Digital Processor Reviews

I should add that I enjoyed hearing the specific aspects of the presentation I've described not for their own sake, but for their musical significance. Music listening was a more fulfilling, complete experience through the No.30. Going back to the example of the click sound being replaced by woodblocks of varying pitch, this minuscule objective change in the signal produced a huge perceptual difference. The No.30 revealed a new dimension of rhythmic interaction in this music. Further, there was suddenly an impression of Flora Purim, who plays percussion when she isn't singing, standing there playing with the band during the long instrumental breaks, rather than just vanishing. It was a different experience.

I could have chosen from among many examples of this remarkable quality. This one sticks in my mind for three reasons: I know the music so well, I've heard it through perhaps 25 digital processors, and significantly, it was the first piece of music heard through the No.30. I'm sure that there are as many examples as there are listeners.

Digital Comparisons
I'll wind up with a capsule comparison of the No.30 and two of the other big boys, the Stax DAC-X1t and Wadia 2000.

Because the Stax has no ST optical input, I used an Anodyne Group digital cable (Madrigal's electrical cable has an XLR on one end). I hadn't heard the Stax for some time, but from my memory, I thought the No.30 would be clearly superior. I was surprised by how close the Stax sounded in some respects—it is a superb processor. But the No.30 didn't have the magic I've just described; it was excellent, but not in a different league. Then I realized that I was listening to the No.30 through an electrical interface, not the ST optical through which I had performed most of the previous auditioning. I thus made three comparisons: the Stax, the No.30 with the same electrical interconnect used with the Stax, and ST optical. The difference between this particular digital cable and ST optical was thrown into sharp relief. The optical interface provided a smoother, more detailed presentation, with more of the qualities described above. The electrical interconnect brought the No.30 down a notch. I will therefore caution prospective buyers to be very careful in selecting a digital cable if the ST optical isn't used. To achieve the kind of performance I've been describing, I recommend avoiding the electrical interface altogether.

But back to the Stax. Its upper mids and treble were less liquid and silky than the No.30's, with a trace of hardness. Overall, the No.30 had a much more analog-like sense of ease and relaxation. In the bass, the No.30 had much better pitch definition, dynamic impact, and rhythmic drive. It was a closer call on soundstaging, but the Stax didn't match the No.30's remarkable image specificity and ability to separate disparate musical lines from the whole.

The Wadia had a very similar bass presentation to the No.30, with a satisfying power and solidity. The No.30 had the edge in dynamics, however, and was slightly tighter. Both processors threw a stunning sense of depth, with air and space surrounding instrumental outlines. The nod goes to the No.30, however, for its greater transparency and image specificity. Like the comparison with the Stax, the No.30 clearly had a better ability to separate individual instruments—spatially and texturally—from the presentation. The No.30 also had a smoother tonal balance, with a greater sense of ease. I also felt the No.30 to be less colored and more transparent, especially through the mids.

Reflections
The No.30's unparalleled performance suggests several generalizations about digital processor design. First, massive computer power, with faster and faster oversampling digital filters, appears not to be a prerequisite for state-of-the-art performance. An off-the-shelf digital filter—in thoughtful implementations—is not the limiting factor in digital playback musicality.

Second, 1-bit conversion techniques would seem to have a long way to go to match the performance of state-of-the-art R/2R ladder DACs like the UltraAnalog D20400 used in the No.30. I tend to think that 1-bit may never equal the performance of ladder DACs, especially in products where cost is secondary to sonic performance.

More important, the No.30 has changed my basic beliefs about digital audio. The hardness, homogeneity, and synthetic characteristic I believed to be inherent in the CD medium (or at least in today's A/D converters) are revealed by the No.30 to be instead primarily artifacts of digital playback processors. This is great news for all music lovers; although very few have the means to own a No.30, we now know that the CD medium is capable of much more than previously imagined. This realization was accompanied by a mixture of relief and joy: digital audio may not be intrinsically unmusical.

Now, if Madrigal would only build an A/D converter with the same dedication and engineering expertise lavished on the No.30. I'd like to hear a CD made through that converter.

Conclusion
The Mark Levinson No.30 Reference Digital Processor sets a new standard in digital audio playback. It provides a musical presentation significantly better in every respect than that of any other processor I've auditioned, regardless of design or price. I believe the No.30 to be so extraordinary that Stereophile's "Recommended Components" should be restructured to recognize this product's pre-eminent position. I therefore suggest that the No.30 be placed in Class A, and all other current Class A processors be moved to Class B. This is a bold step, but one I believe warranted by the No.30's superior performance. This is appropriate, however, only when the No.30 is connected via the AT&T ST-type optical input, or a very high-quality electrical interconnect.

I won't summarize the No.30's strengths in this conclusion—they're detailed in the body of the review and too numerous to reiterate here. As far as weaknesses, I'm at a loss to criticize the No.30. Although every other review I've written of digital processors has included a description of some shortcomings, I find it difficult to find anything wrong with the No.30. It's possible that I've been blinded to the No.30's flaws by its own sheer musicality. More likely, however, is that the No.30's transparency and lack of editorial interpretation leave no specific sonic signature that can be criticized.

In my review 18 months ago of the $12,000 Stax DAC-X1t (Vol.13 No.8), I cautioned prospective purchasers that "the DAC-X1t may be surpassed by a less expensive product within a relatively short period of time. One is therefore cautioned about investing in a five-figure digital processor during this period of rapid advancement."

I have no such reservations about the No.30. It is so far superior to anything else that it may be many years before other manufacturers can begin to catch up. Further, it may be difficult to achieve this level of performance without the No.30's cost-no-object approach. Competing products may thus one day match the No.30, but will probably still be expensive (footnote 4). Finally, I must reiterate the extraordinary build quality, superb user interface, and unprecedented flexibility in accommodating various digital formats. My only regret is that the No.30's astronomical price will limit its availability to relatively few music lovers.

Oh yes, I do have one other regret: I must return the review sample to Madrigal (footnote 5).

I shall miss it dearly.

---

Footnote 4: It is rumored that Madrigal is working on a less expensive processor (my guess: $6500) that incorporates many of the No.30's innovations, but in a more cost-effective package. The No.30's price could be substantially reduced by getting rid of the user interface, eight digital inputs, power-supply packaging (like the 1"-thick front panel), and other niceties.—Robert Harley

Footnote 5: I purchased this review sample and over the years upgraded it to first No.30.5 status, then ultimately to No.30.6 status, where it would decode hi-rez data. It served as my reference for digital sound quality through 2009, when, after almost 17 years of continuous use, the PLS-330 power supply failed.—John Atkinson