Cayin SP-10A Integrated amplifier Measurements
I used Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see the January 2008 "As We See It" and www.ap.com) to measure the Cayin SP-10A amplifier. Because some of the measurements produced anomalous results, I repeated some of the tests with our Audio Precision System One.
With its volume control set to its maximum, the Cayin's voltage gain from its 8 ohm tap was lower than the norm for an integrated amplifier30.7dB into 8 ohmsand even lower from its 4 ohm tap: 23.8dB into 8 ohms. This won't be an issue with CD players, but might be with legacy analog sources such as FM tuners. The unity-gain setting of the volume control was 8:30; the 12:00 position was about 6dB below full volume. The amplifier inverted absolute polarity from both taps, and its input impedance was usefully high at 59k ohms at 20Hz and 1kHz, dropping inconsequentially to 50k ohms at 20kHz.
The output impedance from the 8 ohm tap was very high, at 5 ohms across the audioband. As a result, the variations in response with our standard simulated loudspeaker (fig.1, gray trace) were a very audible ±2.2dB. By contrast, the output impedance from the 4 ohm tap, which is what Bob Reina used for his auditioning, was almost 10 times lower, at 0.57 ohm across the audioband, which gave variations of just ±0.3dB (fig.2, gray trace). As well as the excellent matching between channels (which was maintained at lower settings of the volume control), these two graphs reveal that the SP-10A has a very wide small-signal bandwidth, which correlates with the amplifier's excellent reproduction of a 10kHz squarewave (fig.3). However, some overshoot and ringing at a very high frequency can be seen in this graph, which correlates with the response peak at the far right of fig.2.
Fig.3 was taken from the 4 ohm tap; the 10kHz squarewave from the 8 ohm tap (fig.4) retained the overshoot, but the ringing was absent. The low-frequency squarewave (fig.5) was superbly square, with flat tops, indicating that the Cayin amplifier has a well-designed output transformer with an extended low-frequency response.
The SP-10A's channel separation was only okay, at 65dB at 1kHz in both directions and 42dB LR, 46dB RL, at 20kHz, which is perhaps due to the fact that one of the three small-signal tubes is shared between channels. However, the Cayin is fairly quiet; the wideband, unweighted signal/noise ratio, taken from its 8 ohm tap into 8 ohms with the input shorted but the volume control at its maximum, measured 68.8dB in the left channel and 70.1dB in the right. The ratios improved to 71.1 and 78.8dB, respectively, when the measurement bandwidth was restricted to the audioband, and to 86.4 and 87.5dB when A-weighted.
The traces in fig.6 indicate that with the high output impedance from its 8 ohm tap, the SP-10A behaves more like a current source than a conventional voltage source. The clipping power into 8 ohms (with clipping defined as 1% THD+noise) was 23.9W (13.8dBW), but this dropped to 11.6W into 4 ohms (7.6dBW), and 6.8W into 2 ohms (2.3dBW). All of these figures are less than the specified 38Wpc, though the Cayin does put out 38W into 8 ohms at 5% THD+N.
The picture was very different from the 4 ohm tap (fig.7). The amplifier clips at just 6.1W into 8 ohms (7.85dBW), 11.2W into 4 ohms (7.5dBW), and 18.5W into 2 ohms (6.65dBW). The SP-10A behaves more like a traditional voltage-source amplifier from its 4 ohm tap, at the expense of generally lower power than is usually obtained from a push-pull pair of 6L6GC tubes. I must admit to some surprise that BJR got sufficient loudness from his Epos speakers, which have a measured sensitivity of 87.1dB(B)/2.83V/m.
The THD+noise levels below clipping in these two graphs are respectably low, particularly from the 4 ohm tap, but there is an odd-looking spike in distortion at 1.5W from the 4 ohm tap into 4 ohms (fig.7, bottom trace). The graph was taken from the left channel; the right channel behaved identically. I will return to this behavior shortly.
Figs. 8 and 9 show how the THD+N percentage in the SP-10A's output varies at 2.83V into various loads from the 8 and 4 ohm taps, respectively. The lowest distortion is obtained when the load is twice the nominal tap value, but there are notable rises in distortion at the frequency extremes. The high-frequency rise will be due to the decreasing effect of loop negative feedback on the amplifier's open-loop linearity; the low-frequency rise will be due to saturation of the output transformer's core. However, this is not unexpected with a moderately priced tubed design such as this.
Fig.10 shows the waveform of the THD+N residual at 1.5W from the 4 ohm tap into 8 ohms. The percentage of THD+N was just 0.046%, which is commendably low, and the distortion consists primarily of low-order harmonics. However, when the load was halved, to 4 ohms (fig.11), the distortion rose by a factor of 10, with now a spike apparent at the waveform's positive peak. (Note that the spuriae waveforms in these two graphs are not to scale.) This actually appeared to be a burst of ultrasonic oscillations rather than a single spiketheir frequency was too high for the digital oscilloscope to capture properly. Such behavior suggests that the SP-10A is not unconditionally stable, and that the circuit becomes unstable at a specific combination of voltage and current from the 4 ohm tap.
I wondered if this was a sample fault, but the two channels misbehaved identically. The fault was also unrelated to the volume-control setting. What was odd about this behavior was that increasing or decreasing the output level even slightly eliminated the instability. It's possible, therefore, that as in practical use it is unlikely that the SP-10A will be asked to drive a sustained tone at exactly this level into 4 ohms, this misbehavior looks worse than it sounds. I asked BJR if he had noticed any noise or distortion problems with the review sample driving his Epos speakers from its 4 ohm taps, but he hadn't.
Fig.12 confirms that the SP-10A's distortion spectrum predominantly consists of the subjectively innocuous low-order harmonics. However, the fundamental tone in this graph is accompanied by sidebands spaced at 60Hz, suggesting that the AC transformer is a little undersized. This graph was taken from the 4 ohm tap; the 8 ohm output's behavior was similar.
The Cayin didn't do particularly well on the high-power, high-frequency intermodulation test (fig.13). Even though its output level during this test was below visible clipping on the oscilloscope, many intermodulation products can be seen: the 1kHz difference tone lay at 54dB (0.2%), as did the higher-order products at 18 and 21kHz. In addition, the right channel (red trace) is not as linear as the left (blue).
Summing up the Cayin SP-10A's measured performance is difficult; it must be remembered that this is a relatively affordably priced product when compared against the similar-looking Leben, Luxman, and Quad integrated amplifiers that Art Dudley reviewed in November 2011. But I am concerned about its 4 ohm tap's curious instability and lack of distortion-free power. As Bob wrote, the amplifier just "ran out dynamic gas" when asked to play loudly with rock music from this output.John Atkinson