Digital Processor Reviews

Robert Harley on the No.30.5 HDCD in April 1995 (Vol.18 No.4):

For more than three years, the Mark Levinson No.30 Reference Digital Processor has been the standard against which other processors have been judged. In addition to serving as a reference at Stereophile—both JA and JE have purchased samples—countless competing manufacturers mention their products' performance in relation to that of the No.30. I can't tell you how many times I've heard the boast "It's as good as the No.30." Few products have enjoyed such a universally acknowledged position of sonic superiority, beautiful build quality, and thoughtful user-interface.

Since my original review in February 1993 (Vol.15 No.2), the No.30 has been upgraded to the No.30.5, an action that has kept the Levinson processor competitive in the rapidly improving digital-processor world (footnote 1). When the No.30.5 upgrade was being designed, the Pacific Microsonics PMD100 HDCD decoder/filter chip wasn't quite a commercial reality. Madrigal made the decision to launch the No.30.5 without HDCD rather than to continue waiting for the HDCD decoder/filter.

Madrigal engineers did, however, hedge their bets on HDCD by making the No.30.5 HDCD-upgradeable—the modification is simply a matter of replacing two socketed chips inside the processor. A small board containing the Pacific Microsonics PMD100 HDCD decoder/filter replaces the NPC SM5842 digital filter, and an EPROM containing control software is swapped for one with updated code. The upgrade costs less than $100 (the exact price wasn't available at press time), plus any labor charge the dealer may add. Because the upgrade is so easy to install (it's about a 30-minute job), customers can buy the kit and do it themselves. All No.30.5s now sold have the HDCD upgrade included.

A provision of the HDCD license calls for 6dB of attenuation in the processor when playing non–HDCD-encoded discs, or 6dB of gain applied to HDCD-encoded discs. Without this gain matching, HDCD-encoded discs would sound quieter, because they have a wider dynamic range (a higher peak:average ratio), which is perceived as sounding less loud. You can imagine an uninformed consumer comparing HDCD to conventional playback in a mass-market store, and deciding that HDCD isn't an improvement because it doesn't sound as loud.

This 6dB of attenuation for standard CDs can be performed in the digital domain by the PMD100 decoder/filter. Unfortunately, attenuating the signal with this method throws away one full bit of resolution. A better alternative is to perform the attenuation in the analog domain. This would have been very difficult in the No.30.5, because the towers containing the analog audio circuits are thermally sealed and difficult to access. Moreover, the No.30's output stage topology wasn't designed with gain or attenuation in mind.

Madrigal has solved this dilemma with a clever trick. As shipped, the No.30.5 is similar to other processors using the PMD100 decoder/filter chip in that it performs the attenuation digitally, meeting the terms of the license agreement. If, however, you use the No.30.5 with a Mark Levinson No.38 or No.38S preamplifier and connect them with the communication link, the preamplifier's gain is automatically boosted by 6dB on HDCD-encoded discs. All you need to do is select the "No.30" alias (the source name shown on the No.38's display), and the 6dB boost is added automatically in the preamp. To show you that this gain adjustment is working, the No.38's display briefly shows the legend "HD+6." This technique avoids the sonically degrading 6dB of attenuation imposed by the filter. Owners of older No.38 or No.38S preamps must upgrade the No.38 software for this automatic feature to work.

If you use the No.30.5 with another preamplifier that lacks the No.38's automatic gain-adjustment function, you can still avoid the 6dB of digital-domain attenuation by putting the No.30.5 in the "Manual" mode. When in the "Set-up" mode on the No.30.5, press the polarity inversion button until the legend "Manual" shows in the No.30.5's display. This defeats the 6dB of attenuation provided by the digital filter on conventional CDs. In theory, you're then supposed to turn up the volume on HDCD discs by 6dB. But if you don't, I doubt that No.30.5 owners will mistake a wider dynamic range for poorer musical performance.

No matter what preamplifier you use, avoiding the 6dB of attenuation in the digital filter is essential to getting the best sound from the No.30.5. If you audition the No.30.5, be certain it's been set to "Manual."

The PMD 100 HDCD decoder/filter has eight output dither options to optimize the DAC's performance (see the sidebar to the Sonic Frontiers SFD-2 Mk.II Follow-Up in last month's Stereophile). Madrigal chose not to use any output dither in the No.30.5, perhaps because the rest of their circuitry was designed for a signal without dither.

Other than the filter and control software, the review sample of the No.30.5 with HDCD was the same unit that I reviewed in the October 1994 Stereophile.

System: I evaluated the HDCD-equipped No.30.5 in my reference playback system, and compared it to the Sonic Frontiers SFD-2 Mk.II, another processor that uses the HDCD decoder/filter. The No.30.5's balanced outputs fed an Audio Research LS5 Mk.II preamplifier through AudioQuest Diamond x3 interconnects, and I fed the No.30.5 with the AES/EBU output from the Mark Levinson No.31 Reference CD transport.

Power amplifiers were Audio Research VT150 tubed monoblocks, which drove the ribbon midrange and tweeter panels of Genesis II.5 loudspeakers. The II.5's four 12" woofers were driven by the Genesis servo amplifier. Other interconnects included AudioQuest Lapis, and loudspeaker cables were AudioQuest Dragon II. Power to the preamplifier, transport, and digital processors under audition was conditioned with a Tice Powerblock, and an MIT Z-Stabilizer was plugged-in to the AC outlets (footnote 2).

Listening: With the HDCD decoder/filter installed, the No.30.5 retained the characteristics that have made this processor a long-term reference, while significantly improving the sound in several key areas. The following impressions were made with conventionally coded CDs, and with the No.30.5 in the "Manual" mode.

Replacing the NPC filter with the PMD100 produced a startling increase in soundstage transparency, image specificity, and the ability to keep individual lines separate. The PMD100-fitted No.30.5 had a remarkable resolution of quiet instruments—even in the presence of much louder signals. Consequently, the music had an ease I didn't hear from the NPC-fitted No.30.5. It was as though all the music was presented effortlessly, and I didn't have to strain to hear low-level detail. Further, the No.30.5 now had a terrific sense of space between instruments, with a halo of air around instrumental outlines.

Another improvement in the HDCD-fitted No.30.5 was in the portrayal of instrumental textures. The new No.30.5 produced a greater sense of realism and palpability, rendering the sound of the NPC-based No.30.5 slightly sterile and mechanical by contrast. Massed violins had more smoothness, grace, and ease through the new No.30.5, with less edge and hardness. It sounded as though the harmonics worked together to produce the beautiful, silky, liquid sound you hear in the concert hall from the violins, rather than conflicting with each other to produce a sound that makes your ears close down.

When playing HDCD-encoded discs through the No.30.5, these qualities were simply staggering—and far better than vinyl. But what surprised me was the PMD100's ability to improve conventionally coded CDs in these important aspects of reproduced sound.

The No.30.5's bass was fabulous, with tautness, extension, and articulation. Although the bass wasn't quite as powerful and rhythmically involving as that of the Krell KPS-20i (see my review elsewhere in this issue), it was significantly better than the bass of the Sonic Frontiers SFD-2 Mk.II. The No.30.5 had a leaner, tighter, deeper, and better defined low-frequency presentation, while the Mk.II had more midbass weight and bloom, at the expense of tautness and the ability to convey bass detail. The No.30.5 had better extension in the very bottom end, with more impact and solidity. The new No.30.5 was also more dynamic than the Sonic Frontiers, but didn't have quite the range of dynamic expression I heard from the KPS-20i when the No.30.5 was decoding conventional CDs.

Last month, I compared Sonic Frontiers' $5295 SFD-2 Mk.II tubed processor to the non–HDCD-encoded No.30.5, and found the Mk.II sonically superior on both conventional and HDCD-encoded discs. With the addition of HDCD to the No.30.5, that situation has reversed, with the Levinson processor taking the lead. Specifically, the new No.30.5 had a more refined, subtle, and suave presentation. This was largely the result of the No.30.5's more-set-back spatial perspective, lower midrange grain, better resolution of low-level detail, and slightly tighter image specificity. If price isn't an issue, the No.30.5 is the processor to own. But for $5295, the SFD-2 Mk.II comes awfully close. I could happily live with either unit.

Playing HDCD-encoded discs on the No.30.5 was absolutely magical. Full HDCD encoding and decoding produced an amazing sense that I was in the room with the musicians. There was a real feeling of spontaneous music-making happening right in front of me that I've never heard from conventional digital audio. It was as though I felt the musical experience more than just hearing sound. On an objective level, the difference in the sound may be quite small. But HDCD produced a profound difference in the way I experienced the music. This was particularly true on Mike Garson's fabulous The Oxnard Sessions, Vol.2 CD (Reference RR-53CD). The feeling of being a part of the music-making was stunning, and produced an almost hypnotic involvement in the music.

With conventionally coded digital audio, I've experienced moments of forgetting about the outside world. With full HDCD encoding and decoding, I instantly slip into a deep involvement and stay there. My wife—who also loves The Oxnard Sessions—described exactly the same experience. It's a difficult thing to express in words, but if you spend some time with HDCD-encoded music you like (a prerequisite) under relaxed conditions, you'll know what I mean. And when I heard an HDCD-encoded disc non-decoded, it sounded flat, sterile, and uninvolving.

I attempted to measure a difference in technical performance in the HDCD-fitted No.30.5, but found no measured change. It's the classic case of applying the wrong measurement criteria: Whatever's happening in the PMD100 doesn't show up on standard tests. When all the details of HDCD are eventually revealed, it may be possible to devise new tests and special test signals to quantify some of the differences heard.

Conclusion: Once again, the Mark Levinson No.30.5 Reference digital processor assumes the exalted position of the digital processor against which all others are measured. The No.30.5, now fitted with the Pacific Microsonics PMD100 HDCD decoder/filter, is simply the best-sounding digital processor I've heard. In fact, I'm at a loss to criticize its sound—it'll only be when a superior processor comes along that the No.30.5's shortcomings will be revealed.

There's no question that No.30.5 owners should upgrade to HDCD. Madrigal should be commended for offering this huge increase in sound quality for less than $100—and making the upgrade so easy to install. Moreover, anyone who hasn't converted their No.30 yet will be in for a big surprise when they hear the difference between a No.30 and the HDCD-fitted No.30.5. I think it's wonderful that the No.30 never became obsolete in this rapidly advancing field. It's also a testament to the quality of the No.30's DACs, analog output stage, and power supply that these continuing improvements in the digital section can be resolved and are so musically significant.

At $15,950, the No.30.5 is expensive by any standard. But if you want the unquestioned best digital playback currently available, there's no choice other than the Mark Levinson No.30.5.—Robert Harley

---

Footnote 1: See my Follow-Up on the No.30.5 in October '94 (Vol.17 No.10), and JA's last month (Vol.18 No.3).

Footnote 2: I've found the Z-Stabilizer effective in smoothing the midrange and treble of digital sources. The Z-Stabilizer removes some of the coarseness overlaying instrumental and vocal textures.