LABORATORY TEST REPORT
Newport Test Labs’ measurements proved beyond any shadow of a doubt that the Bel Canto ACI 600 is a powerful amplifier. It delivered more than 300-watts per channel into 8 , irrespective of whether one or both channels was driven, and more than 600-watts per channel into 4 loads, again irrespective of whether one or both channels was driven. It also was also able to deliver this above-specification power right across the frequency band, from 20Hz to 20kHz. The results are tabulated in the test result table, and shown graphically as a bar graph. Note that we haven’t shown the usual two bar graphs (one for single-channel-driven and the other for both-channels-driven) because the two graphs were identical.
Channel separation was superb, besting 100dB at low and midrange frequencies and still 91dB at 20kHz. Channel balance was also excellent, with Newport Test Labs measuring 0.116dB at 1kHz. Interchannel phase was excellent at low and midrange frequencies, as you can see from the tabulated results, and only 1.94° out at 20kHz.
Distortion was very, very low, measured either into 8 or 4 loads and also either at low output levels or at rated output. Graph 1 shows harmonic distortion levels when the amplifier is delivering one watt into an 8 load. You can see a single third harmonic component at –110dB (0.00031%) and then fifth and sixth harmonic components at around –115dB (0.00017%), but that’s about it. Equally remarkable is the low level of the noise floor—down close to –130dB—and the lack of low-frequency noise. Performance was almost as good when the Bel Canto ACI 600 was driving a 4 load. This time there’s a second harmonic at –105dB (0.00056%) and the third harmonic is a little higher in level, at around –108dB (0.00039%). There’s also a fourth harmonic at –122dB (0.00007%), plus the fifth and sixth harmonics are fractionally higher than they were when the amplifier was driving an 8 load, at around –113dB (0.00022%). Overall wideband distortion at one watt was just 0.005%, as you can see in the tabulated figures.
As I noted earlier, distortion was also very low at rated output, as you can see from the measurements made by Newport Test Labs that are shown in Graph 3 and Graph 4. At an output of 300-watts into 8 , apart from the second harmonic at –90dB (0.00316%) and the third harmonic at –92dB (0.00251%), all other distortion components were more than 100dB down (0.001%). Distortion increased slightly when the ACI 600 was delivering 600-watts into 4 (Graph 4) but apart from the second and third harmonics, all other distortion components were more than 100dB down as well. In both graphs you can see the noise floor has dropped even lower—down close to –140dB. The extra distortion components visible in Graph 4 appear to be the result of an increased level of 100Hz signal from the power supply. Overall wideband distortion at rated output was measured at 0.004%.
Intermodulation distortion was exceptionally low, as you can see from Graph 6. The 18kHz and 21kHz sidebands alongside the two test signals are at –95dB (0.00177%)
and –97dB (0.00141%), while the 17kHz and 22kHz sidebands are both around –114dB (0.00019%). The unwanted regenerated signal at 1kHz is nearly 120dB down. Interestingly, there’s an obvious signal at 2kHz, though as it’s 111dB down (0.00028%) it would not be audible. Frequency response was measured as extending from 7Hz to 21kHz –1dB, and from 4Hz to 33kHz –3dB. This response is shown in Graph 5 and was measured with a laboratory standard non-inductive 8 load.
The effect of the Bass Eq control is shown in Graph 7. You can see that it delivers a 3dB boost or cut from 20Hz up to 100Hz, after which the boost or cut rapidly diminishes, so there’s basically no effect on the response above 300Hz.
The effect of the Tilt controls is shown in Graph 8. Although at first glance the boost and cut effect seems to be the same as standard bass and treble controls, if you look closely you can see quite a few differences.
The performance of Class-D amplifiers continues to improve apace and now mostly rivals or exceeds that of linear amplifiers
Firstly, maximum boost and cut is restricted to around 3dB, whereas tone controls usually offer around 8–12dB. Second, the frequency where there’s no boost or cut at all when either of the controls is used is somewhat lower than usual, at 780Hz (it’s usually at 1kHz). Finally, the range of frequencies that is boosted (or cut) is somewhat wider than is usual with tone controls, both in the bass and treble regions.
Signal-to-noise ratios were measured at 81dB unweighted and 91dB A-weighted referenced to one-watt output, and at 94dB unweighted and 100dB A-weighted referenced to rated output. Although these are good results, the reason they’re not better is because of the presence of high-frequency switching noise from the Class-D output stage. Although this noise is so high in frequency that it is not audible, it affects the measurements.
The switching noise certainly affects the square wave measurements, so for the square waves shown with this report, Newport Test Labs used a low-pass filter to remove it. The 100Hz wave shows a degree of tilt, indicating that the Bel Canto ACI 600’s low-frequency response rolls off at low frequencies and does not extend to d.c., but there’s no bending, so there’s no phase shift at low frequencies. The 1kHz square wave shows a slight rise-time limitation, plus an overshoot that would suggest a lift in the frequency response at some ultrasonic frequency. The 10kHz square wave is of course affected by the presence of the low-pass filter used by the lab, but you can see the rise-time limitation, the high-frequency response roll-off, and the ultrasonic lift. I have included a square wave of a the same 10kHz wave without the filter in place, and you can see that the high-frequency switching noise obscures the information the square wave test could otherwise reveal. The square wave showing the Bel Canto ACI 600’s performance into a highly capacitative load shows that the amplifier it will be unconditionally stable when driving highly reactive loads, such as electrostatic loudspeakers.
The performance of Class-D amplifiers continues to improve apace and now mostly rivals or exceeds that of linear amplifiers in almost all areas. The amplifiers inside the ACI 600 are exceptionally good examples of Class-D engineering.