Warning (2005-05-15): I CAN NO LONGER RECOMMEND RITEK MEDIA. This is due to a large decline in quality recently. See my blog for more details.
This page shows some testing results of DVD Recordable media. This testing is done to ensure quality of media I export from Taiwan to the US. The primary application is for video recording, so quality and compatibility are important. While overall DVD Recordable quality has improved a lot, some of the manufacturers still do not have sufficient quality for consistent compatibility with video players.
My test platform consists of a P4-2.53ghz system running Windows XP. It has three DVD Burners in it, all of which are on their own IDE controller as master with no other drives on the same controller. The first burner is a Pioneer DVR-105 4X DVD-R drive. The second is a Lite-On LDW-851S 8X DVD+R 4X DVD-R drive. The third is a Pioneer DVR-A07 8X DVD+-R drive.
Testing of 4X DVD-R media is done at 4X on all drives. Testing of 4X DVD+R media is done at 4X on the Lite-On and A07 drives. Testing of 8X DVD+R is done at 4X, 6X, and 8X on the Lite-On and A07 drives. Nero Burning ROM version 6 is used for all current tests. Some older tests were done with version 5.5.
A video image of 4.35-4.37gb size is used so that the test video completely fills the disc. (Note that while DVD single layer capacity is usually quoted at 4.7gb, this uses gigabyte as equivalent to 1 billion bytes, whereas a true gigabyte is 1073741824 bytes. I use true gigabytes instead which yields a capacity of 4.37-4.38gb depending on the disk.) Full burns are used because bad DVD Recordables often have problems on the outer edge of the disk. A partial burn would not reveal this problem. Rarely a disk will fail with a write error. Disks which fail with a write error almost certainly indicates an unacceptable quality disk
When doing test burns the "Verify written data" option is turned on. With this option, Nero will re-read the disk after it is burned and compare it to the original data to ensure that the disk is readable. A DVD which fails this test is almost certainly of unacceptable quality. A DVD Burner in particular and computer DVD drives in general tend to have the best compatibility with reading recordable media, so when a disk cannot be read even on the drive that burned it is probably not going to work well on anything else.
After the image is burned, it is tested for error rates and read speed in the Lite-On drive using a utility called KProbe. DVDs use an error correction scheme that has three levels of errors, called PI, PO and uncorrectable. PI errors are the first level of error and every disk will have lots of these. PO errors are the second level, and most disks will have some. Some of the best grades of media from Ritek, Mitsubishi or Taiyo Yuden will sometimes test with 0 PO errors, but this is rare. The last level, of uncorrectable are errors that could not be corrected.
The DVD Recordable specifications state that a disk read at 1X speed will have less than 280 PI errors per 8 blocks when calculated over 8 ECC blocks, and less than 4 PO errors per block when calculated over 1 ECC block. And it should go without saying that there should be no uncorrectable errors at all.
KProbe 1.x allows testing of PI and PO errors, but because compliance to the standard for PI uses 8 ECC blocks and PO uses 1 ECC blocks, two runs of the test are run to measure each. In the test results the filename ending in 1PIPO is the PIPO test with calculation over 1 ECC block and the one ending in 8PIPO is the PIPO test with calculation over 8 ECC blocks. In the PIPO test, the drive is locked at a fixed speed, usually 4X, and the media is read from beginning to end without stopping while error rates are recorded.
KProbe 2.x which was recently released allows setting the measuring of PI and PO ECC blocks separately now, which allows testing of standards compliance with one pass. New tests are done with PI over 8 ECC blocks and PO over 1 ECC block and are identified by 8PI1PO in the name.
In addition, while the specification says that error testing be done at 1X, I do testing at the media's rated speed, (i.e. 4X or 8X). There are two reasons for this. First, testing at 1X does not reveal problems which may occur at higher speeds. Second, testing at 1X takes about 1 hour per pass. Burning, verifying, two error check passes, and one speed test would take about 3 hours per piece, which makes it prohibitive to test more than a couple of pieces. Doing the tests at 4X or higher means the entire test takes a little more than 1 hour which is more manageable. But because of this, a little more tolerance should be given to the PO error rates. The PI error rates should still be lower than 4 in the 1PIPO test in my experience. In comparison tests, most disks had less than 20% difference between 1X and 4X disks.
The final test in the KProbe suite is the speed test. Actually there are many speed tests programs out there, and Nero includes their own. I use KProbe's for convenience and consistency of the output format. In the speed test the media is read at top speed for the drive and a graph is made of how fast the data was capable of being read. The top speed is usually variable across the disk because it uses CAV mode. For example, on 4X media the Lite-On reads at about 2.6X initially and ramps up to just over 6.2X at the outer edge. On a good disk the line will be a fairly smooth curve from beginning to end. Slight wiggles in the line are not necessarily a problem and can also be caused by the computer running another program. Large drops in speed can indicate a quality problem. Drops to 4X speed are usually not serious. Drops to below 2X may be a problem, and drops to 1X or lower are almost certainly going to cause problems.
But at the end of the day, real compatibility is whether it actually plays reliably on your target platform. In this case the target platform is consumer DVD video players. Early video players tend to be picky about playing DVD Recordables, and while compatibility has improved, even newer players have trouble with some recordable media. In general the lower the error rate measured above, the fewer the problems. Nevertheless, I also test by playing each video on 2-3 consumer video players, including one manufactured in 1998 which is a bit picky. A quality media will play on each player without problems. This part is also time consuming, so a sampling of scenes are randomly selected to try to spot any problems.
NEW (2004-05-06): Starting at this date, the speed test and the 8PIPO test have been eliminated to save time.
NEW (2004-06-02): Starting at this date the Pioneer 105 drive has been replaced with the Pioneer A07.
NEW (2004-07-23): Bought another IDE controller and moved disk drives around, so all three burners now back in the system. Also upgraded to KProbe2 so tests are now of the 8PI1PO variety now.
The media I've tested below comes from a number of sources. Some of them were purchased retail either in the US or Taiwan, while some were supplied as samples by potential vendors, which could be the manufacturer itself, or a trading house or broker. I've identified each media by the manufacturer code.