Cary CAD-280SA V12 power amplifier Measurements
As Jonathan Scull noted, with its 12 EL34 output tubes, the Cary CAD-280SA V12 sure do run hot. I let it warm up (literally) for an hour after installing the tubes, then adjusted the output-stage bias for each channel via the convenient rear-panel phone jacks. The trimpots' action is rather coarse—even an imperceptible twist with a screwdriver changed the bias current by tens of milliamps. The closest I could get to the recommended 275mA was 278mA in the right channel, 283mA in the left.
The Cary amplifier didn't invert absolute polarity. The signal/noise ratio was okay, at 65dB (unweighted, ref. 1W into 8 ohms) and 80.4dB (A-weighted). In triode mode, the CAD-280SA's voltage gain was low, at 21.6dB (8 ohm tap into 8 ohms) and 18.6dB (4 ohm tap into 4 ohms). Even though the gain in ultralinear mode increased to 25.2dB and 22.1dB, respectively, these are still lower figures than usual. They didn't change for balanced operation, so I performed all the rest of my testing using the single-ended inputs, which was how JS had used the amplifier. The unbalanced input impedance was a very high 130k ohms in the midrange and below, dropping slightly to a still high 100k ohms at 20kHz. The balanced figures were the same.
Fig.1 plots the V12's small-signal amplitude response from the 4 ohm output transformer tap with the amplifier operating in triode mode. The response is well-extended at the frequency extremes, with the -3dB points lying at 10Hz and 70kHz—the Cary obviously uses well-designed output transformers. But note the almost ±2dB variation of the response with our simulated loudspeaker load and the large difference in absolute level each time the load impedance halves. These imply a relatively high output impedance; it's even higher from the 8 ohm tap (fig.2), and especially so with the amplifier's output operating in ultralinear mode (fig.3), where the response variation reaches an extremely audible ±4dB.
Fig.1 Cary CAD-280SA V12, triode mode, 4 ohm tap, frequency response at (from top to bottom at 2kHz): 2.83V into dummy loudspeaker load, 1W into 8 ohms, 2W into 4 ohms, and 4W into 2 ohms (2dB/vertical div., right channel dashed).
Fig.2 Cary CAD-280SA V12, triode mode, 8 ohm tap, frequency response at (from top to bottom at 2kHz): 2.83V into dummy loudspeaker load, 1W into 8 ohms, 2W into 4 ohms, and 4W into 2 ohms (2dB/vertical div., right channel dashed).
Fig.3 Cary CAD-280SA V12, ultralinear mode, 8 ohm tap, frequency response at (from top to bottom at 2kHz): 2.83V into dummy loudspeaker load, 1W into 8 ohms, 2W into 4 ohms, and 4W into 2 ohms (2dB/vertical div., right channel dashed).
In fact, allowing for some variation in the calculated output impedance with the level and load impedance used, we're looking at source impedances of 5.5 ohms (8 ohm tap) and 2.8 ohms (4 ohm tap) in triode mode. While these are both high, they are exceeded by the ultralinear figures: 12 and 6 ohms, respectively! Fortunately, the impedances don't change much across the audioband, but, as figs.1-3 reveal, there will be a large and audible change in frequency response depending on which loudspeaker is used with the Cary and which transformer tap and mode of operation are used.
However, as I said above, the amplifier uses good transformers, which is revealed by the essentially perfect shape of the small-signal 1kHz and 10kHz squarewaves (figs.4 and 5). Surprisingly for a design that appears to site the two channels on opposite sides of the chassis, channel separation was only okay L-R (fig.6), and slightly worse in the R-L direction.
Fig.4 Cary CAD-280SA V12, triode mode, 8 ohm tap, small-signal 1kHz squarewave into 8 ohms.
Fig.5 Cary CAD-280SA V12, triode mode, 8 ohm tap, small-signal 10kHz squarewave into 8 ohms.
Fig.6 Cary CAD-280SA V12, channel separation (10dB/vertical div., R-L dashed).
Plotting the Cary's small-signal THD+noise percentage from the 4 ohm tap against frequency (triode operation) gave the traces in fig.7. For some reason, the right channel is quite a bit more linear in the bass at low levels than the left, though the left channel is slightly better in the midband; a similar picture could be seen from the 8 ohm tap (not shown). Fig.7 clearly shows that the amplifier is more comfortable working into higher impedances, and that its high-frequency linearity is not as good as its midrange linearity.
Fig.7 Cary CAD-280SA V12, triode mode, 4 ohm tap, THD+noise (%) vs frequency at (from top to bottom at 10kHz): 4W into 2 ohms, 2W into 4 ohms, 2.83V into simulated loudspeaker load, and 1W into 8 ohms (right channel dashed).
Of course, what matters most when it comes to distortion is its spectral distribution. The lower trace in fig.8 reveals that the primary harmonic present in the V12's output is the relatively benign third, though higher harmonics do impose an envelope on that waveform. This graph was taken in triode mode; things were not appreciably different in ultralinear mode. The predominance of the third harmonic is confirmed by the spectral analysis shown in fig.9. At this high power level (35W, triode operation, 8 ohm tap into 8 ohms), the third harmonic lies at -40dB (1%). Probably of subjective significance is the well-controlled drop in the level of individual harmonics as they increase in order. This was shown almost 25 years ago by Jean Hiraga to sound musically natural.
Fig.8 Cary CAD-280SA V12, triode mode, 4 ohm tap, 1kHz waveform at 1.2W into 4 ohms (top), distortion and noise waveform with fundamental notched out (bottom, not to scale).
Fig.9 Cary CAD-280SA V12, triode mode, 8 ohm tap, spectrum of 50Hz sinewave, DC-1kHz, at 35W into 8 ohms (linear frequency scale).
This spectrum was taken from the left channel; if you glance back to fig.7, you'll see that this channel was actually worse at the test frequency than the right. Repeating the test with a 1kHz tone at the same power level (not shown) resulted in identical levels of spectral components, revealing that the difference in the channels' low-frequency nonlinear behaviors was of significance only at low power levels.
Switching the Cary to ultralinear operation reduced the levels of distortion harmonics (fig.10), mainly because this mode's 3.6dB-greater voltage gain meant that the drive signal had to be backed off by the same 3.6dB to maintain the same output power. However, while a little lower in absolute level, the harmonic spectrum is effectively identical in both modes of output-stage operation.
Fig.10 Cary CAD-280SA V12, ultralinear mode, 8 ohm tap, spectrum of 50Hz sinewave, DC-1kHz, at 35W into 8 ohms (linear frequency scale).
Although the CAD-280SA's nonlinear transfer function produces levels and orders of harmonic distortion that may well be subjectively benign with music program, the high-frequency nonlinearity seen in fig.7 results in quite high levels of intermodulation with our torture test of equal amounts of 19kHz and 20kHz tones (fig.11). At 17W into 8 ohms (triode operation, 8 ohm tap)—a measured power level just below visible clipping on the 'scope with this stressful signal—the 1kHz difference component reaches -52dB (0.25%), while the higher-order components at 18kHz and 21kHz reach 0.5% in level, with other spectral components apparent. Ultralinear operation at the same power level (fig.12) made the situation worse. It is possible that this behavior correlates with J-10's finding the Cary's high frequencies to sound a little "crispy."
Fig.11 Cary CAD-280SA V12, triode mode, 8 ohm tap, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 17W into 8 ohms (linear frequency scale).
Fig.12 Cary CAD-280SA V12, ultralinear mode, 8 ohm tap, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 17W into 8 ohms (linear frequency scale).
Logistical problems involved with the computer-controlled resistive load I use with the Miller Audio Research Amplifier Profiler software meant that I couldn't perform my usual toneburst test of maximum output power on the Cary. However, on continuous tones with both channels driven, the amplifier didn't quite meet its specification, even with the load matched to the output-transformer tap and at a relaxed 3% distortion definition of clipping. In triode mode into 4 ohms (fig.13), the 4 ohm tap reached 3% THD+N at 45W (13.5dBW). While the available power did increase in ultralinear mode (fig.14), this was still below specification at 92W (16.6dBW). It should be noted, however, that the measured powers are only half a dB or less than the rated power, which is not really much to get concerned about.
Fig.13 Cary CAD-280SA V12, triode mode, 4 ohm tap, distortion (%) vs continuous output power into (from bottom to top at 10W): 8 ohms, 4 ohms, and 2 ohms.
Fig.14 Cary CAD-280SA V12, ultralinear mode, 4 ohm tap, distortion (%) vs continuous output power into (from bottom to top at 10W): 8 ohms, 4 ohms, and 2 ohms.
Cary's CAD-280SA V12 performed quite well on the test bench; its squarewave responses were particularly excellent. However, I was concerned by the highish levels of HF intermodulation and the very high source impedances, particularly in ultralinear mode. Unless you really need the extra 3dB of dynamic range you get from ultralinear operation compared with triode, and the extra 3dB you also get from the amplifier's 8 ohm tap compared with the 4 ohm tap, my advice is to use the CAD-280SA in triode mode with the speakers hooked up to its 4 ohm outputs.—John Atkinson