Mark Levinson No.31.5 Reference CD transport

If there is a component category that causes the "objectivists" in the audio community to splutter uncontrollably over their cups of herbal tea, it is the high-end CD transport. For in their "bits is bits" world, all a transport is required to do is recover the digital data from a disc—much like a grown-up cousin of your computer's $25 floppy-disk drive. The thought of paying up to $10,000 for something so humble—and, in their eyes, unnecessary—typifies what these blinkered folks regard as the insanity of the High End.

"If you hear a difference, it's because your D/A processor is broken or badly designed," they scorn. Yet when I first auditioned the original Mark Levinson No.31 transport in my system in early 1993, prior to its review by Robert Harley in Stereophile (footnote 1), the improvements it wrought—in the stunningly transparent presentation of detail, the delineation of the recorded soundstage, and in the sheer effortlessness of the musical message—were so noticeable, so large in a field where sonic differences can sometimes require much straining to identify, that I immediately reached for my checkbook. As far as digital sources were concerned, I lived happily ever after (at least for two years—see later), particularly after I'd purchased the magazine's original 1991 review sample of the Levinson No.30 processor and updated it to full No.30.5 status. (This gave Wes Phillips many opportunities to point out that my CD player cost more than most people's entire systems.)

To take the digital medium's theoretical perfection on trust strikes me as poor engineering practice. It appears that the real world of digital technology is far removed from the perfect world of mathematical theory. Single-box players can sound poor because of the impossibility of isolating the ultra-sensitive D/A circuitry from noise; while separating the transport and processor sections solves that problem, it raises a new one in the introduction of datastream jitter in the transport output, the data cable, and the DAC's word-clock recovery circuit.

But even as the "obs" still debate the existence of jitter and put down "subs" like you and me, engineering-led companies like Madrigal (and others) have been making great strides to eliminate it. Three years is thus a long time in digital, and at the 1996 Winter CES last January, Madrigal announced the No.31's successor, the No.31.5. Selling for $1000 more than the '31, the No.31.5 superficially looks identical. And, in fact, like all Mark Levinson "Reference" components, original No.31s can be upgraded to the latest status. A "performance only" upgrade, which replaces the transport mechanism, circuit board, one of the power supplies, EPROM, and remote, costs $2995. Replacing the lid assembly adds another $950, while a new faceplate, identical to the original but engraved with "No.31.5," adds another $750.

What's new?
The No.31.5's transport mechanism is mounted on the same sprung 12-lb lead-brick assembly as in the '31. However, whereas the No.31 used the Philips CDM 4 CD player mechanism, with its familiar swing-arm laser pickup, the '31.5 features Philips' new CDM 12 2x-speed industrial CD-ROM mechanism with a linear-tracking pickup and a digital servo. Because this is a CD-ROM drive, it won't by itself access the audio disc's subcode and index information. Madrigal therefore developed their own control software.

More important from a sound-quality point of view, they remove the Philips-supplied 1000ppm-accuracy oscillator from the servo board and replace it with a TCXO (Temperature-Controlled Crystal Oscillator). Spec'd at 5ppm and run at a 256fs rate, this is mounted on its own sprung printed circuit board next to the output. And even more important with respect to sound quality, the transport has changed conceptually. That master oscillator controls not the transport mechanism but the output stage, a flip-flop chip that could be considered a 1-bit FIFO. The output data thus have the maximum timing precision, which is where it is needed. The mechanical drive is slaved to the output stage and therefore needs to be able to respond to its demands, which is why a 2x drive mechanism, with its inherently fast response, comes in useful. (This topology, referred to by Madrigal as Closed-Loop Jitter Reduction, or CLJR, was developed for the less expensive No.37 and Proceed CDD transports.)

Footnote 1: RH's review appeared in June '93 (Vol.16 No.6), with Follow-Ups in Vol.16 Nos.9 & 11 and Vol.17 No.1. Refer to his original review for a full physical description of the Levinson transport.
Mark Levinson
P.O. Box 781
Middletown, CT 06457
(860) 346-0896