Bel Canto SET 80 monoblock power amplifier Measurements

Sidebar 3: Measurements

Bel Canto makes bold claims for good power delivery in the context of SE triode designs. In general, both power and bandwidth are in short supply with this technology, since little or no negative feedback is available to rescue a failing frequency response or a distortion-limited power maximum.

The SET 80 performed well on the power and power-bandwidth tests, notwithstanding a degree of handicap afforded this class of amplifier. It's general to allow a 3% maximum of distortion-limited power in view of the relatively benign low-order nature of SE distortion. Driving an 8 ohm load from the Bel Canto's 8 ohm setting gave a flat-out power of 70W to visible oscilloscope clipping (18.5dBW), and 61W (17.9dBW) for 3% measured distortion.

The available power was substantial, especially for an SE type, even if its topology has some parallels with both bridge and push-pull configurations. As the load impedance rose, so did the output voltage—to a maximum of 33V RMS, or the equivalent of 136W, this a function of the relatively high output impedance and the source regulation of the output stage itself.

The measured output resistance was 1.7 ohms, corresponding to a low damping factor of 4, or better than the 3 quoted in the specification. With the SET 80 switched to 4 ohms (generally, the following results conform to this setting), the "8 ohm" damping factor was 6.2, corresponding to an output resistance of 1.3 ohms, which provides better control of the speaker's frequency response. This was clearly audible on test.

While 8 ohm speaker loads with broad regions of higher impedance will play loudest, the maximum output power on the 4 ohm setting for an 8 ohm load was only 0.8dB lower. Feel free to use the one you prefer.

Distortion at higher power levels, if aurally benign, was nonetheless mildly audible as a shift in tonal balance. Careful checking of the time-distortion spectra showed that the "sweetest" spectra were obtained with load matching—ie, with the 4 ohm load on the 4 ohm setting and the 8 ohm load on the 8 ohm setting—although the difference wasn't that great.

Fig.1 shows the typical distortion spectrum of the Bel Canto at 1W into a matched load. The "push-pull" nature of the output stage was evident in the partial cancellation of second-harmonic distortion, leaving third-harmonic slightly dominant at a innocuous –73dB. Intriguingly, some higher-order harmonics can be seen, as well as some 11th-harmonic, although this is at a very low –100dB.

Fig.1 Bel Canto SET 80, spectrum of 1kHz sinewave, DC–20kHz, at 1W into 4 ohms, 4 ohm tap (linear frequency scale).

Compare this spectrum with that in fig.2, where the amplifier was mismatched (if kindly so), by being asked to drive an 8 ohm load from the 4 ohm transformer setting. Now the output is harmonically rich, if more monotonic in character (smoothly decreasing with increasing harmonic order). The fifth-harmonic was now present at –75dB: interesting.

Fig.2 Bel Canto SET 80, spectrum of 1kHz sinewave, DC–20kHz, at 1W into 8 ohms, 4 ohm tap (linear frequency scale).

Subjectively, it would be hard to separate the "sounds" of these two kinds of linearities. These spectra indicate the changes in load matching due to the different output impedance. From the 4 ohm setting, the output is more current-rich while from the 8 ohm setting, it is more voltage-rich.

At low powers, around 1W, the SET 80 proved to be surprisingly linear. Excluding hum, the 1kHz result was –76dB (0.015%), and was only a little poorer at 0.03% at 20kHz. At 20Hz, it measured –60dB (0.1%), which is still commendable. At an equivalent of 25 watts output, the high-frequency intermodulation result for an equal mix of 19kHz and 20kHz tones (fig.3) was a creditable –57dB (0.027%). This improved to –73dB at 1W, a typical music level.

Fig.3 Bel Canto SET 80, HF intermodulation spectrum, DC–20kHz, 19+20kHz at 10W into 8 ohms, 8 ohm tap (linear frequency scale).

I checked the SET 80's power bandwidth for an 8 ohm load from the 4 ohm setting at 3% distortion. In context, I was impressed. I got 38W at 20Hz, 50W at 1kHz, and 20.3W at 20kHz. (With a more strict 1% criterion, these figures reduced to 23W, 26W, and 15W.) These figures are very good for the genre, indicative of a carefully executed, wide-bandwidth design especially regarding the output transformers. Referenced to 50W at 1kHz into 8 ohms, the –3dB points were 17Hz and 19kHz, which are not bad at all in the context of the power spectrum of naturally recorded music.

Some power supply residual hum was present. Unweighted, the hum and noise was 65dB below 1W. This proved to be just audible at 1 meter from speakers with a sensitivity higher than 90dB/watt. Eliminating the contribution from hum by A-weighting the measurement, the 1W-referenced S/N ratio was 86.5dB. Relative to full power, the unweighted S/N was 85dB and 106.5dB A-weighted. Excluding hum from the unweighted figure, I got 114dB. Really high-sensitivity speakers, such as some horn designs, might expose some background hiss with the SET 80.

The small-signal frequency response (fig.4, solid trace) was wide, with the aurally just-significant –0.5dB points at 13.6Hz and 12.6kHz. The 3dB rolloff occurred at approximately 5Hz and at 28kHz. Of mild initial slope, the high-frequency rolloff was not felt to be a major factor. Shown dashed in fig.4 is the Bel Canto's response on the 8 ohm setting with the Wilson WITT Series II speaker. Channel balance held within 0.3dB.

Fig.4 Bel Canto SET 80, frequency response at 1W into 8 ohms, 8 ohm tap (solid) and at 1W into Wilson WITT Series II (dashed) (1dB/vertical div.).

Checked for stability on squarewaves, the SET 80 was unconditionally stable into any load—even a pure 2µF capacitor—and was clearly powerful enough to drive electrostatic speakers. The Bel Canto monoblock proved to be quite imperturbable—another feature of low- and zero-feedback design. For the record, I have reproduced the 330Hz squarewave for an 8 ohm/2µF load (fig.5). This features absolutely minimal overshoot and ringing.

Fig.5 Bel Canto SET 80, small-signal 330Hz squarewave into 8 ohms in parallel with 2µF.

The SET 80 was easy to drive, its measured input impedance very close to the 100k ohms specified. Note that the sensitivity is fairly low, the SET 80 requiring 3.34V for 60W output into 8 ohms from the 4 ohm output tap and 4V for full clipping.—Martin Colloms

Bel Canto Design
212 Third Avenue N., Suite 446
Minneapolis, MN 55401
(612) 317-4550