Why CD-Rs Fail

Contents of this Document

CD-R Defect Mechanisms: Sources of CD-R field failures.

Sources of Defects: Flaws responsible for field failures.

Defect Types: Tests that detect CD-R flaws.

Conclusions: Summary of findings.

Defect Prevention: How to avoid CD-R field failures.

CD-R Defect Mechanisms

Although discs that are unreadable may have one dominant defect, such as no data, failed CD-R discs normally show multiple flaws when tested to ISO/IEC 10149, Orange Book, and ISO 9660 interchange standards. Read failures are usually caused by tolerance stackup, where combinations of defects are fatal even though no single flaw prevents readability. Read failure is then unpredictable, since some read drives may be forgiving while others are more sensitive to defects.

Unrecorded discs, writers, formats, and handling, are four defect sources responsible for interchange failure. Disc manufacturing problems adversely affect writer servos, or prevent recording of well-defined data marks. Flaws in the recording drive, its mounting and cooling, or power supply produce poor data marks or other imperfections such as improper track location. Recording software design and associated operator selections cause format defects such as invalid gaps or incorrect track, session, volume, or file structures. Improper handling, either before or after recording, results in debris or damage that interferes with write and read operations.

Media Sciences analyzed test results of one hundred CD-R field failures during the twenty-four month period ending August 2002. The most frequent cause of field failure was a combination of improper formats and poor handling, with no evidence of disc or writer problems. Results shown in Table I, where a mark indicates that defects were observed for the associated source, demonstrate that 78% of the failures had two or more identifiable defect sources.

Sources of Defects

Ideally, manufacturers would first verify that products satisfy interchange standards, and then provide confirmation from an independent testing laboratory to their customers. In practice, cost and time-to-market competitive pressures, together with user demands for higher storage capacities and faster recording speeds, result in products containing flaws that are masked from the user. Brand names are not helpful because quality may vary significantly within products from one supplier.

Both pre-groove and dye layer quality of unrecorded discs are important. Manufacturers monitor these characteristics, but test equipment is too expensive for most professional and home office users. As a result, design flaws and production problems from short cycle times, stamper overuse, and dye recycling are not detected until user problems surface. Discs having capacities in excess of 74 minutes often contain track pitch, scanning velocity, or minimum/maximum radius values that violate standards and enhance the risk of interchange failure.

Recording drive servos, pickup heads, firmware, mechanical mounting, cooling, and power supplies affect recording quality. Manufacturers and users seem preoccupied with the rush to higher speeds that place greater demands on laser power, servo stability, clamping accuracy, and vibration characteristics. As a result, writers not only may perform poorly at their maximum speed but even worse at certain lower speeds. Not all drives of a given model and lot may be suitable for recording, and a very limited number are usable for archival or laboratory purposes.

Recording software must not only provide a convenient user interface but must also produce images that conform with ISO/IEC 10149, with the Orange Book, and with ISO 9660 and its extensions. Software engineers increasingly incorporate flawed formats together with improper or non-optimum default selections.

Handling defects are often overlooked because of robust CD design. Obstructions on the entrance surface are usually masked because of the large and defocused laser beam at that plane. Read errors that do occur are usually concealed by powerful error correction methods that are part of every CD. False confidence in robustness leads to severe defects that may not cause read failure themselves but combine with other flaws to prevent successful interchange. Imperfections on the entrance surface during recording cause a permanent defect to be written in the dye layer. Only a thin lacquer layer on the fragile label side protects recorded information that can be damaged by poor handling practices.

Defect Types

Improper handling and incorrect formats were observed more frequently than disc or writer defects. Sixty-four percent of the failed discs exhibited handling defects consisting of scratches, smears, scuffs, debris, and fingerprints. Format defects were observed on 57% of the discs. Incorrect post-gap was the predominant format defect at 35%, closely followed by improper descriptors on 24% of the discs, and wrong session structures on 6% of the field failures.

Disc defects were present on 51% of the field failures. Uncorrectable errors (E22, E32, burst) were most frequent as shown in Table II.

Writer defects were present on 35% of the field failures. Jitter and uncorrectable errors (E22, E32, burst) were most frequent as shown in Table III.

Conclusions

Although users typically expect to discover and correct a single flaw when testing failed CD-R discs, multiple flaws are usually detected when such discs are tested to ISO/IEC 10149, Orange Book, and ISO 9660 standards. Identification and correction of just one defect would then overlook other problems associated with unrecorded discs, writers, formats, or handling, and interchange problems would remain.

Confidence in CD-R interchange and longevity requires comprehensive logical, visual, mechanical, and electrical testing. BLER is often mistakenly identified as the single most important quality indicator. This survey indicated that BLER tests alone would have detected only 19% of the field failures. Forty percent would have been detected by uncorrectable error tests, while the combination of jitter and uncorrectable error testing would have detected 55% of the field failures. In the absence of format, mechanical, and visual inspection, only 64% of the discs having defects would have been identified using electrical parameter and error tests alone.

Defect Prevention

CD-R discs, drives, and software must be qualified by comprehensive testing if field failures are to be minimized. Low cost testers may only evaluate BLER and uncorrectable errors, overlooking important quality indicators such as jitter, asymmetry, I3, and I11. Proper testing requires expensive equipment and operators trained in its correct use. Only volume or specialized users can afford the cost of in-house or third-party testing, and then should provide quality certification to their customers. In the absence of certification by disc, drive, and software manufacturers, most users of CD-Rs for mass storage can only rely on experience, word-of-mouth, or on Media Sciences' offer to test one recorded CD-R at no charge.