Tube Preamp Reviews

I measured the PrimaLuna DiaLogue Three using the Audio Precision SYS2722 system (see www.ap.com and the January 2008 "As We See It"), as well as my vintage Audio Precision System One. The maximum gain at 1kHz into 100k ohms was lower than the specified 12dB, at 9.67dB in the left channel and 9.9dB in the right. The preamp preserved absolute polarity (ie, was non-inverting), and the input impedance was very high, at >150k ohms at all frequencies. (I can't be more precise because the voltage-drop method that I use to measure input impedance becomes increasingly inaccurate for impedances above 100k ohms or so.)

The DiaLogue Three's output impedance is specified as a high 2500 ohms; I measured 2400 ohms at high and middle frequencies but 4000 ohms at 20Hz, which will be due to the limited size of the output coupling capacitor. (A coupling cap needs to have a high value, but the desired plastic-film types are also physically large.) As a result, the frequency response into the punishingly low 600 ohm impedance will roll off prematurely at low frequencies (fig.1, cyan and magenta traces), reaching –3dB at 28Hz. Into 100k ohms (fig.1, blue and red traces), the response is flat to below 20Hz, though the ultrasonic response is curtailed compared with the condition of low load impedance. At –0.25dB at 20kHz, however, this will have no audible consequences. This graph was taken with the volume control set to its maximum, and a 0.23dB channel imbalance can be seen. Repeating the measurement with the control set to unity gain (2:00) gave the same imbalance, which suggests that it is due to an intrinsic difference in gain between the two channels rather than to volume-control mistracking.

The PrimaLuna's low-frequency channel separation was moderately good, at 90dB in both directions at 100Hz, though the separation decreased to 45dB at the top of the audioband. The unweighted, audioband noise levels, measured with the input short-circuited but the volume control set to its maximum, were low, with a signal/noise ratio of 87.5dB in the right channel but 80dB in the left. Switching an A-weighting filter into circuit increased both ratios to 97dB.

The DiaLogue Three appears to use only a limited amount of negative feedback, as the plot of its THD+noise percentage against output voltage reveals a steady increase of the former into all loads (fig.2). While the low-level THD+N is low into the higher impedances, dropping below 0.1% below 330mV or so, this graph indicates that loads below about 3k ohms are best avoided with this preamp. Even so, the THD reaches 0.58% at 2V into 100k ohms, about the highest level the preamplifier will be required to deliver in practice. The small-signal THD+N percentage remains constant with frequency (fig.3).

Fortunately, the spectrum of the distortion is heavily second-harmonic in nature (fig.4). This graph also shows the slightly higher level of random noise in the left channel (blue trace), and the highest power-supply component is the third harmonic at 180Hz, which lies at –97dB left and –100dB right. This is presumably due to magnetic leakage from the power transformer, but is too low in level to be audible, even when the signal level drops to 300mV (fig.5). The only disappointing aspect of the DiaLogue Three's measured behavior was with the high-frequency intermodulation test, where, even at 300mV, the equal mix of 19 and 20kHz tones resulted in a 1kHz tone at –46dB (0.5%). However, the higher-order products at 18 and 21kHz lay below –70dB (0.03%).

The PrimaLuna DiaLogue Three's measured performance is related to its low-feedback tubed circuitry; but taking that into consideration, it gets a clean bill of health.—John Atkinson