CD Player/Transport Reviews

Sidebar 4: JA Takes a Listen

It's hard to sum up the measured performance of the expensive four-box player. In many respects—its poor rejection of word-clock jitter, its nonflat response, its nonlinearity at low frequencies, its lack of ultrasonic image rejection, its higher-than-usual noise floor—it is the worst-measuring digital product I have encountered. Yet Michael Fremer was totally bowled over by its sound. Assuming Mikey is not deaf—and I have taken part in enough listening sessions with him to know that that is not the case—what the heck is going on with the Zanden pair? Is this a case of "horses for courses"? (See John Marks' "As We See It" in this issue.)

I hooked the 5000 Signature up in my regular reference system—Mark Levinson No.326S preamp and No.33H monoblocks, Revel Ultima Studio speakers—and did some level-matched comparisons with the Mark Levinson No.30.6 D/A processor, which competes price-wise with the Zanden 5000S (or would, if it were still available) (footnote 1).

I started off driving both with the AES/EBU output of the Classé cdp-202 player. Strange. Even though I knew how badly the 5000S measured when I did the comparisons, there was nothing immediately or obviously bad about its sound other than a slight hum that was audible at high playback levels. However, on pure, digitally generated low-frequency tones around –3dBFS, the Zanden could be heard to fur up the sound with a spray of low-order harmonics, particularly if I moved around the room to find a location that nulled the fundamental. Similarly, if I played the mix of 19kHz and 20kHz tones used to generate fig.10, I could hear both the 1kHz difference tone and a buzz of other spurious products, even though I had the preamp's volume low to avoid smoking the Revel Studios' tweeters. The Levinson produced no audible distortion spuriae in either case.

But what about the Zanden's reproduction of music? The Fender Bass channel-identification tracks on Editor's Choice (Stereophile STPH016-2) sounded lightweight in comparison with the Levinson, but also had a bit more bite. The Mozart Flute Quartet track on this CD sounded rather old-fashioned, in that it seemed a little band-limited. It also seemed a little quieter than the Levinson, despite the levels being matched at 1kHz. Perhaps more importantly, the soundstage seemed diffuse and deep rather than focused and forward, as it did through the Levinson. This effect wasn't unpleasant, but was not what I thought I was capturing when I made the recording.

There was a similar presentation on this CD's solo violin track. There was noticeably more space around the instrument through the 5000S, but the image of the violin was a little diffuse. The presentation of Robert Silverman's piano on his Beethoven Sonatas set (OrpheumMasters KSP830) was very similar, with more reverberation and more image depth apparent through the Zanden than through the Levinson. The Bösendorfer did sound a little "smaller," however. In fact, while the 5000 Signature's lightweight low frequencies didn't sound unpleasant, the bass guitar on Sting's Mercury Falling (CD, A&M 31454 0483-2), which I was listening to while writing the article on the making of the new Cantus CD in the December issue, had significantly more weight through the Levinson and less of a furry quality. The deep organ pedals in the finale of the BBC recording of Elgar's Enigma Variations with Tadaaki Otaka conducting the BBC National Orchestra of Wales (CD, BBC Music 250) were missing in action compared with the Levinson, and the big bass-drum smack at the very end sounded a bit furry.

But I must say that the Zanden's presentation was not unpleasant, and was even rather seductive. When music had no low-bass content, I didn't notice the lack, and the midrange did seem very clean. Only in the highs did I consistently become bothered by a pervasive grain. Not only did Richard Lehnert's voice on the spoken introductions to the test tracks on Editor's Choice acquire a tad extra sibilance, but Art Baron's plunger-muted trombone on "The Mooche" had a bit more "bite" through the Zanden, and the massed strings on the Enigma Variations sounded too wiry.

But this was all with the Zanden 5000S used as a standalone DAC, and I might have been hearing the effects of the poor rejection of word-clock jitter. When I drove the 5000S with the 2000P transport using the I²S connection, as MF found, the treble grain was much reduced in level, leaving me with the clean midrange, the spacious soundstage, and the lightweight, but rather furry, low frequencies.

So how come Mikey loved the Zanden pairing's sound, and even I was not too bothered much of the time with what I knew was going wrong?

All I can surmise is that the apparent discrepancy comes from the fact that the 5000 Signature actually performs quite well in the critical midrange. When you play music with strong midrange content, that quality is what you will hear. Yes, its measured shortfalls in performance below 100Hz and above 10kHz are relatively severe, but music with strong content in those regions is relatively rare. Yes, synthesizer-based rock and organ music will expose the Zanden's bass limitations, but even in rock music, low-frequency tones peaking much above –20dBFS are not common. At the other end of the spectrum, perhaps it doesn't matter that loud cymbal crashes will be accompanied by bursts of what also sound something like cymbal crashes.

Personally, I wouldn't recommend the Zanden pairing. Both psychologically and practically, I need to use components that neither editorialize nor are on the verge of misbehaving. And the Zandens' positive impression could be punctured by playing the system too loud. While the Mark Levinson DAC's presentation became more magnificent when played at high levels, the Zanden just sounded more fuzzy. But again, I was surprised by how approachable the Zandens' presentation was at moderate playback levels, by how seductive this transport and DAC could sound. Score another one for the subjectivists, I suppose.—John Atkinson

---

Footnote 1: All my auditioning was performed with the correct absolute polarity (red LED illuminated). This turned out to be significant, as you will read in this issue's "Manufacturers' Comments" (p.144).—John Atkinson