Digital Processor Reviews

How does the new Mark Levinson No.35 compare with its slightly senior, somewhat more sophisticated, and much more expensive older brother, the No.30? The latter has not lacked for well-earned accolades extolling its superb performance but in no way restrained by its wallet-numbing, precedent-bursting price and sheer physical presence. Not that the No.35 is exactly shy and retiring in appearance; but the latter's smaller size and single-chassis construction conspire to make it just a bit less awe-inspiring.

We still had a No.30 on hand, so a comparison merely required prying it away from RH. Not easy—I'd already separated him from the No.31 for the duration of this review—but necessary. With both processors in my listening room, both driven by Levinson's own transport, the games began. It was humbling to realize that it probably cost less to build the Stereophile listening room itself than it would cost to purchase these three Levinson pieces.

Before I could begin, however, another problem raised its shaggy head: which digital cable to use. The obvious choice seemed to be to continue to use the Kimber, at least initially, for continuity's sake. But that was not easily done; the No.30 does not have a dedicated S/PDIF (RCA-type) coaxial digital input. Madrigal does have adapters to enable the use of one of the AES/EBU inputs in this mode, but I didn't want to contend with that added variable.

The choice had to be between optical ST and AES/EBU. I listened briefly to the No.35 with Levinson's own AES/EBU cable connecting it to the No.31 and definitely preferred the sound of the system with the Kimber AGDL. With the AES/EBU, the sound became slightly more bloated (those liking it may call it bloom), slightly fuller, and definitely more forward. There was less apparent depth and dimensionality to the soundstage, less overall focus.

Since the Kimber was not a candidate for the No.35/No.30 face-off, I turned back to the AudioQuest ST optical link. Comparing it briefly again to the Kimber, I still felt a marginal preference for the Kimber, but the ST was definitely closer than the AES/EBU, and perhaps, on further reflection, superior in some ways to the Kimber (a bit sweeter on top though with a bit less detailing overall). ST it would be.

Now I was ready to compare the No.35 with Levinson's flagship No.30. The rest of the system remained unchanged, including the TARA Labs Master RSC unbalanced interconnects between the processors to the Rowland Consummate preamplifier.

To my surprise, I found that I preferred the less expensive No.35 to the No.30. The choice was hardly clear-cut, with the No.31/No.30 making a powerful case for itself. But I found the No.35 to have, in my system, the tighter, more detailed sound. The No.30 certainly didn't lack for detail, but I found it to be, by a thin margin, the softer and sweeter of the two processors, rendering a slightly cleaner sound on many recordings but doing so at a slight loss in apparent upper-end extension. I went back and forth in my preferences, leaning to the No.30 on some discs, the No.35 on others. The No.30 sounded more liquid on the sometimes cool and dry Jurassic Park soundtrack (MCA MCAD-10859), but the No.35 was the more visceral, with a marginally tighter, better-defined bottom end. Nevertheless, I'd be hard pressed to prattle on about the differences.

On "Jolie Louis" (Music from Northern Exposure, MCA MCAD-10685), excerpted from Daniel Lanois's Acadie (Opal 25969-2), the No.35 brought out the more interesting, finer shadings and natural textures of Lanois's voice. This impression was reinforced by other first-rate vocal recordings. Again, however, neither processor clearly took the upper hand. Comparing them using the usual audiophile checklist was frustrating.

Which, in the long run, was the more accurate? I cast my vote with the No.35, but not without a lot of soul-searching and hairsplitting. There was, I feel, a larger difference between AES/EBU and ST interconnects than between these two top-rank D/A processors. Further listening only reinforced this opinion, but also added to my marginal preference for the No.35. On Britten's Noye's Fludde (Virgin Classics VC 7 91129-2), the No.35's sharper sound was not necessarily a plus on this slightly bright recording, but its clean grainlessness helped considerably in minimizing any recording-generated irritation. The No.30, again, came across as the warmer-sounding unit. I also began to feel, as the comparison wore on, that the No.30 sounded somewhat more forward than the No.35, particularly in the upper midrange/lower treble. At the same time, the No.35 came across as more dynamic—punchier and tighter overall.

To ensure that a change to balanced output cables would not dramatically alter these conclusions (recall that, up to this point in the evaluation, I was using an unbalanced processor-to-preamp link), I briefly compared the TARA Labs unbalanced with a pair of AudioQuest balanced Lapis interconnects between the No.35 and the preamp. (Unfortunately, the added variable of a different brand of cable was unavoidable, as balanced TARA Labs RSC were not on hand.) The Lapis had a bit less bloom and a bit more brightness than the Master RSC. I continued to prefer the latter—not by a huge margin, but by enough to continue using it for the remainder of the tests.

Again, although my preferences wavered a bit depending on the recording, I always seemed to come back to the No.35. I'm just as certain, however, that, given the nature and the degree of the differences, another listener might well prefer the No.30. Both D/A processors are superb devices. Given the gap in their prices, however, no one shopping for the No.30 should automatically assume that they will find it to be sonically superior; make the comparison for yourself. Matched levels in such a test will be no hindrance; both processors have identical output levels.—Thomas J. Norton