Bel Canto USB Link 24/96 USB-S/PDIF converter Follow-Up, May 2010

John Atkinson returned to the USB Link in May 2010 (Vol.33 No.5):

In May 2009 I reviewed Bel Canto's USB Link 24/96...I found that the USB Link 24/96 had quite a high level of jitter on its S/PDIF output, but with a modern DAC like the Benchmark DAC 1 or Bel Canto's own e.One DAC3, both of which have superb jitter rejection, this wasn't an issue. The USB Link 24/96 became a regular member of my system, and Stereophile writers Wes Phillips and Larry Greenhill are also fans. A drop in price from $495 to $249 at the beginning of 2010 made it an even better value, though it no longer comes with a Stereovox S/PDIF cable.

I used two different D/A converters to compare the Lindemann and Stello converters with the Bel Canto USB Link 24/96: the Benchmark DAC 1 and the older Musical Fidelity X-24K. I also ran the converters straight into the NAD M2 Direct Digital integrated amplifier. In addition, I compared each of them feeding the dCS Puccini SACD player ($17,000), used as a D/A converter, with the dCS Puccini U-Clock ($5000), which offers a true asynchronous USB input. All four USB-S/PDIF converters were connected with a 9" Stereovox datalink. Files were played with iTunes 9.0.3 running on my Mac mini. My comments apply to what I heard from the converters when used with all four DACs, but in general, the differences were smallest with the Puccini and largest with the Musical Fidelity. To my surprise, the differences with the Benchmark were larger than I expected, given that DAC's jitter-rejecting circuit topology. In absolute terms, the dCS U-Clock feeding the dCS Puccini DAC gave the best sound quality, though all three converters directly driving the NAD amplifier were not far behind.

As good as it sounded using 44.1kHz files, the Bel Canto's soundstage was a little flattened in direct comparison with the Stello and Lindemann, and its presentation had reduced LF extension and a slightly less-defined upper bass. The double bass on Astral Weeks: Live at the Hollywood Bowl, Van Morrison's 2009 revisiting of his classic 1968 album (from CD, Listen to the Lion 93423), was a little too generous in terms of LF bloom through the Bel Canto compared to the other two converters, both of which gave the instrument's sound better-defined leading edges. The same things were true for Nicky Scott's more reticently balanced bass guitar on Van's 1994 A Night in San Francisco (from CD, Polydor 314 521 289-2), which was just that bit better defined with the Lindemann.

Handling hi-rez audio, the Bel Canto has the advantage in that it will correctly process 88.2kHz data, whereas the Lindemann and Stello must rely on the host computer to up- or downsample it in real time to 96kHz or 48kHz. Either degrades the sound to a quality below that of good CD sound. But with 96kHz data, such as the Beatles' Love (from DVD-A, Apple/Capitol 3 79810 2 5), both the Stello and Lindemann converters gave excellent sound. Perhaps the Stello gave slightly more soundstage depth with the bird sounds that accompany the a cappella "Because." Perhaps. Perhaps the Lindemann was a bit better at differentiating the sound effects at the end of "I Am the Walrus." Perhaps. I am talking small differences here, but ultimately, with 96kHz files, my preference was still for, first, the Stello, then the Lindemann, then the Bel Canto. And it is fair to note that, with both the dCS and Benchmark DACs, the Bel Canto converting 24/96 data sounded more palpable than the Stello converting CD data.

Fig.1 Bel Canto USB Link 24/96, eye pattern of S/PDIF data output carrying 16-bit J-Test signal (±500mV vertical scale, 175ns horizontal scale).

When I measured the Bel Canto USB Link 24/96, I looked at the quality of its digital output by connecting it to one of the Audio Precision SYS2722's S/PDIF inputs. The jitter was a fairly high 5.85 nanoseconds peak, and the "eye pattern" of the data waveform correspondingly showed some uncertainty at its starts and finishes. I have had the SYS2722 upgraded and recalibrated since that May 2009 review; looking again at the Bel Canto's S/PDIF jitter, I now get 2.91ns peak, which is still fairly high, and the eye pattern still shows some blurring at the beginning and end (fig.1).

I looked at the effects of datastream jitter in the reconstructed analog signal with a Musical Fidelity X-24K D/A processor, sourcing 16-bit J-Test audio data from the Mac mini's USB output via the converters under test and connecting each to the Musical Fidelity with a 9" Stereovox datalink. The X-24K offers only moderate rejection of incoming datastream jitter and has a noise floor with each FFT bin at the –140dBFS level, about 6dB lower than the Compact Disc's theoretical limit.

Fig.2 Musical Fidelity X-24K, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, S/PDIF data from Bel Canto USB Link 24/96. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).

The spectrum of the Bel Canto USB Link 24/96's analog output showed significant broadening of the central peak that represents a high-level tone at exactly one-quarter the sample rate (fig.2), corresponding to the presence of low-frequency random variations of the sample rate; there were also sidebands of unknown origin at ±78 and ±122Hz. The Miller Analyzer calculated the jitter level to be a high 4.57ns peak–peak; as I wrote in my original review, it is important for the Bel Canto to be used with a DAC offering superb jitter rejection.

Summing Up: Having used the Bel Canto USB Link 24/96 ($249) for a year at the time of writing this review, I am very appreciative of its merits. But overall, it is outclassed by both the Lindemann and Stello converters. But neither the Lindemann nor the Stello can cope with 88.2kHz data, which the Bel Canto does handle correctly.—John Atkinson

Bel Canto Design
221 N. First Street
Minneapolis, MN 55401
(612) 317-4550