There really isn't a foolproof and good way to test this.
Scratches, misprints, dirt and other tracking/reflectivity problems on a CD take so many forms that it would take a really mastefully produced test disc to simulate all these in a reproducible and consistent manner (there are some commercial and expensive test discs out there that try to simulate these kinds of scratched/dirty discs).
The best tests I've seen (performed by Ian here at CDRLabs) use AFAIK a commercial audio CD with scratches on it (if I remember correctly). Or Ian has used at least two different discs, one that had more scratches and the current used disc that has less error inducing scratches.
With his current test disc, the best performing drives should IMHO report as high number of blocks on the test CD as OK (not 'damaged' or 'unreadable') WHILE also at the same time consistently reading the discs with same results (number of % errors should be small) AND also at the same time do this with a speed that is still bearable. All drives should also be able to report back C2 error data for the OK/damaged/unreadable numbers to be meaningful.
If a drive only reports 100% ok, but produces a lot of errors, that doesn't make it a good reader for scratched discs. Also, if it produces no errors, but has a high number of unreadable/damaged blocks, then it's also no good.
Theoretically it is also possible for a drive to report a low number of damaged sectors (lousy or non-existent C2 reporting with undetectable error concealment) while still doing this with a low number of read errors (it just reads the disc poorly, but in a consistent manner). However, I find this an academic issue as I have not seen such a drive in Ian's tests. Drives with no C2 usually fail in the CD DAE error count test as well.
Based on the above explained criteria (which I hopefully have explained faithfully here), the best performing CDRW burners for Digital Audio Extraction of scratched discs is (based on Ian's testing) are:
- Code: Select all
CD Blocks status DAE
Writer OK Damaged Unreadable Errors Speed
Samsung SM-348B 44% 56% 0% 0.28% 8.9x
Samsung SW-252B 38% 62% 0% 0.32% 9.1x
Sony CRX220A1 40% 60% 0% 0.37% 10.3x
LG GCE-8520B 32% 68% 0% 0.43% 8.9x
Some of the otherwise well received drives did not perform that well
compared to the above drives:
Plextor Premium 30% 70% 0% 12.35% 23.2x
LiteOn 52246S 30% 70% <1% test failed
TDK veloCD 522448 29% 67% 4% test failed
Teac CD-W552E 26% 38% 7% test failed
(with Teac test the percentages do not add up to 100%,
because the test was not able to complete due to the drive freezing up)
What I can deduce from above is:
Both Samsung drives do a very good job IMHO. So does the LG.
What is interesting is that the Sony drive did very well, but the LiteOn drive based on the same mechanism did quite poorly. AFAIK, both drives use the same mechanism manufactured by LiteON. Also, afaik, the test disc was the same. I have no idea if the differences can be explained by firmware differences or due to drive units being good and bad examples of what that mechanism can do (i.e. quality variance in manufacturing, slight misalignment in shipping or some other unit-to-unit variance).
It would be nice to see the test of another LiteOn 52x drive and see how it fared with these tests (I'm very eagerly waiting for Ian's 52327S review along with the aforementioned DAE test data).
As we do not have enough statistical data, I'm not sure we can fully trust the 'bad' results for some drives models for which there is test data inconsistency (like the LiteOn above).
However, if we believe that this is a rare anomaly, then it's pretty safe to say that Plextor premium is not perhaps the best drive to use for scratched disc DAE WITH maximum speed settings. It is not known how well the drive does if the speed is manually set to a lower setting (say 12x) or if another program is used to control the DAE process (say Plextools Professional).
This also raises another issue with the DAE test results applicabability: if the DAE performance of a drive can improve with a different software (that tests for errors on the fly in a different manner and manually sets the speed of the drive for difficult to read parts of the CD), then how well do the CD Speed and CD DAE tests correspond with the optimal performance of the drive. To tell the truth, I have no idea. Probably the correspondence is quite high for most drives, but for drives using special commands there may be interesting differences (i.e. Plextor being a prime candidate).
I hope I didn't bore you with my ramblings. Please feel free to correct any mistakes you find. I'm not an expert on these issues and I'm still learning a lot.