CD-R Media Survey

Jerome L. Hartke, Media Sciences, Inc.

Published in Software Fulfillment News, May 18, 1998 and Updated May 22, 2000

Contents of this Document
Purpose: Background of this study.
Sample Selection: Compromises and definitions of test lots.
Test Methods: Description of equipment and calibration procedures.
Test Results: Pass/fail results by test type and by lot.
Table of Test Results: Results for browsers that support tables.
Summary: Conclusions of this CD-R media evaluation.
Return to Home Page

Purpose

CD-R users experiencing problems may believe that all discs are defective, while others may feel that all discs are equally good. Neither expectations are accurate. Quantitative tests by Media Sciences show that the quality of CD-R discs from experienced manufacturers has improved from 30% defective discs in 1998 to 13% in 2000. Failures for other discs increased from 33% in 1998 to an alarming 60% in 2000, mostly for high radial tracking and jitter.

Falling prices, new suppliers, conflicting vendor claims, and silver - green - gold - blue alternatives present a bewildering matrix to CD-R buyers requiring both high quality and low cost. Information to guide procurement is often lacking, since readability in a few drives or the absence of coasters are not effective quality indicators. Costly and embarrassing field failures are often the result.

One laser beam in the read drive sustains multiple operations that are sensitive to CD-R disc quality. Beam intensity is bit detected in the data channel, and data bytes are then error corrected. Beam variations control the clock and spindle servos. One differential beam detector supports the track following servo. Another controls the focus servo. Since drive standards do not exist, readability is not an indicator of disc quality because various read drives can respond differently to media defects. Only conformance to disc standards provides confidence in readability and interchange.

CD-R recording is a sophisticated process involving many components that can result in highly variable quality. Discs, writers, system hardware and software, and recording software can degrade quality. Extensive error correction methods in read drives can mask severe flaws, therefore readability alone does not predict interchange and longevity. Quality can be established and maintained only through in-depth testing using properly calibrated equipment and trained personnel.

This study provides an example of CD-R quality evaluation that properly forecasts interchange and longevity. Detailed results are reported for each of many tests instead of an ambiguous pass/fail conclusion for the disc. Results of this study are not reported by brand name nor are consumers provided with a list of approved or recommended suppliers, since brand quality may be inconsistent.

Return to Top of Page

Sample Selection

Compromises were necessary in order to achieve timely and cost-effective results. The exact origin of each sample was not tracked, although suppliers may have various manufacturing plants or may resell discs from another source. Regular variations with time, or from lot-to-lot, were not explored. Some vendors were not included because they only resell privately labelled discs made by others. The following samples were selected to represent a significant and diverse part of industry capacity.

Lot A, tested in 1998, consisted of 23 samples of 74 minute discs from seven different experienced manufacturers. Cyanine dye (green with gold metal or blue with silver metal) samples from Denon, Verbatim, Taiyo Yuden, and TDK were evaluated. Phthalocyanine (gold) samples from Kodak, Mitsui Toatsu, and Ricoh were tested. Ten discs passed, six were marginal, and seven discs failed.

Lot B, also tested in 1998, contained twelve samples of 74 minute green discs from six manufacturers, generally from the Asia-Pacific region, including Anton, CMC, KAO, Lead Data, Mega Media, and Ritek. One disc passed, seven were marginal, and four discs failed.

Lot C used six samples of 63 minute discs manufactured by Mitsui Toatsu and TDK. These potentially had of higher quality because of lower data density associated with higher linear track velocity. All six discs passed.

Lot D tests during 1999 and 2000 used 87 samples from 17 different manufacturers, adding 80 min. discs, silver discs, and 8X recording. Fifteen samples were from manufacturers represented in Lot A; seven passed, six were marginal, and two failed, one for jitter and one for length deviation. Seventy-two were from manufacturers typical of Lot B; eleven passed, 18 were marginal, and 43 failed, mostly for high radial tracking and jitter.

Return to Top of Page

Test Methods

Samples were recorded at 2X, 4X, and 8X in a high quality recording system of Media Sciences using selected components. The recording system was periodically retested using known media and provided consistent results. Each recorded sample was evaluated for parameters and errors in accordance with ISO/IEC 10149, with certain limits modified for CD-R. Philips calibration and correlation discs were used to certify the test system.

A full suite of electrical tests was used to evaluate CD-R quality. Nominal results indicate high quality, while marginal or non-conforming results disclose problems that predict interchange and longevity failures. Reliance only upon pass/fail tests, or upon qualitative tests such as readability, often results in misleading results and erroneous conclusions.

Parametric tests were conducted for Itop (reflectance), radial tracking (push-pull), radial noise, I11/Itop, I3/Itop, asymmetry, jitter, length deviation, and radial contrast before. Local defects were evaluated using BLER, E22, E32, and burst error measurements.

Environmental tests were not conducted, although heat, humidity, sunlight, and chemicals may degrade CD-R quality. Experience indicates that longevity is best attained by assuring high initial quality. Discs can then be stored under the proper conditions with confidence that they will always be readable in any system of reasonable quality.

Return to Top of Page
Test Results - Itop
Intensity of the reflected laser beam as measured by Itop is important to readability. Reflectivity is affected by both the type and thickness of metallization and by attenuation in the dye layer.
Lot A Results: No discs failed, two were marginal, and 21 discs passed.
Lot B Results: No discs failed, two were marginal, and ten discs passed.
Lot C Results: All six discs passed.
Lot D Results: Seven discs failed, four were marginal, and 76 discs passed.
Test Results - Radial Tracking
This measure of sensitivity of the radial error signal predicts whether or not the radial servo of a read drive will accurately position the laser. Marginally high results are not uncommon, but very high values can result from pre-groove or dye problems, and such discs may be readable in some drives but not in others.
Lot A Results: No discs failed, seven were marginal, and sixteen discs passed.
Lot B Results: No discs failed, four were marginal, and eight discs passed.
Lot C Results: All six discs passed.
Lot D Results: Fifteen discs failed, 24 were marginal, and 48 discs passed.
Test Results - Radial Noise
Excessive noise caused by defects can disrupt the radial servo of a write or read drive, resulting in unpredictable errors.
Lot A Results: No discs failed, two were marginal, and 21 discs passed.
Lot B Results: All twelve discs passed.
Lot C Results: All six discs passed.
Lot D Results: All 87 discs passed.
Test Results - I11/Itop and I3/Itop
Measurement of the amplitude of the return laser beam at both the lowest and highest data rates provides an important quality indicator of the dye layer. Weak signals predict readability problems, especially if subsequent dust or scratches degrade the entrance surface of the disc.
Lot A Results: No discs failed, two were marginal, and 21 discs passed.
Lot B Results: No discs failed, two were marginal, and ten discs passed.
Lot C Results: All six discs passed.
Lot D Results: Two discs failed, three were marginal, and 82 discs passed.
Test Results - Asymmetry
CD-R recorders perform optimum power calibration (OPC) prior to writing. Asymmetry evaluates the match between the dye layer and OPC, and also evaluates the radial uniformity of the dye.
Lot A Results: Three discs failed, none were marginal, and twenty discs passed.
Lot B Results: One disc failed, two were marginal, and nine discs passed.
Lot C Results: All six discs passed.
Lot D Results: Two discs failed, seven were marginal, and 78 discs passed.
Test Results - Jitter and Length Deviation
Error-free data recovery requires that tolerances be maintained in the time intervals between mark-land transitions. Length deviation evaluates averages of eighteen different time intervals, while jitter measures random variations from those averages. Deficiencies usually result from either pre-groove or dye problems, and are a frequent cause of media failure.
Lot A Results: Five discs failed, four were marginal, and fourteen discs passed.
Lot B Results: Four discs failed, three were marginal, and five discs passed.
Lot C Results: All six discs passed.
Lot D Results: Twenty discs failed, seventeen were marginal, and 50 discs passed.
Test Results - Radial Contrast Before
This important quality indicator for the unrecorded disc measures sensitivity to off-track position of the return laser beam. Radial position errors may be excessive during recording if it is low. Problems usually point to an incorrect pre-groove. No maximum limit exists in the standards. Samples did exhibit 3:1 variations that could cause problems in some recorders.
Lot A Results: All 23 discs passed.
Lot B Results: No discs failed, two were marginal, and ten discs passed.
Lot C Results: All six discs passed.
Lot D Results: All 87 discs passed.
Test Results - BLER
BLER denotes frame error rate averaged over ten seconds. It accurately evaluates small errors caused by noise or microscopic manufacturing defects. The long averaging time masks large defects such as scratches, debris, or black spots.
Lot A Results: No discs failed, two were marginal, and 21 discs passed.
Lot B Results: All twelve discs passed.
Lot C Results: All six discs passed.
Lot D Results: Two discs failed, none were marginal, and 85 discs passed.
Test Results - E22
Large defects generate errors that approach uncorrectable levels. Standards specify maximum defect sizes that indirectly forbid E22 errors, and directly prohibit them for archival storage. Since defects cause not only data errors but also adversely affect radial, focus, clock, and spindle servos, their effects are not predictable. Problems observable only as E22 errors in certain drives may generate more serious errors in other drives.
Lot A Results: Two discs failed, one was marginal, and twenty discs passed.
Lot B Results: No discs failed, three were marginal, and nine discs passed.
Lot C Results: All six discs passed.
Lot D Results: Eight discs failed, none were marginal, and 79 discs passed.
Test Results - E32
Very large defects in the substrate, pre-groove, dye, or metallization generate E32 errors that are uncorrectable by C1/C2 circuitry in the read drive. All Standards require zero E32 errors that can cause read failure, low data rates resulting from read retries, or make the disc unsuitable for archival storage purposes.
Lot A Results: One disc failed, none were marginal, and 22 discs passed.
Lot B Results: All twelve discs passed.
Lot C Results: All six discs passed.
Lot D Results: Five discs failed, none were marginal, and 82 discs passed.
Test Results - Burst
Defects or scratches along the track result in contiguous errors. Standards forbid burst errors that contain seven or more sequential error frames.
Lot A Results: Three discs failed, none were marginal, and twenty discs passed.
Lot B Results: All twelve discs passed.
Lot C Results: All six discs passed.
Lot D Results: Five discs failed, none were marginal, and 82 discs passed.
Table of Test Results for Newer Browsers

Return to Top of Page

Summary

Test results clearly indicated that all discs were not alike, even if their colors were similar. Cost pressures have resulted in a broad matrix of stampers, dyes, metallizations, and processes. No correlation was observed between CD-R quality and dye type (cyanine or phthalocyanine), metallization (gold or silver), or recording speed (2X-8X). Quality is primarily determined by efforts at the manufacturing facility, and depends less on types of dyes or metallizations.

Good discs would be expected to satisfy all interchange and longevity requirements. Marginal or defective discs might be readable in high quality drives, but could fail in others. Degradation from handling or storage might cause poor quality discs to become unreadable while better discs could still function.

Updated Lot D results from 1999-2000 indicate that the quality of discs from experienced manufacturers, similar to those evaluated in Lot A, has improved from 30% failures to 13%. Quality in Lot D from manufacturers comparable to previous Lot B has declined. Failures in this group increased from 33% to an alarming 60%. High radial tracking and jitter were the most commonly observed deficiencies. BLER was not a meaningful quality indicator.

Readability in a few drives does not confirm quality. Even reliance upon brand name can be ineffective unless each manufacturing location and product type is qualified and regularly monitored to assure consistency. Only in-depth testing can qualify media and assure interchange and longevity. The expense of such an effort is quickly repaid when recording processes flow smoothly and field failures are minimized. Establishment of a quality baseline enables further cost savings to be achieved by reducing the frequency of testing while maintaining a high level of confidence in the process.

This study will be expanded from time-to-time. Current results will be posted on this web site.

Return to Top of Page

Media Sciences, Inc. –– Dedicated to Quality