ie, RCA). The maximum gain for full balanced operation (ie, XLR to XLR) was 19.3dB for the left channel, 18.8dB for the right, both figures significantly higher than the specified 14dB. For unbalanced operation (RCARCA), the maximum gain was closer to specification at 13.2dB left and 12.8dB right. There is no balance control, so I must assume the channel imbalance may well have been to tube problemsaccording to the front-panel display, the Conductor had had 109 hours of use, and the right channel consistently measured less well than the left. Unless stated otherwise, my comments refer to the better-performing left channel, which I assume is more representative of the Conductor's ultimate performance.
Though it was slightly lower than the specified 47k ohms, the Conductor's unbalanced input impedance was fairly high, at 41k ohms at low and middle frequencies, dropping slightly to 31k ohms at 20kHz. The balanced figures were twice the unbalanced figures, as expected. The balanced output impedance was very low at high and middle frequencies, at close to 100 ohms. However, it did rise to 2900 ohms at 20Hz, presumably due to the finite size of the output coupling capacitors. The unbalanced output impedance was significantly higher than the balanced, at 1600 ohms at 1kHz. However, while this figure did rise at the frequency extremes, it was only to 1800 ohms or so, meaning that the response will not be affected when driving an amplifier with a low input impedance.
Although not marked as such and not mentioned in the skimpy manual, the RCA jacks that appear to be Input 7 are actually the Tape Output jacks. (While these RCAs are accompanied by XLRs, these are the wrong gender for outputs.) The Tape Out RCAs pass through the input signal at unity gain for unbalanced inputs, 6dB for balanced inputs, and are not affected by the volume control. The source impedance is a low 84 ohms across the audioband, implying that they are actively buffered.
The Audio Valve's increase in balanced output impedance at low frequencies is not unusual for tubed designs, but it meant that the frequency response rolled off prematurely in the bass when the preamplifier was tested into the very low 600 ohms impedance (fig.1, bottom pair of traces below 1kHz). As long as the Conductor is used with power amplifiers having an input impedance of 30k ohms or more, its bass extension will be fine. At the other end of the spectrum, fig.1 indicates that the Conductor's frequency response extends very high, being just 1dB down at 150kHz. However, this graph was taken with the volume control set to unity gain and an input of 1V at 1kHz; with the volume control set to its maximum and the input voltage reduced to give the same 1V output, the bandwidth decreased somewhat, to 1dB at 80kHz. There was an insignificant change in output at 20kHz and below, however, so the dependence of the preamplifier's bandwidth on its volume-control setting should be irrelevant to its sound quality.
Fig.1 Audio Valve Conductor, balanced frequency response at 1V into 100k ohms (top two traces) and 600 ohms (bottom two traces) with volume control set to unity gain (left channel blue, right red). (1dB/vertical div.)
The Conductor's channel separation with the volume control at its maximum was good rather than great, at 80dB at 1kHz. This decreased to 67dB at 20kHz, most likely due to capacitive coupling somewhere in the circuit, probably at the volume control. The Audio Valve preamp was also very quiet. With its input short-circuited but its volume control set to its maximum, the unweighted, wideband signal/noise ratio was a fine 86.8dB ref. 1V output, increasing to 95.5dB when A-weighted.
Fig.2 shows how the percentage of THD+noise in the preamplifier's output changes with the output level of a 1kHz tone into 100k ohms (bottom trace) and 600 ohms. The amplifier doesn't clip (defined as 1% THD+N) until a very high voltage, even into the demanding 600 ohm load, where it delivers 11V. And while the actual distortion with 600 ohms can be seen to begin rising out of the noise floor above 300mV output, it remains below 0.1% at all practical levels the Conductor will need to output with real-world power amplifiers. Into 100k ohms, the distortion doesn't climb above the noise until 1V output, and is just 0.005% at its minimum.
Fig.2 Audio Valve Conductor, THD+N (%) at constant 1V output level vs 1kHz balanced input level.
While performing this last measurement, I noticed something unusual: I could change the shape of the traces in the graph by adjusting the volume control. Fig.2 was taken with the volume control at its maximum. But if I backed off the volume control and increased the input level to give the same output voltage, I got higher minimum distortion. This suggests that something is adding distortion upstream from the Conductor's output stage. I therefore measured the distortion in the preamp's output at 1V output, increasing the input voltage but each time backing off the volume control to keep the output level constant. The results are shown in fig.3. The percentage of distortion increases in a linear manner with input voltage, and the preamp clips when the input voltage is 2.6V and the volume control is set to 8.3dB, a not unusual setting. And even at 2V, the standard output for a CD player, the THD+N reading is 0.58%. If our sample of the Conductor was not brokenand I have no other reason to suspect that it was (I did reseat all the tubes before starting the measurements)it looks as if the input stage ahead of the volume control is overloading prematurely.
Fig.3 Audio Valve Conductor, THD+N (%)vs 1kHz balanced output level into (from bottom to top above 1V): 100k, 600 ohms.
This disappointing behavior can also be seen in fig.4, which plots THD+N against frequency at 1V output into 100k ohms and 600 ohms but with two input levels: 200mV (bottom four traces) and 1.2V (top four traces). As mentioned earlier, the right channel (red, gray, and magenta traces) is significantly worse than the left (blue, cyan, green), but the preamp actually performs quite well into the very low impedance. Unfortunately, the higher input level results in more than 10 times the level of THD in the output compared with the lower input level.
Fig.4 Audio Valve Conductor, balanced THD+N (%)vs frequency at 2V into: 100k ohms and 600 ohms with 1.2V input level ((top) left channel blue, right red); and 100k ohms and 600 ohms with 200mV input level (bottom, left channel green, cyan; right gray, magenta).
Peculiarly, the distortion of the signal present at the Tape Out jacks was not affected by the level of the input signal, other than having an increasing proportion of noise as the level was reduced. This suggests the Tape Out signal is taken from before the overload-prone input circuit.
I further investigated this behavior by looking at the spectra of the Conductor's output under various conditions. Fig.5, for example, shows the spectrum of its output while it drives a 1kHz tone at 2V into 100k ohms, about the highest level it will be called on to deliver in practice. The input level was 800mV, which is not unreasonable. Both the second and third harmonics in the right channel lie at 60dB (0.1%), and higher harmonics can also be seen. While some low-frequency, power-supplyrelated spikes are evident, these all lie at or below 120dB and will therefore be irrelevant. (Their levels were not affected by experimenting with the system grounding.) For comparison, fig.6 was taken with the input level reduced to 220mV and the volume control adjusted to give the same 2V output into 100k ohms as before. The second and third harmonics have now dropped by 10 and 20dB, respectively, in both channels, and while some higher harmonics can still be seen, these are almost exclusively in the right channel (red trace).
Fig.5 Audio Valve Conductor, spectrum of 1kHz sinewave, DC10kHz, at 2V balanced into 100k ohms and 800mV input level (linear frequency scale; left channel blue, right red).
Fig.6 Audio Valve Conductor, spectrum of 1kHz sinewave, DC10kHz, at 2V balanced into 100k ohms and 220mV input level (linear frequency scale; left channel blue, right red).
Confirming that the Conductor's output stage copes well with low impedances, fig.7 shows the spectrum of its output taken under the same circumstances as fig.6: ie, a low input level. While the levels of the harmonics have risen with the increased demand for current from the output stage, they still remain below 60dB.
Fig.7 Audio Valve Conductor, spectrum of 1kHz sinewave, DC10kHz, at 2V balanced into 600 ohms and 260mV input level (linear frequency scale; left channel blue, right red).
Finally, fig.8 shows the Audio Valve's behavior on the punishing high-frequency intermodulation test under worst-case conditions: an input level of 2V typical of CD players, and the volume control set to unity gain. A large number of intermodulation products are visible, with the 1kHz difference component lying at 54dB (0.2%), and the higher-order components at 18 and 21kHz at 46dB (0.5%). Not good.
Fig.8 Audio Valve Conductor, HF intermodulation spectrum, DC24kHz, 19+20kHz at 2V peak balanced into 100k ohms (linear frequency scale).
While the Audio Valve Conductor is an impressive-looking product, has very low noise, and its output stage seems capable of driving low impedances without breaking too much of a sweat, its measured performance is compromised by its inability to handle high-level sources in a linear manner. Why wasn't Bob Reina bothered by this behavior? I suspect that it is the fact that the increased distortion comprises the lower harmonics and that they increase linearly with input level. So with a CD player having the usual 2V maximum output, with classical and jazz, the average signal level will stay below 500mV almost all the time, meaning that the Conductor's input will not be audibly overloading. It will progressively overload for the top 12dB of the music's peaks, but with the distortion signature consisting of the second and third harmonics, the perceived effect will be more of a "fattening" of those peaks rather than distortion as such.
Nevertheless, and again assuming that our review sample was not broken, I was disappointed by this expensive preamplifier's measured performance.John Atkinson