Digital Processor Reviews

The DAC3 uses a single PCM1792, though it should be noted that the chip can also be used in dual-differential mode (one per channel) to give 3dB lower noise and thus greater resolution. The DAC's four current outputs (left hot/cold and right hot/cold) are each transformed into a voltage signal by what appears to be a combination of a high-precision Caddock resistor (footnote 1) and a Linear Technology high-speed op-amp; this stage in turn feeds the output stage, which is based on high-quality Burr-Brown OPA1632 differential-output op-amps.

Sound
I mainly used the e.One DAC3 taking an AES/EBU digital feed from my Ayre Acoustics C-5xe universal player and a TosLink feed from my Slim Devices Transporter network player. I auditioned only the balanced outputs, mostly using them set to Fixed Level.

Though not to the standard set by the Ayre playing well-recorded SACDs such as our May 2006 "Recording of the Month," Daniel Gatti and the Royal Philharmonic performing Tchaikovsky's Symphony 6 (Harmonia Mundi HMU 807394), CDs played back through the Bel Canto had silky-smooth high frequencies. The violins on the CD layer of the Tchaikovsky disc were as natural-sounding as I've heard from 16-bit digital. Even recordings whose high frequencies have been definitely overcooked, such as Queen's Live at Wembley '86 (CD, Hollywood HR-61104-2)—great music in dreadful sound—sounded smoother than I'd expected.

At the other end of the spectrum, low frequencies didn't have quite the extension or the impact of the Levinson No.30.6, but that's hardly surprising: the long-discontinued Levinson, with its massive separate dual-mono power supply, remains unrivaled in that area, in my experience. But I had to give the nod to the far-less-expensive Bel Canto when it came to midrange and treble purity. There was a sweetness to its presentation that allowed the music to communicate more directly.

This hadn't been achieved by the processor smearing detail or muffling the highs. In fact, the DAC3 dug deep into recordings with which I am well familiar to bring into contrast aspects of the mix that I had recognized without fully appreciating. For example, in the coda to "Beyond and Before," from Yes's eponymous 1969 debut album (CD, reissued as Rhino R2 73786), reverb tails appeared on the guitar and organ, with the result that the images of those instruments moved noticeably back in the soundstage. On the same album, there was the subtle doubling of Tony Kaye's piano figure by vibes in the final measures of "Yesterday and Today," and the peculiar, hi-hat–dominant balance given Bill Bruford's swing-tinged drums on the cover of the Byrds' "I See You." All were laid bare, but without the feeling that anything was being unnaturally spotlit.

It was in its reproduction of soundstage width and depth that the Bel Canto excelled. In fact, I heard more reverberation than I'd expected from recordings I thought I knew well. The string players on the Mozart Flute Quartet movement on Editor's Choice (CD, Stereophile STPH016-2) were set farther back in the stage than I was expecting, as were Billy Drummond's drums on "The Mooche," on the same CD. Particularly when I replaced the Parasound Halo JC1 monoblocks with the Boulder 860 amplifier, the players on Attention Screen's Live at Merkin Hall (CD, Stereophile STPH018-2) were set back slightly more within the subtle hall acoustic. And pianist Bob Reina's sotto voce mumblings during Mark Flynn's drum solo at the start of "Blizzard Limbs" were clearly audible. (Who was he talking to?)

Was this aspect of the DAC3's presentation enhanced accuracy or an exaggeration? I suspect the former. On my recording of Mozart's Piano Quartet in g from Bravo!, a selection of performances from the 1998 Santa Fe Chamber Music Festival (CD, Stereophile STPH014-2; the final movement also appears on Editor's Choice), the installation of a noisy air-conditioning system in St. Francis Auditorium meant that I'd had to abandon my purist philosophy and close-mike all the instruments. In the mix, I therefore had to add a judicious amount of Lexicon-sourced artificial reverberation. I had tried hard to match the character and level of that ambience to that of the actual hall, in which I had also run a pair of distant mikes. The Bel Canto, however, revealed the artifice, the slight disparity between the signature of the artificial reverb and that of the hall being more apparent than I had wished—or than I had experienced when mixing and monitoring this album with the Levinson DAC.

Against the benchmark Benchmark
The Benchmark DAC1 has set the standard for affordable D/A processor performance for three years now, and has recently been updated to add a USB input. I will be reporting on the new version in a Follow-Up, but for this review, I compared the Bel Canto e.One DAC3 with the original version of the Benchmark DAC1, which I had purchased after writing about it.

To permit instantaneous comparisons, I fed the Ayre C-5xe's AES/EBU output first to the Levinson No.30.5, then from that processor's two AES/EBU data outputs to the two DACs under test using identical lengths of Madrigal AES/EBU cable. Levels were matched to within 0.05dB at 1kHz by keeping the Bel Canto's output at its maximum and reducing the Benchmark's with its analog level control.

No doubt about it, the Benchmark DAC1 is still an excellent-sounding product, with well-extended, well-defined lows, a natural midrange, and clean highs. And, of course, it doubles its utility by having two pairs of headphone outputs. But the Bel Canto DAC3 scored with its slightly silkier high frequencies and its wider, deeper soundstage.

Promise fulfilled?
The real promise of a product such as the DAC3 is that it can be used to make silk purses out of the audio world's sow's ears: the host of el cheapo DVD players that may be able to pull the bits off of discs but have sonically compromised analog output stages and jittery digital outputs.

A few years ago I purchased a Pioneer DV-578A universal player ($150 at the time, now discontinued) to use as a benchmark in this arena. It's a serviceable player, but I'd been disappointed when I tried using it as a transport, even with the Benchmark and Levinson DACs, both of which have excellent measured jitter rejection.

I've mentioned in other reviews how much I've been enjoying David Gilmour's On an Island (CD, Columbia 2876-8028-2), released in 2006. "Red Sky at Night" has an addictive slow-drag feel; played on the Ayre C-5xe feeding the Bel Canto via an AES/EBU link, this track produced a wide, deep soundstage, with such details as Georgie Fame's Hammond-organ riffing and the room reverb on the snare drum well distinguished. Again to my puzzlement, given the DAC3's superb measured jitter rejection, when I played back the same track on the Pioneer hooked up to the Bel Canto with 10' of AudioQuest VSD-4 S/PDIF coax, the soundstage flattened a little and the bass didn't reach quite as deep. If I had to swear on it, however, I thought the Pioneer-AudioQuest combo was actually a slight bit better in ultimate definition.

Replacing the high-quality AudioQuest S/PDIF cable with the worst datalink I had to hand, a 15' length of generic plastic TosLink, gave a sound quality from the DAC3 that was hard to distinguish from the electrical connection. Again, if I had to swear on it, the balance using the Pioneer transport was perhaps slightly lighter overall with both the electrical and optical datalinks.

For my casual listening these days, I use a Slim Devices Transporter to access my iTunes library. It's definitely overkill to feed the Transporter's data output to the Bel Canto; the Transporter's DACs and analog circuitry are in the same league as the DAC3's. But, my goodness, feeding the TosLink output from the bargain-priced Squeezebox to the Bel Canto resulted in a sound that kept distracting me from the computer.

Computer Music
To assess the Bel Canto's sound quality from its USB input, I connected it to my five-year-old Apple TiBook, which recently survived a hard-drive replacement. Selecting the USB Audio Codec as the computer's default audio output in the System menu and playing back 44.1kHz-sampled files with iTunes, the sound was accompanied by what sounded like FM "birdies"; ie, silence when no music was playing, but a random whistling when it was. I couldn't eliminate this no matter how I tried—wondering if it was a grounding issue, I tried using the laptop in battery-powered mode, but no luck. So I switched to driving the DAC3's USB input from a more recent computer: the Mac mini that serves (ha!) as my household's music server.

Now the Bel Canto performed without audible problems. However, while the sound quality of the DAC3 used in this manner was okay, it wasn't in the same class as the processor fed data via its conventional digital inputs. The soundstage was flatter, less dimensional, the music less involving, the high frequencies less silky. Despite the more complex data-transmission path, I preferred the presentation of the same files played back via WiFi and the Squeezebox, again feeding the latter's optical output to the Bel Canto D/A.

Summing Up
I so wish that Bel Canto's e.One DAC3 had been priced closer to $1250 than to $2500. The former would make it a no-brainer credit-card purchase for almost any audiophile wanting digital playback quality that is the state of the art ca 2007. At $2495, its purchase should be a more considered decision. But its superb sound quality, excellent engineering, and the fact that its volume control makes a system preamplifier optional, all make the DAC3 a must-have product. Erick, you were so right!

---

Footnote 1: In some ways, a single low-value resistor to ground is the ideal means of converting a current signal to a voltage one. However, it will introduce noise due to the need for the resistor to have a finite value, which is why you almost never see this topology used. As the DAC3 has a very low analog noise floor, they must be doing something different.