Sharp SM-SX100 digital integrated amplifier Measurements
Although the Sharp SM-SX100 is really a power D/A converter, I first examined its performance as a conventional integrated amplifier. With analog sources and the volume control set to "128," the '100's maximum voltage gain (into 8 ohms) was 36.7dB with the rear-panel switch set to "8 ohms," 34.75dB set to "4 ohms." A volume-control setting of "81" resulted in clipping with a digital 0dBFS input signal, while the unity-gain setting for analog sources was an indicated "33" ("8 ohms") or "34" ("4 ohms"). The amplifier didn't invert signal polarity via the digital or single-ended analog inputs, but did so through the balanced input. (The XLRs are wired with pin 3 hot.) The single-ended input impedance was a usefully high 130k ohms.
Fig.1 shows the amplifier's small-signal frequency response in 8 and 4 ohm resistive loads and into our standard simulated speaker load, taken with the unit set to "4 ohms." Into 8 ohms, the response is flat in the audioband, with a very slight rise above 15kHz and a peak (somewhat down in level) just above 100kHz, before a rolloff to -10dB at 200kHz. These features are due to the passive low-pass filter on the SM-SX100's output. While the shape of the curve is the same into 4 ohms, there appears an early rolloff of 2dB down at 20kHz. Into the dummy load (top trace at 3kHz) the HF response is maintained out to 20kHz, but the voltage-divider action between the speaker impedance and the source impedance of the amplifier results in the usual modification of that response, in this case by ±0.5dB.
Fig.1 Sharp SM-SX100, 4 ohm output, frequency response at (from bottom to top at 3kHz) 2W into 4 ohms, 1W into 8 ohms, and 2.838V into simulated loudspeaker load (1dB/vertical div., right channel dashed).
When I measured the impedance, it was the same—0.2 ohms at 1kHz and below—from both the "4" and "8 ohm" settings, suggesting that all this switch does is reduce the output-stage rail voltage for the lower impedance setting. However, the impedance did increase dramatically in the top octave, reaching approximately 1.3 ohms at 20kHz. I say "approximately" because the amplifier could not sustain its output voltage at 20kHz. I estimate source impedance by measuring the amplifier's output voltage with no load, then measuring the voltage drop when an 8 ohm load is connected. With the Sharp, connecting the load not only caused the voltage to drop immediately, but it then gradually declined with time. This lack of output-voltage regulation can be seen in an oscilloscope trace taken with the amplifier reproducing a 1kHz toneburst (fig.2). The first cycle of the burst is noticeably higher in level than the subsequent cycles.
Fig.2 Sharp SM-SX100, 4 ohm output, waveform of 10-cycle 1kHz toneburst at 7.65W into 2 ohms.
The Sharp's reproduction of a 1kHz squarewave (not shown) was textbook, other than the flat tops of the waveform being overlaid with quite a high level of ultrasonic noise. This noise can also be seen overlaying the 10kHz squarewave (fig.3). (The shape of this squarewave's leading edges correlates with the passive filter's dip-and-peak ultrasonic response seen in fig.1.) Channel separation (not shown) was hard to measure, due to the presence of the ultrasonic noise, but appeared to be better than 60dB below 10kHz.
Fig.3 Sharp SM-SX100, small-signal 10kHz squarewave into 8 ohms.