CD Player/Transport Reviews

The Madrigal-developed control software for the mechanism runs on a combination of an Analog Devices DSP chip and a Xilinx field-programmable device (FPD). The DSP chip is also used to process the audio data—as, once the timecode, track and index information has been read, its presence in the output data potentially increases jitter. Not only are this data stripped out from the output, but the subcode bits in each data frame are replaced by random bits generated by the DSP (footnote 2).

Madrigal had first tried calculating the appropriate bits for each frame to make it DC-neutral (as originally suggested by Dunn and Hawksford in their 1992 AES paper, reprinted in the March '96 Stereophile), which will inherently minimize jitter. Surprisingly, this was found not to sound as good as adding random data, perhaps because the calculated extra bits, while giving the lowest measured jitter, are still correlated with the signal data. The random data will give slightly higher residual jitter in the AES/EBU interface, but this will be completely decorrelated. (Note that the audio data recovered from the CD, bits 1 through 16, are left untouched by this process.)

How is the world treating you?
The first thing you notice when you audition the No.31.5 is the sound, not of the CD, but of the lid. Gone is the idiosyncratic whirring noise. Instead, the new lid lifts noiselessly, driven by a high-speed, low-torque motor via a 64x reduction gear. Its speed is also under software control: it gracefully slows down near the two ends of its travel. It also opens and closes if gently pushed.

The second thing you notice is the remote, which is the same size and shape as the '31's, but adds volume and source-selection buttons. For a transport? The new remote and programming allow the No.31.5 to control the volume level and source of a No.38 or '38s preamplifier in the system as well as selecting between the digital sources connected to a No.30.5. The volume and source settings are briefly shown on the transport's alphanumeric display. The preamp can thus be sited out of sight, if so desired. This simple ergonomic improvement proved mighty valuable in the listening room.

But sound quality is what matters most to this magazine's readers, and therein lies a tale. As impressive as I had found the original No.31, after about two years of living with it I found myself experimenting with outboard jitter-reduction units. In chronological order, I tried the Digital Domain VSP, the Sonic Frontiers UltraJitterbug, the Audio Alchemy DTI•Pro and DTI•Pro 32, the Meridian 518, and the Genesis Digital Lens. The sonic changes wrought by these devices were mainly positive, though the effect of the original DTI•Pro's enhancement algorithm was very dependent on the CD being played, and sometimes made the sound worse.

In general, the palpability of the No.31's sound was made greater by these outboard devices: the midrange sounded even more natural, and the highs were better differentiated. Cymbals, for example, sounded less noise-like, and more textured in the manner of the real thing. The two best-sounding units were the most recent, the Meridian and the Genesis, when both were set to redither to a 20-bit data word length. While the Digital Lens gave the greatest improvement in sound quality, this came at a price: the Genesis made the No.31's always-authoritative bass sound more bloated, less tight. As Martin Colloms would phrase it, the Genesis reduced the '31's sense of pace.

Ivo Papasov and His Bulgarian Wedding Band's Orpheus Ascending (Hannibal HNCD 1346), for example, has been spending a lot of time spinning in the '31 in recent weeks. The bass guitar that drives the crazy combination of clarinet and rhythm guitar—or "structural" guitar, as the Dead's Bob Weir calls it—through the music's angular and irregular time signatures lost some of its essential forward momentum even while the midrange-dominant instruments, the snare drum in particular, gained palpability.

As much as I enjoyed the sound of the No.31, I guess I knew that there was more to be gotten from digital, as listens to vinyl reminded me. I set the No.31.5 up with anticipation: Would it offer the improvements to the realism of the midrange and treble that I had become used to from the Digital Lens while preserving the No.31's sense of pace and bass authority?

In a word, yes. The No.31.5 gave me the best of both worlds. It only took me about five seconds of Orpheus Ascending to realize that with the '31.5 recovering the bits, and without the Genesis, I was hearing the best I have yet to hear from the 16-bit CD standard. Yes, Michael Fremer, it's still not vinyl, but it's damned close. And when there is great music only available on CD—Charlie Watts's new jazz standards collection, Long Ago & Far Away (Virgin 8 41695 2), for example—the Levinson No.31.5 driving the '30.5 relieves the pain of vinyl withdrawal and eliminates the strain.

Lovely as ever, i must admit
With a high-quality audio version of DVD looking to be years away, I see no reason not to invest in a high-performance CD transport (especially as I don't see it as too difficult a task to upgrade the '31.5 to take a DVD mechanism). And the No.31.5, a "reference" component if ever I heard one, defines "high performance." However, it is fair to point out to owners of the No.31 that even the cheapest upgrade to 31.5 status, replacing as it does everything other than the chassis and half the power supply, is quite expensive at a hair under $3000. By comparison, the $1800 price tag of the Genesis Digital Lens or the Meridian 518 looks tempting.

But you've guessed it: I've ordered the upgrade kit for my No.31. The combination of improved sound and greater functionality/versatility in a Levinson-based system proved impossible to resist. I can fully recommend the No.31.5 as a supreme example of a state-of-the-art, potentially future-proof CD transport. And let the obs be damned!

---

Footnote 2: Four years after this review was published, I examined the validity of this claim using RME's DIGICheck program, which shows the status of every bit in the frames that make up the digital audio datastream. As fig.3 on that page shows, it appears that the Levinson transport does not make the bottom 8 bits active in the manner I described in my 1996 report.—John Atkinson