Sonic Frontiers SFD-2 D/A processor Page 2

I was also impressed by the SFD-2's lack of treble etch and grain. The presentation was bright, immediate, and full of detail, yet not fatiguing. In fact, the SFD-2 has the cleanest, most pristine treble I've yet heard from digital. There was no trace of a hashy character overlaying treble textures or of cymbals that splash over the soundstage; instead, the treble was as pure and clean as mountain spring water. The Gen.III's treble seemed hard and grainy next to that of the SFD-2.

Compared with the No.30, the SFD-2 seemed to have more lower-treble energy, but a slightly softer—and cleaner—treble. The SFD-2's upper midrange tended to be a bit prominent and the upper treble slightly subdued. The snare drum on the Michael Ruff disc had more air, snap, and immediacy through the SFD-2, although the cymbals were more forward through the No.30. The SFD-2's brighter rendering infused the presentation with a sense of air and openness that made the soundstage bloom. The combination of brightness and detail without etch or grain seemed to allow more of the music's life to emerge. Moreover, the treble was well integrated with the rest of the spectrum. I never got the feeling—as I do with lesser processors—that the treble was a separate component riding on top of everything else.

Midrange textures, however, were not quite as liquid as those heard from the No.30. The SFD-2 was more forward in the mids, and lacked the senses of ease and lushness that characterize the No.30's interpretation. In addition, the SFD-2's overall perspective was more up-front than that of the No.30. The SFD-2's combination of slightly harder timbre and a more forward balance in the mids made the SFD-2 more aggressive than the No.30. Although the SFD-2 had a startling clarity in the mids, the presentation tended to be immediate and incisive rather than relaxed and gentle. On some music, the SFD-2's greater sense of immediacy and palpability outweighed the slightly less velvety rendering—rock and blues, for example. But on naturally miked acoustic instruments, the No.30's more laid-back perspective was more musically satisfying—and less fatiguing.

The Gary Schocker, Flutist CD (Chesky CD46) was a good example. Through the No.30, the musicians were set farther back in the soundstage, with tighter image focus; I had more of a feeling of being drawn into the music than of the music coming at me. On some classical music—particularly chamber music—I preferred the No.30's ease and liquidity over the SFD-2's greater incisiveness. I reiterate that this description relates to the midrange; I found the SFD-2's treble extraordinarily clean—a significant factor in a converter's musicality (footnote 3).

The SFD-2 continued to surprise; its soundstaging was more transparent and holographic than even the Theta Gen.III. The SFD-2 threw a vast sense of space before me, with a beautiful feeling of air between instruments. Instrumental images floated in three-dimensional space, completely detached from the loudspeakers. Moreover, the images were surrounded by a delicious bloom that added an impression of life and palpability. Recorded music often lacks this dimension, although analog comes much closer than digital to reproducing it. Most digital sound has a flatness and sterility that detract from the impression of instruments existing in the listening room. Through the SFD-2, I heard an amazing degree of this hard-to-define sense of life from the music.

The Sonic Frontiers also resolved an abundance of spatial cues. Reverberation decay was particularly well portrayed, with a smooth, distinct decay that seemed to hang in space for a long time. These characteristics made the SFD-2 particularly adept at conveying the recorded acoustic and presenting a stunning sense of depth. Moreover, the SFD-2 superbly separated the instrumental image from the hall reverberation. Some processors fuse an instrument's direct sound with that of its reverberation, making the overall sound a modified version of the instrument's. The SFD-2 presented instrumental images as existing within an acoustic space, with the hall sound spatially differentiated from the instruments' sounds.

The excellent recording of Ulrike-Anima Mathé playing Max Reger's Violin Sonata in D (Dorian DOR-90175) dramatically illustrated just how well the SFD-2 resolved the recorded acoustic. The solo violin was bathed in the rich acoustic of the Troy Savings Bank Music Hall, yet retained its distinct spatial outline. The instrument was surrounded by a palpable feeling of air and bloom, conveying the impression of being transported to the hall. In conveying space, the SFD-2 was nothing short of stunning.

The SFD-2's image focus was less tight and sharp than those of the three other processors. Instrumental and vocal outlines tended to be slightly diffuse rather than razor-sharp. The No.30 was better able to convey correct image size, changing from tight and compact to large and billowy, depending on the recording. The SFD-2 had less range in presenting image size, suggesting that the No.30 more accurately reveals what's on the recording—even though the SFD-2 revealed more space and depth. The No.30 also had slightly finer resolution of front-to-back depth, with more gradations of distance.

The SFD-2's portrayal of dynamic contrast and the sheer physical force of music was the best I've heard from digital replay. Even compared to the Theta Gen.III, the SFD-2 revealed a greater range between loud and soft, and more impact on high-level transients. This sense of slam and power combined with its first-rate bass reproduction to make the SFD-2 extraordinarily involving rhythmically. The SFD-2 was a paradigm of Martin Colloms's description of pace and rhythm in Vol.15 No.11.

Although the SFD-2 didn't congeal and thicken during loud passages, it did have a tendency to lose some liquidity at high signal levels. Instrumental textures became a little hard and grainy when the music got loud. In this regard, the No.30 was clearly better than the SFD-2; the No.30 maintained its composure when confronted with loud passages.

Despite the slight loss of liquidity, the SFD-2 had a remarkable sense of transient coherency. It was as if the music's dynamic structure was lined up in time, giving the presentation a "tight" feeling. The rhythm locked in, with the music's drive powerfully conveyed. The SFD-2's "tightness" was like a perfectly tuned car engine: no hesitation, lag, or reticence. The whole presentation had a feeling of cohesion and musical rightness, rather than sounding smeared or made up of separate components. These qualities conveyed an excitement in the music—an "up" feeling—that I found most satisfying.

In an overall assessment of the SFD-2 compared to the Theta Gen.III, it wasn't even a horse race. The SFD-2 had a vastly cleaner treble, more liquid textures, finer resolution of detail, better pitch definition in the bass, and a more musically involving presentation.

The comparison with the No.30 wasn't so easy to define. The SFD-2 and No.30 sound very different: the SFD-2 is more dynamic, open, "up," incisive, and rhythmically involving than the No.30; it reaches out and grabs you. Conversely, the No.30 is more polite, refined, subtle, understated, and less visceral. Although I would choose the No.30 for some music, I preferred the SFD-2 on a significant portion of the music I most enjoy. That says a lot about the SFD-2's overall musicality—and is quite an achievement for a $4695 processor.

These listening impressions were all made with the SFD-2's balanced outputs. I then compared the balanced and unbalanced outputs by driving an Audio Research LS2B preamp in both modes at matched levels. The unbalanced interconnect was a 3' pair of AudioQuest Lapis; the balanced interconnect was a 4' run of Lapis. The LS2B's balanced input is the "Direct" input, which bypasses the input switching for a shorter signal path, giving the balanced outputs a slight advantage.

I found that the SFD-2's musical magic disappeared when I listened through the single-ended outputs. The treble became less clean and pure, instead sounding brighter, hashy, and a little metallic. The bass drive I'd so enjoyed wasn't nearly as powerful, and the soundstage lacked the expansiveness and bloom heard from the balanced outputs. This unsubtle difference made me yearn to go back to balanced outputs. Although the No.30 sounds its best balanced, and the Theta Gen.III is almost a different converter balanced, the SFD-2 lost more than either of these processors when switching to unbalanced outputs. The SFD-2 was still a good processor unbalanced, but not a stunning one.

If you audition the SFD-2, be certain you're hearing it in balanced mode. If you buy one, don't even think of using its single-ended outputs—it's that much better balanced.

I've rarely been as excited about a product as I am about the Sonic Frontiers SFD-2. Its sound is clearly of Class A caliber; it is perhaps the best-sounding digital processor I've heard—regardless of price. What makes the SFD-2 so exciting is that this level of digital sound quality won't cost you as much as a decent car. Although the SFD-2 isn't inexpensive, it nevertheless is much more affordable than the five-figure processors that previously defined the state of the art (footnote 4).

The SFD-2 is, however, not nearly as lavish, elegant, or functional as the Mark Levinson No.30. In fact, the SFD-2 is rather plain, and its tiny front-panel toggle switches do not befit a product of this technical and musical sophistication. Looking at the two units and considering their "feel," features, and function, the No.30 is light-years ahead of the SFD-2—as it should be, considering its much higher price. If you want the level of opulence provided by the No.30, by all means buy the No.30. Although they are different sonically, both the SFD-2 and No.30 represent digital replay at the highest level.

I won't reiterate the SFD-2's sonic strengths—and minor faults—here; you'll have to read the full review. But I will say that the SFD-2 is a stunning achievement in digital processor design, and one that probably won't be topped until a processor equipped with the High Definition Compatible Digital (HDCD) chip comes to market in late 1994.

The SFD-2 should turn the competitive digital processor market on its head. I'll look forward to hearing everyone else's attempts to top the extraordinary Sonic Frontiers SFD-2.

Footnote 3: Even in its unbalanced mode, I felt the Sonic Frontiers DAC to have the most natural-sounding massed violin tone I have heard from digital replay, with only the Mark Levinson Nos.30 and 35 coming close.—John Atkinson

Footnote 4: Sonic Frontiers is working on a less expensive processor, the $1995 SFD-1. The SFD-1 is essentially a scaled-down version of the SFD-2, with the AES20 input receiver, one UltraAnalog D20400A DAC, and the same tubed output stage. The unit will have a less extensive power supply, lower parts quality, and won't be balanced in the digital domain. The ST-type input will also add $200 to the price.—Robert Harley