What is an Extended Resolution Compact Disc (XRCD), and how is it different from a conventional CD?
First, XRCD is not actually a “format” that requires special players or decoding. Rather, an XRCD conforms to the Red Book specification that defines the compact disc, making XRCD compatible with all CD players. XRCD is simply a mastering and manufacturing process that attempts to extract the highest possible sound quality from the CD format.
JVC has examined each step in the CD mastering and manufacturing processes and designed specific equipment to improve those processes. Every combination of equipment, connections, AC power regulation, clocking, mastering format, and compact disc construction was evaluated technically and by listening tests. XRCD was created by a team that included Alan Yoshida, Akira Taguchi, Shizuo Nomiyama, Dave Collins, and Andrew Garver.
The XRCD process starts by converting the analog signal (most XRCDs are re-issues from original analog tapes) to digital with JVC’s custom 24-bit analog-to-digital converter. A considerable amount of human skill is involved in this process to extract the best sound from the analog mastertape. This includes researching the tape’s provenance to be sure it’s not a multi-generation “master.”
The A/D converter employs JVC’s proprietary K2 Interface (see sidebar) that clocks the signal with high precision. The 24-bit signal is then reclocked in the digital domain with JVC’s Digital K2 before being stored on a magneto-optical disc. This magneto-optical disc is the master from which the CD master glass is cut. Before the CD master is cut, one-off test discs are created from the magneto-optical master and must be approved.
The XRCD laser cutting system is unlike any other CD mastering system in the world. The 24-bit data from the magneto-optical disc are converted to 16 bits with a process called “K2 Super Coding,” a noise-shaping technique that maintains much of the 24-bit signal’s dynamic range in the 16-bit signal. This 16-bit audio signal is then encoded into the bitstream that will be recorded on the glass master. The signal is again reclocked with a circuit called “K2 Laser” before driving the crystal that modulates the laser beam.
All these efforts to create a signal with such precise timing would go to waste if the speed of the spinning glass master were not equally precise. Consequently, JVC developed a turntable speed control it calls Extended Pit Cutting Technology. Without precise speed control, the pit and land lengths would vary. (See sidebar for a more detailed description of CD mastering and the sonic effects of turntable speed accuracy.)
All the components in the mastering chain are clocked from a single K2 Rubidium Clock. There are no crystal oscillators anywhere in the system. Moreover, all the AC power supplying the mastering system components is generated locally. That is, a precision oscillator creates a very clean low-level 60Hz sinewave that is amplified to 110V to power the components in the mastering chain. This assures perfectly clean power that’s completely isolated from the power grid.
Finally, the stamper that presses an XRCD is created by a one-step rather than a three-step process. In conventional three-step stamper creation, the glass master is coated with a thin layer of silver and then electro-plated with nickel (step #1). Next the layer of nickel is peeled off to create a metal mother (step #2). Then the metal mother is electro-plated with nickel to make a stamper (step #3). Multiple stampers are created from a single mother. In the XRCD process, the glass master is itself electroplated to create the stamper. Eliminating two electroplating steps produces cleaner and more precise pit-and-land structures on the replicated disc. But because only one stamper—a stamper that wears out—can be created from a glass master, the quantity of discs made by the XRCD process is limited. Before the production run of discs, however, a few discs are pressed so that the sound quality can be evaluated. Even after the production run, discs are sampled at random and auditioned.
CD Mastering: Converting Data to Modulated Light to Physical Structures
The digital signal that is to be recorded on the CD master doesn’t actually turn the cutting laser on and off. Rather, it drives an electro-optical modulator, a crystal whose lattice structure “twists” when a voltage is applied across it, diverting the laser beam’s path. This phenomenon allows the crystal to convert an electrical signal into a modulated beam of light.
The digital bitstream to be recorded on the disc drives a power amplifier that can swing very high voltages very quickly. The power amplifier’s output is applied across the crystal. With no voltage applied across the crystal (corresponding to zeros in the digital data stream) the crystal allows the beam to pass straight through to the glass master, creating a “pit.” But when a voltage is applied across the crystal by the power amplifier (corresponding to “ones” in the digital datastream), the crystal’s lattice structure twists and diverts the beam’s direction away from the glass master, leaving the master unexposed and creating the “land” between the pits. (This explanation is simplified for clarity.)
The cutting laser is always on; its beam is modulated by the bitstream to be recorded on the disc by the crystal. This is how an electrical signal is converted to a modulated laser beam, which in turn creates the pit and land structures on the glass master. Incidentally, the “rejected” beam that doesn’t reach the glass master contains the identical data as the recorded beam. This beam can be picked up by a photodetector, decoded, and listened to.
Variations in turntable speed introduce variations in the pit and land structures on the CD. One would assume that if the signal can be recovered from the CD with no data errors, subsequent reclocking could completely remove any traces of timing errors introduced by these pit-and-land-length differences. But one would be wrong about that. I’ve heard two CDs containing the same data (verified to be bit-for-bit identical) but cut on two different mastering machines, one of which used a vastly more sophisticated turntable rotational-servo system. The disc cut on the machine with the more precise turntable speed sounded more open, spacious, and smoother; it also had a greater sense of ease. An analysis of the variations in pit and land lengths (in essence, jitter embedded in the disc itself) revealed that the better sounding disc had less variation in its pit and land lengths.
SIDEBAR: JVC’s History in Digital Audio
JVC has a long track record of making digital audio sound better; it’s no surprise that JVC is behind this significant technical effort to improve CD quality.
In the early days of the compact disc, JVC developed a mastering system that was sonically superior to the Sony PCM-1600 format. JVC’s format, called the DAS-900, stored CD masters on 3/4" U-Matic tape as did the Sony system, but JVC’s analog-to-digital converters, digital signal processing, and digital-to-analog converters were markedly better sounding. There was a brief format war which Sony won, but a few mastering engineers (notably Doug Sax of The Mastering Lab) continued to use JVC’s superior system for many years after the industry had regretfully abandoned it.
The first time I became aware of jitter in digital audio was on a 1989 tour of JVC’s R&D laboratory in Japan. Two of its engineers created a circuit called the “K2 Interface” that produced a cleaner and more precise clock for digital-to-analog conversion. I had been bewildered by the incontrovertible fact that two digital bitstreams with the same ones and zeros exhibited an analog-like sonic variability. This phenomenon was particularly vexing because I’d spent the previous three-and-a-half years working in CD mastering and the topic was starting to become a wedge issue with my engineering colleagues, some of whom dismissed the phenomenon purely on theoretical grounds.
During the very technical presentation on the K2 Interface, I had one of those “ah ha!” moments when I realized that the timing precision of the analog-to-digital and digital-to-analog conversion process was crucial. JVC had known about the detrimental sonic effects of jitter for years, and developed the K2 Interface to solve the problem. To the JVC engineers, jitter was simply another engineering challenge to solve. Moreover, JVC employed professional listeners who worked with, and guided, the engineers. I met one of them, who told me that he compared all of the laboratory’s new digital designs to LP playback. It’s rare to see a major company’s audio research laboratory where the attitude isn’t “If it can’t be measured, it can’t be heard.”
You can read Wayne Garcia's accompanying article here.