Conrad-Johnson Premier 16LS preamplifier Measurements
Measurement-wise, Conrad-Johnson's Premier 16LS was very similar to C-J's Premier 17LS, which Brian Damkroger reviewed for Stereophile in May, in that many of its "static" parameters varied with the setting of the volume control. Fortunately, because of the very high voltage gain—a maximum of 26.6dB, due to the lack of loop negative feedback—the volume control will be used in a position where things are basically okay. Fig.1, for example, shows how the high-frequency rolloff changes from when the volume is set to its maximum, "99" (lower traces), and to the unity-gain position, "66." With the control at "99," the highs are down by a very audible 3dB at 19kHz, whereas at "66" they are just 0.5dB down at 58kHz.
Fig.1 Conrad-Johnson 16LS, frequency response (from top to bottom) at volume-control settings of "66" and "99" into 100k ohms (1dB/vertical div., right channel dashed).
Like the 17LS, the 16LS inverted signal polarity. The input impedance with the volume control set to "99" was a high 64k ohms across most of the audioband, dropping to just below 20k at 20kHz. With its extra tubes, the 16's output impedance at 1kHz was slightly lower than the 17's, at 743 ohms, and even lower—474 ohms—at 20kHz. At 20Hz, however, it rose to a high 1.8k ohms, which suggests that the Premier 16LS should be used with power amplifiers with input impedances greater than 10k ohms, if the bass is not to start sounding lean. This is shown graphically in fig.2, which plots the C-J's response into loads varying from a high 10k ohms, where the infrabass is down just 0.5dB at 12Hz, to 600 ohms, where it is 3dB down at 30Hz. (Also shown in this graph is the fact that the ultrasonic response is more extended into the lower loads, due to the lower source impedance at high frequencies.)
Fig.2 Conrad-Johnson 16LS, left-channel frequency response (from top to bottom at 20Hz, bottom to top at 20kHz) at volume-control setting of "66" into 10k, 3k, 1k, and 600 ohms (0.5dB/vertical div.).
The Premier's A-weighted signal/noise ratio (ref. 1V output with the volume control at "99" and the inputs shorted) was an excellent 82dB, though this worsened to 54dB with an unweighted measurement covering a 10Hz-500kHz bandwidth, due to the presence of both ultrasonic and infrasonic noise. The channel separation was good below 1kHz, and showed some dependence on the volume-control setting, particularly in the R-L direction. This can be seen in fig.3, where the L-R leakage (solid trace) varies from -90dB with the control at "66" to -49dB with it at "99." There is much less variation in the other direction. It should also be noted that the linear increase of crosstalk with frequency in this graph is due to the usual capacitive coupling between channels.
Fig.3 Conrad-Johnson 16LS, channel separation (from top to bottom) with volume control at "99" and "66" (10dB/vertical div., R-L dashed).
Fig.4 plots the percentage of THD+noise present in the 16LS's output against output voltage, into three loads: 100k, 10k, and 1k ohms. Below about 900mV into the two higher loads the reading is dominated by noise, revealed by the upward slant of the traces with decreasing voltage. (The noise remains constant, and so steadily becomes an increasingly larger proportion of the measured percentage as the signal decreases.) Above 1V, the reading is dominated by true distortion. The oscilloscope display indicated that this was almost pure second harmonic, resulting from the parallel triodes' slightly "bent" transfer function. The THD increases smoothly with increasing output voltage up to 15V or so into these two higher loads. The kink in the traces at this level is due to the negative-going half of the sinewave starting to round off. ("Clip" is too harsh a word for what appears to be relatively gentle change.)
Fig.4 Conrad-Johnson 16LS, THD+N (%) vs output level (V) into (from bottom to top at 1V): 100k, 10k, and 1k ohms.
The minimum distortion percentage is just above 0.2% into 100k ohms, 0.24% into 10k ohms—a little higher than the 17LS's—and remains constant with frequency across the audioband (not shown). However, into the 1k load (top trace in fig.4), the minimum distortion is 0.4% at a low 350mV output, and the proportion reaches a quite audible 1% at a low 1.3V. As with the output impedance discussion above, amplifier loads of less than 10k ohms are contraindicated for the 16LS.
At 2.2V—about the highest output level the Premier 16LS is going to be asked to deliver—the distortion into 100k ohms rises to around 0.3%. But, as shown in figs. 5 and 6, this is still pure second harmonic, which is subjectively benign at low levels and even preferable at higher levels, provided it isn't accompanied by significant intermodulation distortion. Fig.7, taken at the same overall level into 100k ohms as fig.5, indicates that the 1kHz difference product with an equal mix of 19kHz and 20kHz tones reaches -54dB (0.2%), which is probably acceptably low.
Fig.5 Conrad-Johnson 16LS, spectrum of 50Hz sinewave, DC-1kHz, at 2.2V into 100k ohms (linear frequency scale).
Fig.6 Conrad-Johnson 16LS, spectrum of 50Hz sinewave, DC-1kHz, at 2V into 10k ohms (linear frequency scale).
Fig.7 Conrad-Johnson 16LS, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 2V into 100k ohms (linear frequency scale).
Somewhat surprisingly, the Premier 16LS measures a little worse than the less-expensive 17LS. However, also like the 17LS, the 16LS's high potential voltage gain means that, in real-world situations, the subjective effect of the preamplifier's measured failings will be minimal.—John Atkinson