Is There A CD-R Media Problem?
Originally published in CD-ROM Professional, February
Information about other CD-R media tests can be found at:
- OSTA - Optical Storage Technology Association
Table of Contents
- CD-Recordable versus Replicated Compact Discs
- Tenets of Media Testing
- The Message in the Media: Test Results
- CD-R Media? No Problem
The relative effectiveness of CD-Recordable media is an issue often bandied about in industry and business circles, where the technology is used and increasingly relied upon. Much of the controversy surrounds finding some useful way of evaluating the blank discs of various brands and types used in CD recorders today. Frankly, what the attention given this issue often reflects is a free-floating anxiety about whether the CD-R discs being used are any good.
Several criteria go into evaluating disc usefulness: readability, compatibility with recorders and players, and expected lifespan. Results amassed in a series of tests performed by One-Off CD Shops International between early 1993 and mid-1995 on a variety of disc brands and types shed a great deal of light on the topic, even though the tests were done only to evaluate readability of recorded discs, and not media longevity or suitability of specific brands or types for use on every system. But the methodological rigor of the narrow focus afforded yielded considerable data that bodes well for the effectiveness of current disc-evaluating mechanisms.
Not every question has been answered by any means, but one finding is clear: worry about the quality of CD-R media seems largely unfounded.
One source of confusion and concern about CD-Recordable discs is their notable physical differences from "normal" pressed compact discs. Each CD-R blank is designed to meet standards regarding function, but the way each achieves the function of storing digital information in a manner that can be read by standard CD players and drives is quite distinct. From the top side and bottom, replicated discs are similar; it's what comes between the polycarbonate substrate and the top's lacquer coating that makes the difference.
Replicated discs have their data encoded during injection molding, with pits and lands pressed directly into the substrate. The data side of the transparent disc is metalized, usually with aluminum sputtered onto the bumpy surface, which is spincoated with lacquer to protect the metal from corrosion, and then it is usually labeled in some fashion, generally with a silkscreened or offset printed design.
CD-Rs are polycarbonate underneath, too, but the substrate is molded with a spiral guide groove, not data's pits and lands. This side is then coated with an organic dye, and gold or silver (instead of aluminum as on manufactured discs) is layered on top of the dye as the reflective surface, which in turn is lacquered and sometimes labeled just as replicated discs are. The dye forms the data layer when the disc is recorded, having a binary information image encoded by a laser controlled from a microcomputer using a premastering and recording program. Where the recording laser hits the dye, the equivalent of a molded "pit" is formed by the laser beam reacting with the photosensitive dye, causing it to become refractive rather than clear or translucent. When read by a CD player or CD-ROM drive, the affected area diffuses the reading laser's beam, causing it to not reflect back onto the reader's light-sensor. The alternations between the pickup laser's reflected light and refracted light make up the binary signal transmitted to the player's firmware for unencoding, error detection, and correction, and further transmission to the computer's processor or the audio player's digital/analog converter.
But the feature that really distinguishes recordable media from replicated discs is the dye layer. The polymer dye formulas used by manufacturers are proprietary or licensed, and are one of the distinguishing characteristics between brands. Two types of dye formulas are in use at this time, cyanine (and metal-stabilized cyanine) and phthalocyanine. One (cyanine) is green, the other appears gold because the gold metalized reflective layer is seen through the clear dye.
Return to top of the page
At least in theory, however, these differences should have little or no impact on readability, since CD-R and CD-ROM media share a common antecedent, the "Red Book" standard for CD-DA (Digital Audio.) Red Book specifies a number of testable measurements that collectively are supposed to determine whether a disc should be readable. The Red Book describes audio CD; CD-ROM media, designed for multipurpose data storage and delivery, are expected to meet a higher standard of correctness. Consequently, the Yellow Book, or CD-ROM standard, requires some additional tests.
Since CD-Recordable discs, described in the Orange Book, are supposed to be functionally identical to CD-ROM, it is logical to assume that the same test equipment and standards should be applied to them as to Yellow Book discs, so no new readability criteria were specified in the Orange Book. Several companies have built machines that are used for testing discs during and after the manufacturing process using these criteria, and only recently have new testing devices made specifically for CD-Recordable become available.
There are several ways to measure what the player or drive's laser pickup "sees" as it reads the data off of CD-ROM and CD-R. These tests include BLER, BRST, E32, and several others. [Editor's note: See sidebar on "CD Testing: The Digital and Analog Sides."] The testing breaks down into major groups, including digital testing, analog testing, and logical testing.
Tests were performed on seven brands of CD-R media written on Sony CDW-900e, Sony CDU 920S, Ricoh 1060C, Yamaha CDR100, and Philips CDD-521 recorders. (The Philips recorder was used both before and after installing the Audio/ROM upgrade). The CD-R brands tested were as follows:
- Taiyo Yuden
- Mitsui Toatsu (MTC)
All discs tested were recorded using OMI's QuickTOPiX for Windows running on an IBM-compatible 486DX/50 with 16MB of RAM. Two disc image files were used: one of 59:34 minutes for 63-minute blanks and one of 73:42 minutes for 74-minute media.
All tests were conducted on the CD Associates Quick Test CD Analyzer between March 1994 and May 1995. The CD Associates Analyzer uses three-beam CD players, which ensures performance similar to that of all CD-ROM drives and most CD-Audio players, although the SL-25 CD-R tester had not been released into full production yet when these tests were undertaken.
Return to top of the page
"We were made aware that there were discs that went through the analysis without any issues that would not play in a CD-ROM player," says CD Associates' Mark Worthington. "I don't want people to get the idea that this is a major issue -- only about 3% of unreadable discs have this kind of undetected error. There are two methods of playing a disc. The read mode analysis utilizes the extra information in Mode 1 CD-ROM, like a third level of decoding, as you translate across a disc to detect errors. In play mode we don't use this information. This is a very technical issue, and it really gets down to whether the problem is source material, manufacturing quality, or the player. The SL-25 is designed to detect these previously undetected errors by using read mode instead of play mode. Once the errors are detected and a sufficient body of data is developed about these errors, the problem can be addressed and solved."
Of the four types of errors -- digital errors (BLER, BRST, uncorrectables), analog errors (quality of analog signal that enters the decoder), physical errors (track pitch, speed, and the like), and format or logical error -- only the last type is new. The format or logical error has to do with the encoding and how the mastering machine communicates with the source. It is very rare and it only occurs on CD-ROM, not audio discs.
It is important to note that because of the limited number of discs tested and the fact that several CD recorders were used in the testing, these results may not be considered statistically valid, but they may prove useful nonetheless as a benchmark for further research. The results also provide interim information in a field that has been remarkable for its lack of published qualitative and comparative analysis.
Return to top of the page
CD recorder performance is directly affected by the type of media used and its quality. All recorders used in this testing produced discs that met Orange Book specifications, but individual advantages and disadvantages of each device were noted.
The Yamaha writer was the first 4X recorder marketed. It was very comfortable writing at 4X and 2X speeds, and the discs written at higher speeds in particular performed quite consistently. When discs were written at 1X, however, the Yamaha writer showed an increase in overall digital errors. Uncorrectable errors were experienced only in 1X mode on discs recorded with this machine.
Sony CDU 920S
The Sony CDU 920S was relatively new when the testing crew began using it in the latter part of the test period. It was introduced at under $2000 retail, but contrary to some negative reports about similarly priced writers, the Sony CDU 920S was an exceptionally good performer among all the writers used.
The Sony 900e, although it is now produced only in limited quantities, continues to be a competitive performer, as did the other elder statesman on the writer roster, the late Philips 521. Although both had been around for several years, they tested well as dependable and consistent writers.
Return to top of the page
The Message in the Media:
The tests undertaken by the One-Off CD Shops report average scores for each test undertaken on several discs from each manufacturer. Over the time these tests took place, one brand of CD-R media, Plasmon/Airy, was discontinued, and media from another manufacturer, Ricoh, ceased to be available for testing after the early testing period. While the test results did vary across time, almost without exception, testing results improved; for the one "AWOL" brand (Ricoh), the early results are probably not indicative of the way the current media product from the company now behaves. And even though Ricoh has some of the worst scoring (especially in the BLER results), it is important to note that like all other tested media (except for the discontinued Plasmon/Airy), Ricoh discs tested well within the accepted specifications. For instance, none of the E32 scores for all tested media had average scores that indicated errors that were uncorrectable by all three levels of error correction.
The BLER test yielded average scores for all the discs well within the standard specifications. For instance, Ricoh, which had the worst score with an average BLER count of over 60, was still well within the maximum value of 220 BLER.
Ricoh media tested second worst (Plasmon held the highest score) for the average BRST score, right at the maximum value of 0.7.
All the media averages were well within specifications for Average I3/Itop -- ratios between 0.3 to 1 and 0.7 to 1 -- showing that the minimum pit lengths compared to the T length was correctly formed. The I11/Itop measurements were also within the specification, and in all the media brands tested, exceeded the minimum ratio of 0.60 to 1, which meant that the pit length maximums were correct.
Physical errors pertaining to drive performance, track pitch, jitter, speed, and so forth were not addressed in these tests. Also, a new measurement called format error or logical error was not used.
Return to top of the page
Keep in mind that the most important performance issue for CD-R media and writer is controlling the levels of digital errors. While all other parameters contribute to high or low rates of errors, the actual error rates directly indicate a disc's "readability" and control the stability of the written disc from drive to drive. These tests showed that all the media brands tested that are currently available meet or exceed the standards accepted in the industry -- a very encouraging conclusion for the users and providers of this technology.
The Mitsui, Plasmon, and Verbatim discs were manufactured under Mitsui's phthalocyanine dye "gold and gold" process. The TDK, Fuji, Ricoh, and Taiyo Yuden discs were manufactured with a cyanine dye ("green and gold" discs). Both dyes are similar in basic function. In earlier tests, the cyanine discs performed significantly better on the HF signal measurements (I3 and I11), however, as they have been brought to 4X-compatiblity, the performance of all media has become much more consistent across all manufacturers.
This product is no longer available and it performed relatively poorly last year. An improved version of this media is now marketed directly by Mitsui Toatsu Corporation.
TDK had the lowest average BRST rate and also had the lowest maximum per second and average 10 second BLER. Although the average BLER was not the best, TDK discs displayed the most controllable maximum errors. TDK discs also scored well in symmetry and radial noise maximum, but were lowest for radial noise average.
These discs scored best on the maximum BLER and BRST tests for a one second interval, but their overall average BRST was behind TDK's. Taiyo Yuden discs measured second highest on the I11/Itop test, otherwise, these discs "split the pack" on most tests.
Verbatim media were the top performers for Cross Talk, and were very close to TDK in controlling maximum BLER rates. Their average I3/Itop was lowest but still well within specifications. Verbatim discs had no outstanding problems relative to all other brands tested.
Mitsui has emerged as a much stronger performer in 1994, in the second part of the testing period, showing improvements in all areas due to an aggressive approach to product improvement. Although the product has its roots in the "Airy/Plasomn" manufacturing process, the Mitsui discs far outperform their predecessors, especially in reduced BLER and BRST averages.
While performing well in analog signal testing (they were best in the I11/Itop and I3/Itop categories), the Ricoh discs had some trouble with high digital errors, with the worst averages and maximums of all digital errors in the test set. The most notable problem was the discs' relatively high tendency to have uncorrectable errors. Ricoh test samples were not available for testing in the last test period, early 1995, which may well account for some of this media's low scores in some tests.
Samples of Fuji media were available only days before completing the tests, so only a small number were included in these results. The samples were written at 4X on the Yamaha writer. Fuji topped the "best" list for average BLER and lowest Radial Noise. Further testing will give a more accurate assessment of this brand, but the initial numbers are very positive.
Return to top of the page
CD-R media, in general, are improving in overall quality. All brands tested showed some improvement, and some made dramatic strides in quality over the time of the tests. The availability of several high-quality media brands is encouraging, and demonstrates aggressive research and product improvement by competitive manufacturers. All discs and writers are making the transition to high-speed writing with little compromise in quality or consistency. Both dye formulations -- cyanine (green) and phthalocyanine (gold) -- used in manufacturing media are resulting in acceptable media, and in the past year have become much less of a differentiating factor in readability than at the beginning of these tests.
On the whole, media variations influence quality much more than performance of the writers used for this test; variances in writer performance between manufacturers in this case was minimal. "My biggest wish for this industry," Scott Bracken says, "is that the reader manufacturers would work harder to make their readers CD-R compatible. In spite of the writers being able to cut a disc that is within specifications as shown by the testers, some players still can't read them well. This is extremely difficult to explain to customers." Bracken says that too many current readers are only expected to comply with the Red Book and Yellow Book, and not the newer Orange Book. "CD-R discs usually have lower error rates than replicated CDs, but that doesn't guarantee that they will work on every reader," he says.
Bracken, Pat Smith, and Mark Worthington believe there and other tests being conducted in the field are helping drive manufactures find out what they need to do to improve their players in the areas of readability and compatibility. These tests have shown that the manufacturers are refining their products, and improving the quality as time goes on. And these test are needed. Managers of some major archiving applications are today choosing other storage technologies that may not be as successful as CD-R in the long run, simply because they cannot justify using largely untested technology.
Admittedly, by any legitimate testing standard, CD-R readability and compatibility have a way to go. Early testing efforts such as those undertaken by the One-Off CD Shops, while not guaranteeing long-term CD-R readability, go a long way toward easing the anxiety that inspired the tests in the first place.
[Author's Note: Testing was performed by Scott Bracken of The One-Off CD Shops International, Inc., and Patrick Smith, then of Cassette Productions. These two companies are divisions of The Duplication Group of Salt Lake City, Utah. Special thanks to Mark Worthington of CD Associates for his assistance during the testing and in explaining some of the issues involved.]
Return to top of the page