Bel Canto Design DAC 1 D/A processor Measurements

Sidebar 3: Measurements

I performed all my measurements on the second sample of the Bel Canto processor (S/N D002078A). The processor successfully locked onto datastreams with sample rates between 44.1kHz and 96kHz, though it ignored a 32kHz feed. It seemed very prone to picking up RF interference when fed an electrical S/PDIF datastream, so, except where noted, I drove it using an AudioQuest TosLink datalink.

Maximum output was a hair higher than the "Red Book" standard at 2.09V RMS, while the absolute phase was correct with the front-panel button pushed in. Output impedance was a low 50.1 ohms throughout the midrange and treble, rising insignificantly to 50.5 ohms in the very low bass.

Frequency response was flat (fig.1, bottom pair of traces), but the processor didn't apply any de-emphasis with pre-emphasized signals (top pair of traces). While only a handful of CDs are pre-emphasized, this is a tad sloppy. Channel separation, however, was superb, at better than 110dB across the band (not shown).

Fig.1 Bel Canto DAC 1, frequency response with (top) and without (bottom) de-emphasis at -12dBFS (right channel dashed, 0.5dB/vertical div.).

Fig.2 shows the spectral analysis of the Bel Canto's output while it decoded data representing a dithered 1kHz tone at -90dBFS. The peak pokes its head just above the -90dB line, hinting at a shade of positive linearity error. More significant is the presence of power-supply components at 60Hz, 180Hz, and 300Hz, which imply some magnetic coupling between the AC transformer and the audio circuitry. (Electrical coupling usually gives rise to spuriae at the fullwave rectified frequency of 120Hz and its harmonics.) I tried moving the DAC 1 well away from all other pieces of electrical equipment, with no change in the spectrum, suggesting this is an internal problem. While increasing the bit depth from 16 to 24 bits drops the noise floor in the treble by an excellent 15dB, equivalent to dynamic range between the 18- and 19-bit level, the hum pickup means that the midrange and bass are not improved by the increase in word length. The same phenomenon can be seen in fig.3, a wideband spectral analysis of the DAC 1's output while it decoded 16- and 24-bit digital black.

Fig.2 Bel Canto DAC 1, 1/3-octave spectrum of dithered 1kHz tone at -90dBFS, with noise and spuriae, 16-bit data (top) and 24-bit data (bottom). (Right channel dashed.)

Fig.3 Bel Canto DAC 1, 1/3-octave spectrum of "digital black," with noise and spuriae, 16-bit data (top) and 24-bit data (bottom). (Right channel dashed.)

The linearity error (fig.4) was very low to below -110dBFS, if very slightly positive below -80dBFS, as suggested by fig.2. The excellent linearity, coupled with the very low level of analog noise, allowed the processor to very accurately reconstruct the waveform of an undithered 16-bit/1kHz sinewave at -90.31dBFS (fig.5). The three discrete voltage levels are well-defined, though the overall slope of the waveform is, again, due to the presence of some 60Hz hum. Increasing the word length to 24 bits gave a good facsimile of a sinewave (fig.6).

Fig.4 Bel Canto DAC 1, left-channel departure from linearity, 16-bit data (2dB/vertical div.).

Fig.5 Bel Canto DAC 1, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit data.

Fig.6 Bel Canto DAC 1, waveform of undithered 1kHz sinewave at -90.31dBFS, 24-bit data.

Levels of distortion were very low, either of harmonic distortion (fig.7) or intermodulation (fig.8), even into the severe 600 ohm load.

Fig.7 Bel Canto DAC 1, spectrum of 50Hz sinewave, DC-1kHz, at 0dBFS into 600 ohms (linear frequency scale).

Fig.8 Bel Canto DAC 1, HF intermodulation spectrum, DC-22kHz, 19+20kHz at 0dBFS into 600 ohms (linear frequency scale).

As I said earlier, the Bel Canto DAC 1 was susceptible to picking up interference when fed an electrical S/PDIF datastream. This can be seen in fig.9, which shows the high-resolution spectra of the DAC 1's analog output when fed an electrical (grayed-out trace) or a TosLink datastream. (The transport in both cases was a PS Audio Lambda.) The optical connection reduces the measured jitter from 321 picoseconds to a low 179ps, and eliminates all the narrow noise spikes in the spectrum. Data-related jitter itself (red numeric markers) is very low, but sidebands at the power-supply-related frequencies of ±60Hz, ±120Hz, and ±180Hz can be seen (brown markers).

Fig.9 Bel Canto DAC 1, high-resolution jitter spectrum of analog output signal (11.025kHz at -6dBFS with LSB toggled at 229Hz, CD data). PS Audio Lambda CD transport connected with 1m AudioQuest TosLink. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz. (Grayed-out trace is from the Lambda connected with 6' of Apature S/PDIF coaxial cable.)

The presence of low-level hum in the DAC 1's output was a disappointment, as was its sensitivity to noise pickup using an electrical feed. Otherwise, the Bel Canto's measured performance was excellent.—John Atkinson

Bel Canto Design
212 Third Avenue North, Suite 345
Minneapolis, MN 55401
(612) 317-4550