40
DVX
Instruction
Manual
CD Errors
E11 errors are the smallest errors, and are essentially random. High E11 error rates can be caused by excessive jitter, or
low signal-to noise ratio. E21 errors represent slightly larger defects like dust or small scratches. E31 are again larger
defects that cannot be corrected at the first stage of error correction (C1), and are de-interleaved and passed to the
second stage (C2). De-interleaving of the data turns one E31 error burst into as many smaller errors. Each E31 error
can produce as many as 28 E12 errors. It is not possible to predict how many E12 errors will be generated for each E31
error unless you know the exact location of the bad bits. This de-interleaving explains why E12 peaks can reach many
time higher than the other errors.
E22 and E32 errors are caused by fairly large defects, or mis-tracking. A common cause of occasional E22 errors on
CD-R’s is a spec of dust on the disc while recording. This will block the recording beam and create a void where there
are no pits. Recordable media should be absolutely clean when recording.
DVD Errors
DVD error correction is much more powerful due to the larger block size, but this also makes it difficult to determine
the size of defects, and very small defects will not show up as clearly as on a CD. The “Sum1” test detects the number
bad bytes in rows and columns, plus the number of uncorrectable rows and columns in one ECC block. This is the
highest resolution view, so small defects should show up best here. On a good disc, PI Bytes will represent random
errors, and so should be fairly low and uniform across the disc. This is comparable to BLER on a CD, so it gives a
general view of disc quality, but as with BLER, does not show the severity of defects. On a good disc, PIF error will
not exceed four per block, and of course there should be no POF errors.
Similar to C2 errors on a CD, the PO errors can be much larger than PI errors. For example, one PIF error means that
one row was uncorrectable, so in principle, there could be as many as 208 bad bytes in that column, so PO Bytes could
be as much as 208 times PIF.
The “Sum8” test detects the number of bad rows and columns in 8 ECC blocks. A small defect that only affects one
block will be averaged out over the 8 blocks, and therefore not appear as prominent. As with the Sum1 test, POE errors
can be, in principle, as much as 208 times as big as one PI error. If all 8 blocks contain PIF errors, then POE could be
as large as 1664. The upshot of this is that POE error rates are typically much higher than the others.
On a good disc, you should see PI Bytes and PIF fairly low and uniform across the disc. PO Bytes and POE will be
larger and more random, but still should be fairly uniform over the disc. Good discs should have no POF errors (see
Fig. 26 below).
