It must also be taken into account that the distortion content is almost entirely pure second harmonic, even at high levels of a low-frequency tone into a high impedance (fig.4). And while the third harmonic rises almost to -80dB (0.01%) into the punishing 1k ohm load (fig.5), the distortion spectrum still remains predominantly second harmonic.

Fig.4 Conrad-Johnson Premier 17LS, spectrum of 50Hz sinewave, DC-1kHz, at 2V into 100k ohms (linear frequency scale).

Fig.5 Conrad-Johnson Premier 17LS, spectrum of 50Hz sinewave, DC-1kHz, at 2V into 1k ohms (linear frequency scale).

Fig.6 confirms that this picture doesn't change at higher frequencies. Note that while the 120Hz rectified power-supply artifact remains below -96dB (0.0015%) in these graphs, the third harmonic of the powerline frequency at 180Hz is the highest in level at -90dB, due, perhaps, to magnetic coupling from the power-supply transformer. Not visible in these graphs is the presence of very-low-frequency noise, which caused the oscilloscope trace to flicker slightly. Audiophiles using direct-coupled solid-state amplifiers and reflex-loaded speakers with the Premier 17LS might see some very slight pumping of the woofers.

Fig.6 Conrad-Johnson Premier 17LS, spectrum of 1kHz sinewave, DC$20kHz, at 2V into 100k ohms (linear frequency scale).

Finally, the slightly bent transfer function, which gives rise to the subjectively benign second-harmonic distortion, also results in moderate levels of intermodulation distortion (fig.7); at this admittedly high output level, the 1kHz difference component reaches -54dB (0.2%). Into the 1k ohm "torture" load, this rises to -52dB (fig.8), though higher-order intermodulation products remain acceptably low.

Fig.7 Conrad-Johnson Premier 17LS, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 5V into 100k ohms (linear frequency scale).

Fig.8 Conrad-Johnson Premier 17LS, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 3V into 1k ohm (linear frequency scale).

Other than its relatively disappointing channel separation, the Premier 17LS's measured defects—volume-control-dependent bandwidth and signal-level-dependent distortion—reach significant levels only under conditions that will be extremely unlikely in normal use. This is sensible engineering, in my opinion.—John Atkinson