The Analog Compact Disc Page 4
Jitter has recently become a hot topic in the CD-manufacturing business. When the CD specification was created (codified in the Philips "Red Book" standard), jitter was never even mentioned, much less specified. Now manufacturers of CD-mastering machines are touting their models as having "low jitter," and disc replicators are beginning to measure jitter in the HF signal as part of the quality-assurance process. Philips even plans to revise the Red Book (the first revision ever) to include a maximum jitter specification. The new jitter spec is rumored to be 50 nanoseconds—about half the amount of jitter that would create a bit error. When Philips decides to make as dramatic a move as adding a new specification to the Red Book after more than 15 years, there must be a good reason behind it.
CD manufacturers have recently adopted double-, triple-, even quadruple-speed mastering, which cuts down on expensive mastering machine time, and increases a CD factory's throughput. I don't have any data, but it's hard to imagine that high-speed mastering improves the disc's jitter performance.
In the electroforming steps of CD manufacturing (converting the glass master to stampers), the primary threats to CD quality are contamination and scratches. Given the pits' tiny sizes, a speck of dust would obliterate hundreds of tracks. Any scratches on the metal parts will also wipe out data, and increase the disc's error rate.
The injection molding process also influences CD quality in that it can create problems in the polycarbonate which introduce an optical phenomenon called "birefringence." Birefringence is a double refraction of the playback beam caused by lack of homogeneity of the polycarbonate. Part of the beam travels at a different velocity and polarization, due to changes in the polycarbonate's refractive index. This varying refractive index is caused by stress on the polycarbonate during the molding process. The polycarbonate temperature and injection pressure must be held to tight tolerances to prevent excessive birefringence. Birefringence, which you can see as striations or swirls in the polycarbonate, degrades the HF signal's quality (footnote 6).
The injection molding process can create disc eccentricity—ie, the disc's spindle hole wasn't punched out at the exact center of the disc (disc eccentricity introduced when the stamper's center hole is punched). This causes the CD transport's tracking servo to work extra-hard to stay on track, drawing lots of current from its power supply and causing power-supply droop. If that supply rail also feeds other circuits, an eccentric disc can affect other subsystems within the transport. This is one reason why high-end CD transports use separate power-supply regulation stages for each transport servo system.
Metallization affects the HF signal in an analog-like way. First, the metallization layer on some CDs isn't thick enough. If you can see through the CD, then insufficient aluminum has been deposited. Aluminum becomes opaque at a thickness of about 800 Angstrom units. The result of a too-thin aluminum layer is lower reflectivity, which lowers the HF signal amplitude.
A separate problem introduced by the metallization process is "pinholes"—tiny holes in the reflective layer. Any spurious particles on the unmetallized disc surface become metallized, then fall off, introducing areas of no aluminum. Pinholes create dropouts in the HF signal, and increase the disc's error rates. You can check for pinholes by holding a disc up to a strong light.
Some CDs—particularly very old ones or of PolyGram Hannover manufacture—are metallized all the way to the outer edge. This can allow air to reach the metallization layer, causing the aluminum to oxidize. The oxidized area slowly eats its way toward the disc center, degrading the HF signal and even rendering the disc unplayable. It can take many years, however, for oxidation to ruin a disc.
Sam Tellig sent me a batch of CDs he'd bought that developed horrible brown blotches around the outside edge. The discs were all made at the same factory, the Philips-DuPont Optical (PDO) plant in Blackburn, UK. These discs were unlike any I'd ever seen—most oxidized discs turn a shade of gray, not brown. I'll report on the effects of these brown blotches later (footnote 7).
In short, CD manufacturing isn't just a simple matter of stamping inviolate digital ones and zeros into a disc, but a series of interdependent processes that produce an analog-like variability in the disc's signal quality.
Data errors on CDs
The raw data recovered from a CD are never identical to the data recorded on the disc. The CD system introduces data errors and dropouts through contamination, scratches, pinholes, and the other manufacturing defects described.
Footnote 6: Birefringence is measured as the relative phase shift between the beam components. A disc must have less than 100nm (nanometers) of birefringence (double-pass) to meet Philips's Red Book specification.
Footnote 7: If you have any of these discs, which were primarily on the ASV label, you may return them to the distributor, Koch International, for replacement. Send defective discs to: Koch International, Attn: Kathy Zagorski, 177 Cantiague Rock Rd., Westbury, NY 11590. Tel: (516) 938-8080. Fax: (516) 938-8055. You should, however, first try to get replacement discs from your local dealer. If you've got bad CDs, you can thank Sam for his efforts in finding an entity willing to accept responsibility and replace the discs. Note that Koch will accept only those discs which it distributes, not all brown discs made by PDO in Blackburn.