Audio Research VS110 power amplifier & SP16L line preamplifier Measurements

Sidebar 3: Measurements

The Audio Research SP16L preamplifier offered a moderate but in my view sensible maximum voltage gain of 11.5dB. The unity-gain setting of the volume control was with the "14" LED illuminated. The preamp didn't invert absolute polarity, and the input impedance was a moderately high 43k ohms over most of the band, this dropping to a still high 38k ohms at 20kHz.

The output impedance was a usefully low 262 ohms at mid and high frequencies, this rising to 549 ohms at 20Hz. As a result, the frequency response into a very low 600 ohm load (fig.1, bottom pair of traces) rolled off prematurely, reaching -3dB at 18Hz. The ultrasonic response can also be seen to roll off earlier into the low impedance than into the 100k ohms featured by the VS110 power amplifier (see later). This rolloff also occurred when the volume control was lowered. At unity gain, for example, the -3dB point into 100k ohms was reduced from above 200kHz to 128kHz. Even so, the rolloff in the audioband was still just 0.1dB, which is negligible.

Fig.1 Audio Research SP16L, frequency response at 375mV into 100k ohms (top) and 600 ohms (bottom), both with volume control at max (0.5dB/vertical div., right channel dashed).

The A-weighted signal/noise ratio (ref. 1V output, with the input shorted but the volume control full) was superb, at 96.6dBA. This worsened slightly to 90.7dB for an unweighted audio-bandwidth measurement, but much more so, to 63.4dB, when the measurement bandwidth was increased to 500kHz. Crosstalk (not shown) was buried beneath the noise floor below 2kHz, but did increase at higher frequencies, due to capacitive coupling between the channels. As the channel separation was still a good 65dB in both directions at 20kHz, this should be fine.

The downward slope of the traces in fig.2 indicates that distortion also lies beneath the noise floor for much of the preamp's dynamic range. The broad plateau in the measured THD+noise percentage between 2V and 3V in fig.2 reveals that the amplifier's gain vs noise vs distortion equation has been finely tuned, at least into 100k ohms. THD products start increasing in level only at output voltages above the maximum the SP16L will be required to deliver. Sensible engineering. Defining clipping as 1% THD, the maximum output level was a high 16V into 100k ohms, though this did decrease to a still respectable 9V into 10k ohms, which is typical of many solid-state power amplifiers. Just 785mV was available into 1k ohm, which suggests that the SP16L not be used to drive power amplifiers with input impedances much below 10k ohms.

Fig.2 Audio Research SP16L, distortion (%) vs output voltage at 1kHz into (from bottom to top): 100k, 10k, 1k ohms.

Fig.3 shows how the SP16L's THD+N percentage changes with frequency at 2V into 100k ohms, a level where the measurement is looking at true distortion rather than just noise. The level is respectably low over most of the band, but does start to rise above 10kHz, presumably because the amount of available open-loop gain for negative feedback to work is decreasing. The spectrum of the distortion products accompanying a 1kHz tone is shown in fig.4. This measurement was taken into an 8k load rather than 100k ohms, which is why the sum of the distortion products—0.0105%—is higher than the measured THD at 1kHz in fig.3. The second harmonic is the highest, at -80dB (0.01%), with then the third and fourth increasingly lower in level, which is the recommended behavior for good sound quality. (The low-frequency spuriae evident in this graph are due to a ground loop with the measurement computer that I couldn't eliminate, and should be ignored.)

Fig.3 Audio Research SP16L, THD+N (%) vs frequency at 2V into 100k ohms (right channel dashed).

Fig.4 Audio Research SP16L, spectrum of 1kHz sinewave, DC-10kHz, at 1V into 8k ohms (linear frequency scale).

Despite the increasing level of THD above 10kHz seen in fig.3, the SP16L offered low levels of intermodulation distortion with the demanding full-scale mix of 19kHz and 20kHz tones (fig.5). Even into the 8k ohm test load used, the 1kHz difference product lay at -88dB (0.0041%). Overall, this is excellent measured performance; other than its lack of ability to drive unrealistically low load impedances, nothing reveals the SP16L's use of tubes. The preamplifier's low level of noise, sensible gain architecture, and correspondingly high dynamic-range capability are particularly commendable.

Fig.5 Audio Research SP16L, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 1V into 8k ohms (linear frequency scale).

The VS110 power amplifier had an input impedance of around 100k ohms at low and middle frequencies, dropping to a still high 80k ohms at 20kHz. It will therefore be a good match with the SP16L. The voltage gain into 8 ohms was 28.35dB from the 8 ohm output transformer tap, 25.02dB from the 4 ohm tap. The amplifier didn't invert absolute polarity.

The output impedance from the 4 ohm tap was a fairly high 0.72 ohm across most of the audioband, this rising to 1.3 ohms at 20kHz. The equivalent figures from the 8 ohm tap were 0.93 ohm and 1.9 ohms, respectively. As a result, the modification of the amplifier's frequency response due to the Ohm's Law interaction between its output impedance and the way in which a speaker's impedance changes with frequency will be quite large. With our simulated speaker load, for example (fig.6, top dotted trace), there were response aberrations of ±0.7dB, which would be audible. From the 4 ohm tap (not shown), the aberrations were smaller, at ±0.5dB.

Fig.6 Audio Research VS110, 8 ohm tap, frequency response at (from top to bottom at 2kHz): 2.83V into simulated loudspeaker load, 1W into 8 ohms, 2W into 4 ohms, 4W into 2 ohms (1dB/vertical div., right channel dashed).

This graph also shows that the VS110's bandwidth is modulated by the load impedance. Into 8 ohms (top pair of traces above 4kHz), the output is down by an innocuous 0.5dB at 20kHz. Into 4 ohms, this rolloff increases to 0.9dB, while into 2 ohms it reaches -3.3dB. These rolloffs were less from the 4 ohm tap, suggesting that the VS110's owner not use an output tap much below the impedance of his speakers. This ultrasonic rolloff does slow down the risetime of a 10kHz squarewave (fig.7), however.

Fig.7 Audio Research VS110, 8 ohm tap, small-signal 10kHz squarewave into 8 ohms.

Channel separation (fig.8) was excellent in the R-L direction, at better than 80dB over much of the band, but worse in the other direction, particularly at high frequencies. Unweighted signal/noise ratios (ref. 1W into 8 ohms) were good, at 68.7dB wideband and 68.4dB audioband, these improving to 82.7dB when A-weighted.

Fig.8 Audio Research VS110, channel separation (10dB/vertical div., R-L dashed).

Audio Research
3900 Annapolis Lane North
Plymouth, MN 55447-5447
(763) 577-9700