dCS Elgar D/A processor:
Page 3
Better sound
But it was when I abandoned 16-bit CDs and DATs that the Elgar scaled the highest sonic heights. Not only did my 20- and 24-bit master tapes sound delightful in their analoglike sense of ease; that delicious sense of detail when the processor was used straight into the power amplifier was invaluable when I was editing the 24-bit hard-disk masters for Stereophile's new Rhapsody CD (see June '97, pp.70-81). Choosing the right note on which to splice between two takes is an exacting business. Not only do you need the musical flow to match each side of the edit point, the sounds must also match. Or if they don't match exactly, the difference must be musically justifiable. A slight change of timbre on a major attack will still sound natural, for example. The more sonically transparent the playback chain, the less stressful the editing. With the Elgar, I benefited from very low stress levels.

But it was the experience of auditioning 96kHz-sampled tapes, made with the dCS 902D A/D converter, that finally sold me on the Elgar. The latest software for the Nagra-D 4-channel digital recorder allows it to be used as a 2-channel 96kHz or 88.2kHz machine. For playback of a 96kHz-sampled two-channel tape, two AES/EBU cables (footnote 1) connect the Nagra's digital outputs to two of the digital inputs of the Elgar processor; when both of its AES/EBU inputs are simultaneously selected, the Elgar remultiplexes the two 48kHz two-channel datastreams into a single 96kHz two-channel stream.

You wouldn't expect that increasing the recorded bandwidth by just over an octave would have any effect. But yes, although the improvement in quality was, on the face of it, subtle, going back to 44.1k was more of a degradation than you'd expect. At 96k there seemed to be more "air" around individual objects within the stereo image, the sense of the acoustic space captured on tape was more tangible, cymbals seemed more lifelike in that they more resemble a vibrating metal sheet, a clarinet sounded more like a column of vibrating air excited by a reed, and voices sounded simply more palpable. And surprisingly, one of the biggest perceived differences when comparing 96kHz and 44.1kHz versions of the same material was in the bass. With the higher sample rate, there seemed to be another half-octave of low-frequency extension. Weird stuff!

Summing up, the sound of 96kHz tapes reconstructed by the dCS Elgar was simply more real. And that is what the High End is all about.

Summing up
I'll get the bad stuff out of the way as quickly as possible: the dCS Elgar doesn't have HDCD decoding, doesn't have a digital output, costs $12,000, and...that's it. On the plus side, the Elgar sounds simply superb and has a measured performance to match. It joins the hallowed ranks of Class A digital processors in Stereophile's "Recommended Components." It is also future- and DVD-proof in that it will decode two-channel 24-bit digital recordings recorded at sample rates of 88.2kHz and 96kHz (footnote 2). A nice if pricey one from the digital guys in Cambridge (Cambridge England, that is).

---

Footnote 1: There is conflict in the pro audio community as to how to handle high-bandwidth digital data. Nagra, dCS, Genex, and SADie have chosen to use two standard AES/EBU connections, each carrying one channel's worth of data; Sonic Solutions and some other companies run a single AES/EBU connection at twice the speed. I'm sure imaginative companies like Bob Katz's Digital Domain will introduce the necessary adaptor boxes. (The dCS 972 DDC also does this.)

Footnote 2: If you already own a Mark Levinson No.30.5, you should note that Madrigal is shortly to introduce a revised input board for both that champion and the No.36/36S, allowing them to accept two-channel 96kHz data sources, probably with a double-speed AES/EBU input, as well as two-channel versions of Dolby Digital (AC-3), DTS, and MPEG/Musicam audio data. (You can find full details at http://www.madrigal.com .)