Cary Audio Design CAD-572SE monoblock power amplifier Measurements
During the course of this review the Cary CAD-572SE was loaned back to Cary's UK agent for a local show. Unfortunately it was mishandled during transit, and was returned to me with several broken tubes. After these were replaced, one amplifier showed higher distortion than the other, though there was no significant change in overall sound quality. Channel balance remained almost perfect, while there was no difference in output power. Perhaps the replacement tube wasn't as linear as the first.
The CAD-572SE achieve its rated power with a mildly relaxed 3% criterion for distortion. I measured exactly 13dBW into 8 ohms, which is 19.995W. Not surprisingly, this was at a 1kHz midband frequency. The usual 20Hz and 20kHz power points are out of reach at this level for a constant distortion limit. While the 572SE managed close to full power down to 50Hz (12.3dBW) and up to 20kHz (12.25dBW), the 20Hz frequency reading was 7.5dBW, 5.5dB below the maximum level.
Just how much this shortfall matters will depend on several factors. First, does the program material being played contain significant 20Hz content? Second, will the music's 20Hz component reach full level? Third, could your speaker reproduce it if it were present? Finally, is a 20W power ceiling sufficient to make the 20Hz components loud enough to be satisfying or even audible?
For the record, I think the Cary's power bandwidth is pretty good for a single-ended tube design. When driving a 4 ohm load it didn't give up the ghost, though the 4.5dB shortfall in level was greater than a typical halved-impedance speaker can make up (usually a gain of 3dB). In terms of available loudness, the 572SE is a 6-7W amplifier when driving more difficult "4 ohm" loads; ideally, these should be of greater than 95dB/W nominal sensitivity.
Conversely, 90-92dB of rated sensitivity will be sufficient for a "good" 8 ohm load and moderately loud playback in medium-size rooms. There was some small improvement in power bandwidth for 4 ohm loads, though the significant output impedance will mean that load interactions will be proportionately increased. For example, the 8 ohm damping factor is 4.3, while for 4 ohms it is half this at 2.15. This is likely to significantly change the intended bass alignments of most speakers.
The directly measured output impedance was pretty constant, and moderate for the class: at 2.1 ohms, 20Hz; 1.85 ohms, 1kHz; and 1.91 ohms, 20kHz. Peak output current was a modest ±2.7A, eminently appropriate in view of the clean current and voltage clipping, and modest output voltage.
Pulse power testing has little effect on class-A amplifiers, and only a fraction of a dB increase was available on equivalent peak program—perhaps 21W/8 ohms. Overdrive testing confirmed that this amplifier was inherently very fast, showing well over 100kHz of 1W bandwidth. The inherent recovery from overload was rapid and well behaved. This measured behavior agreed with the convivial clipping character heard in the listening tests.
The dominant distortion component was the second harmonic, while the decay in amplitude through to the higher harmonics was monotonic (smoothly decremented). As noted earlier, the distortion at rated power was typically 3%, aurally quite innocuous even on pure tones. I thus adopted 10W (-3dB) as the nominal upper-level working point for measuring distortion and got figures of -45dB, or 0.56% THD+Noise in the midband. The THD measured -37.5dB (1.33%) at 20kHz, -40dB (1%) at 50Hz, and -20dB (10%) at 20Hz. These are satisfactory figures, just on the verge of accepted audible thresholds for low-order harmonic-distortion spectra such as the Cary's.
Total harmonic distortion was greatly improved all around by 1W, typically -60dB (0.1%) or better at 1kHz (fig.1), and was held to a more than satisfactory 0.7% at 20Hz, 0.28% at 20kHz.
Fig.1 Cary CAD-572SE, spectrum of 1kHz sinewave, DC-10kHz, at 1W into 8 ohms (linear frequency scale).
To confirm that the distortion continued to decrease at lower and lower powers, I measured at 100mW—still a moderately loud, conversation-equivalent level—and got better than -70dB (0.015%) of distortion in the midband. At the frequency extremes, I measured 0.05% at 20kHz and 0.25% at 20Hz. Clearly, the 572SE's low-level purity is quite excellent.
At the 10W level, a very fair result was obtained for high-frequency intermodulation, with -44dB (0.63%) of difference tone and with a low proportion of higher-order sidebands. At a more representative 1W level (fig.2), the results were -61dB for Channel A and -55dB for Channel B—actually quite good. This helped confirm my subjective view concerning the 572SE's "sweet, clean" treble.
Fig.2 Cary CAD-572SE, HF intermodulation spectrum, DC-20kHz, 19+20kHz at 1W into 8 ohms (linear frequency scale).
Channel separation was understandably exemplary given the dual-mono construction. Residual hum measured -66dB for 1W unweighted, while the A-weighted reading was more than satisfactory at -83dB—sufficient for all but the most sensitive loudspeakers. I heard no residual noise at the listening position in quiet surroundings at an equivalent sensitivity of 94dB/W. Relative to full level, 20W/8 ohms, the noise measured -80dB. When this was A-weighted, a fine -96dB result was obtained (this the average of the two channels).
Channel balance was excellent, measuring ±0.02dB—probably a fluke for this particular set of tubes. Input sensitivity was quite normal—250mV for 1W and 1.18V for full power—so passive line controls are possible for use with higher-output CD players and digital sources. (There are, of course, Cary preamplifiers; the company suggests the SLP-50.) The input impedance is an easy-to-drive 140k ohms.
Unconditionally stable, the amplifier was happy driving a pure 2µF capacitor load, the overshoot increasing only to 12% and with no ringing at all. In contrast to that of many feedback amplifiers, this loading of 2µF had no measurable effect on the audio-range frequency response; the Cary 572SE is indeed highly compatible with electrostatic loudspeakers. On a pure resistive load (fig.3, top trace above 6kHz), the 572SE's response measured ±0.2dB from 12Hz to 22kHz, -0.5dB at 10Hz and 28kHz, and -3dB at 6Hz and 44kHz—a notably wide, flat bandwidth for 1W. In the ultrasonic range there was a mild "sub" resonance at 80kHz, this an inaudible out-of-band ripple that added a small 6% overshoot to the otherwise excellent squarewave response.
The other trace in fig.3 is the amplifier response driving a two-way "6 ohm" loudspeaker load, in this case a B&W CDM1SE, the impedance of which is shown in fig.4. The small B&W has a minimum value of 5.2 ohms at 200Hz and a peak of 28 ohms at 2kHz, plus a 5 ohm treble range. Driving this real loudspeaker, the Cary has an nonflat frequency response. By the standards for a loudspeaker, typically 50Hz-15kHz, ±3dB, the Cary's result doesn't look too dramatic, holding to 15Hz-30kHz, ±1.5dB. Nevertheless, the upper midrange is somewhat lifted—does this correlate with my feelings of a "slightly lightweight" midrange?—the upper treble is attenuated by 1.5dB—is this my "sweet upper treble"?—and the bass-response effect added some unevenness to the combined speaker/room response.
Fig.3 Cary CAD-572SE, frequency response at (from top to bottom at 10kHz): response at 1W into 8.2 ohms, 2.83V into a B&W CDM1SE loudspeaker load (1dB/vertical div.).
Fig.4 B&W CDM1SE loudspeaker, electrical impedance compared with that of an 8.2 ohm resistor (2 ohms/vertical div.).
Choose the right cables, a compatible speaker, and tweak the speaker placement in your listening room, and these errors do not generally amount to very much in the context of the overall sound quality.
All in all, this very well-balanced SE triode design delivers realistic power, bandwidth, and distortion performance in a compact package.—Martin Colloms