Tube Power Amp Reviews

The first thing I do when I hook up an amplifier in the test lab is to set the Audio Precision System One to measure distortion. Then, to get the lowest possible reading, I experiment with all the different ways of grounding the device being tested to the test gear. That way, I know that the noise and distortion levels I measure are truly indicative of the amplifier's performance and are not due to some freak of grounding in the lab.

However, when I manhandled the first sample of the monster Cary CAD-1610-SE down the stairs into my basement lab—straight back, lift with the legs—and hooked it up, I couldn't get a distortion reading of below 0.7%, no matter the signal level, output transformer tap, or grounding arrangement. Yes, it is a single-ended amplifier, but given the presence of that two-story Kronzilla tube, I had hoped for a little more refinement, a little less distortion. I proceeded with measuring the beast, but couldn't help but wonder if a tube had gone bad.

Accordingly, I picked up the second amplifier from Jonathan and a spare set of tubes and started all over again. This sample performed a bit better, so all the published measurements refer to this amplifier. Jonathan did have a 6SN7 on the CAD-1610-SE's front-end go bad, so I tried replacing first one, then both of these, with no significant change in the measured performance. I didn't have spare 6BH7s to try, but I did try the second sample of the T-1610 output tube. Again, the measured performance didn't differ significantly.

But then I replaced the 300B tube: distortion and output impedance rose, and voltage gain and output power both dropped. Putting back the original 300B dropped the distortion and increased the available power. It appears, therefore, that the Cary amplifier's performance is critically dependent on this tube.

All measurements with the "good" 300B were taken with the volume control at its maximum setting. The output-stage bias was set to 200mA, as recommended by the manual. Overall voltage gain was low, at 20.9dB into 8 ohms (8 ohm transformer tap) and 19.3dB into 8 ohms (4 ohm tap). (Both figures were measured via the unbalanced input; the balanced input gave 0.5dB more gain.) The Cary didn't invert signal polarity, and its unbalanced input impedance was a high 92k ohms at 1kHz. The balanced input impedance was 150k ohms. Source impedance from the 8 ohm output was high at 3.4 ohms across most of the band, rising to 4 ohms at 20kHz. Things were better from the 4 ohm tap, at 1.75 ohms and 2.5 ohms, respectively.

Even so, the interaction between the amplifier's source impedance and the impedance of the loudspeaker with which it is used will be strong, as can be seen by the top trace in fig.1. (This was from the 4 ohm tap; from the 8 ohm tap, the modification exceeded ±1.5dB.) Into resistive loads, the amplifier is respectably flat down to the low bass. The high treble rolls off to an extent dependent on the load resistance. About 1dB of treble rolloff at 20kHz is visible into 16 ohms in this figure, while into 4 ohms, the rolloff is closer to –2dB. Perhaps more important, a number of ultrasonic peaks can also be seen. These suggest the presence of various parasitic resonances and perhaps imply some incipient instability. These are much better damped with lower load impedances, and, peculiarly, even more so with the balanced input (bottom trace).

Fig.1 Cary CAD-1610-SE, 4 ohm tap, frequency response at (from top to bottom at 20kHz): 2.83V into 16 ohms, dummy loudspeaker load, 8 ohms, 4 ohms, and 2 ohms (2dB/vertical div.).

A small-signal 1kHz squarewave (not shown) had an excellent square shape, though with some small ultrasonic oscillations visible on the leading edges. These can also be seen in the 10kHz waveform (fig.2), overlaying the slowed-down leading edges and the tops and bottoms of the wave.

Fig.2 Cary CAD-1610-SE, 4 ohm tap, small-signal 10kHz squarewave into 8 ohms.

Fig.3 shows how the THD+noise percentage varied with frequency at low levels. This was from the 4 ohm output tap; the 8 ohm tap was similar, but with more of a rise into the lower impedances. The 16 ohm figure is respectable, but the THD approaches 1% into 2 ohms. Figs.4 and 5 reveal that the distortion content is very strongly second-harmonic, which will somewhat mitigate the high level of distortion. You can also see from fig.5 that the higher harmonics descend in level with increasing order, which Jean Hiraga showed a quarter-century ago was subjectively desirable.

Fig.3 Cary CAD-1610-SE, 4 ohm tap, THD+noise (%) vs frequency at (from top to bottom at 4kHz): 2.83V into 2 ohms, simulated loudspeaker load, 4 ohms, 8 ohms, and 16 ohms.

Fig.4 Cary CAD-1610-SE, 4 ohm tap, 1kHz waveform at 2W into 8 ohms (top), distortion and noise waveform with fundamental notched out (bottom, not to scale).

Fig.5 Cary CAD-1610-SE, 4 ohm tap, spectrum of 50Hz sinewave, DC–1kHz, at 10W into 8 ohms (linear frequency scale).

Why does the amplifier have such a high level of distortion, and why is that distortion second-harmonic in nature? Fig.6 shows the waveform of the amplifier's output as it approached clipping into a 4 ohm resistive load. You can see that while the positive-going peaks reach 2.8V, the negative-going peaks reach only –2.5V. The Cary has a single-ended output tube; this cannot deliver less than 0mA on the negative peaks—200mA below its standing bias level—whereas there is a higher current limit with respect to positive peaks. So when the negative peak output current reaches –200mA, the signal is starved of current beyond that point, even though the positive peaks can continue to get larger. (If you're worried that the graph indicates the output current is much higher than 200mA RMS, remember that the 200mA is the current output by the tube, which will be amplified by the output transformer.) The result of this asymmetry is the introduction of heavy even-order distortion products.

Fig.6 Cary CAD-1610-SE, 4 ohm tap, waveform of 1kHz sinewave at 9.1W into 4 ohms (THD = 4.14%, current = 1.51A RMS).

As with the Legend Starlet amplifier Chip Stern reviews elsewhere in this issue, high second-harmonic distortion may be subjectively benign or even preferred, as long as it is not accompanied by high levels of intermodulation distortion. But other than at low levels, the big Cary does produce a lot of intermodulation. Fig.7, for example, shows the spectrum of its output while it drove an equal mix of 19 and 20kHz tones into 4 ohms from its 4 ohm tap just below the visible clipping level. The 1kHz difference component reaches a very high –30dB (3%). However, while many other products are visible, only two of these, at 18kHz and 21kHz, are higher than –60dB (0.1%), which may work in the amplifier's favor.

Fig.7 Cary CAD-1610-SE, 4 ohm tap, HF intermodulation spectrum, DC–22kHz, 19+20kHz at 8W into 8 ohms (linear frequency scale).

Returning to fig.5, you can see that the AC supply components at 60Hz and its harmonics are also well down in level, the one at 120Hz being at –70dB (0.03%). However, the overall noise floor was not that low, lying at –57.5dB ref. 2.8V into 8 ohms. A-weighting the measurement improved the S/N ratio to 74.5dB, which is still not as good as claimed in the specification.

With the progressive limiting of the negative peaks, an amplifier with a single-ended output stage doesn't necessarily hard-clip; especially into low impedances, the distortion just increases with level. This can be seen in figs.8 and 9, which show the percentage of distortion plotted against output power from the 8 and 4 ohm transformer taps, respectively. There are clipping "knees" into the higher impedances; looking at the power at these "knees" indicates that the amplifier does meet or even exceed its power specification, but only at a high 10% THD figure and into a load double the nominal transformer tap value.

Fig.8 Cary CAD-1610-SE, 8 ohm tap, distortion (%) vs continuous output power into (from bottom to top at 2kHz): 16 ohms, 8 ohms, 4 ohms, and 2 ohms.

Fig.9 Cary CAD-1610-SE, 4 ohm tap, distortion (%) vs continuous output power into (from bottom to top at 2kHz): 16 ohms, 8 ohms, 4 ohms, and 2 ohms.

These figures were taken under continuous sinewave drive, but using a low-duty-cycle 1kHz toneburst more typical of music program (fig.10) doesn't change the picture appreciably. Even at the more relaxed 3% THD figure, the Cary belies its size by giving out 8.8W into 8 ohms (???dBW) from its 4 ohm tap, and 4.8W into 4 ohms (???dBW). However, you can see from fig.10 that if you choose a generous 5% figure, which is really about where the amplifier runs out of grunt, it will drive 13.3W into 4 ohms (8.23dBW) and 38.5W into 8 ohms (15.9dBW), both, again, from the 4 ohm output tap.

Fig.10 Cary CAD-1610-SE, 4 ohm tap, distortion (%) vs 1kHz burst output power into 16 ohms (black trace), 8 ohms (red), 4 ohms (blue), 2 ohms (green), and 1 ohm (magenta).

The black and red traces in fig.10 show that the Cary does hard-clip into higher impedances, as revealed by the sudden rise in distortion at the right of each trace. But note the shape of the curves: there are actually decreases in distortion before the clip point. Watching the 'scope display, I could see that this was because, at around 14V RMS output, the positive halves of the waveform were starting to become squared-off. This reduced the waveform asymmetry, hence the even-order distortion, even though it introduced odd-order harmonics.

Measuring an amplifier like the big Cary, which did well in the listening room despite its gross objective imperfections (assuming that neither sample broke on the test bench), calls into question everything you think you know about audio. It's as if—in this age of 10-cylinder, 3-liter Formula 1 engines, capable (with their tiny jewel-like pistons and hydraulic valve lifters) of revving to 19,000rpm, but with a narrow useful powerband—you encountered a race entrant that featured a powerplant with a single 3-liter cylinder. You might get only one powered piston stroke every second or so, but boy, would you experience some wideband torque!

If the audio equivalent of that torque is what's important to you, then this silly and silly-priced amplifier might be what you need. Me, I just can't ignore what it does wrong on the test bench, no matter how glorious it might sound.—John Atkinson