The Colossus of Audio John Atkinson Comments

John Atkinson Comments, from May 1988 (Vol.11 No.5)

In Vol.11 No.4, JGH enthusiastically reviewed a series of recordings made with the new Colossus recording system (p.153). The reason for his enthusiasm was that these recordings totally lacked any sign of the dreaded "digititis" so prevalent in modern classical recordings, leading him to proclaim that the Colossus "could justifiably be included in any contemporary listing of the seven wonders of the audio world."

My appetite whetted, I listened to one of these recordings, Stephen Kates playing Rachmaninov's Cello Sonata in G Minor, accompanied by Carolyn Pope Kobler on a Bösendorfer piano (Bainbridge BCD6272), one of my favorite pieces of chamber music. JGH had declared that this was the most natural-sounding CD recording of a cello that he'd heard. I have to agree that the tonality of both instruments is very natural, although the acoustic surrounding the instruments is very close and dry. In my opinion, a less closed-in, more reverberant recording venue, one sounding less like a small, modern, heavily draped hall, would have been more sympathetic to the music. That, however, is a matter of taste, and not germane to this followup report.

Something I found much more disturbing was the fact that the music was accompanied by often quite severe distortion at climaxes, sounding just as if an amplifier had been clipped. Instances from the first movement are at 0m 44s, 0m 55s (left channel only), 7m 20s and 7m 24s (right channel only), 7m 50s, 8m 34s, 8m 41s and 12m 12s. To ensure that this was not due to the playback system being driven into overload, the disc was auditioned on three systems, at a variety of levels, the final listening being done via a pair of Stax Lambda Pro headphones. Checking the recorded level against a 0dBFS reference indicated that the distortion appeared to be due, not to the recorded waveform "banging its head" against the maximum possible digital encoding level on the CD, but against a level 1.27dB below the maximum (fig.1). A maximum-level sinewave was reproduced by the same CD players without clipping.

Fig.1 Rachmaninoff Cello Sonata, example of digital clipping from the first movement. Note the waveform's flat tops and bottoms at a voltage equivalent to –1.27dBFS.

The 1981 copyright date on the CD jacket and the ADD SPARS code imply that the Colossus digital processor had just been used as a transfer medium for the mastering of an original analog recording (footnote 1). In view, therefore, of what I can only term an obvious error on the part either of the engineer who made the digital copy of the original analog tapes or of the engineer who prepared the CD master, Stereophile's recommendation for this recording must be withdrawn, as natural as the recorded sound may be.—John Atkinson

Footnote 1: Bainbridge says that the SPARS code is, in fact, incorrect; the original recording was digital. Engineer Leo Kulka confirms that it was made with a highly modified Sony PCM1 PCM processor. Regarding the "clipping," the disc is being remastered, Mobile Fidelity Productions of Nevada's Manufacturer's Comment elsewhere in this issue giving the facts of the matter.—John Atkinson