This article is from the CD-Recordable FAQ, by Andy McFadden (email@example.com) with numerous contributions by others.
An excellent reference for is Ken Pohlmann's mammoth _Principles of Digital
Audio, 4th edition_ (ISBN 0-07-134819-0), especially chapter 9 (on compact
discs) and chapter 5 (on error correction). If you want something a little
slimmer, try his older _The Compact Disc Handbook, 2nd edition_, 1992
Another good book is _The Art of Digital Audio_, 2nd edition, by John
Watkinson, Focal Press, 1994 (ISBN 0-240-51320-7).
http://www.ee.washington.edu/conselec/CE/kuhn/cdmulti/cdhome.htm has a
number of interesting pages. In particular, there's a good page about CIRC
and http://www.ee.washington.edu/conselec/CE/kuhn/cdaudio/95x6.htm has a
nice explanation of disc construction and optics, especially the three-beam
The page at http://www.tc.umn.edu/~erick205/Papers/paper.html provides
some background information on sampling, aliasing, dither, DACs, and other
You can get a copy of ECMA-130 from http://www.ecma-international.org/.
This document describes the format of a CD-ROM, including physical dimensions
and optical characteristics, as well as sector formats and Q-channel specs.
It also features some interesting annexes:
- Annex A: Error correction encoding by RSPC
- Annex B: Scramble (a description of the pre-EFM scrambler)
- Annex C: Error correction encoding by CIRC
- Annex D: 8-bit to 14-Channel bit conversion (has the full table)
- Annex E: Merging bits (algorithm for computation)
Standards documents, as a rule, are terse and difficult to understand.
ECMA-130 is actually quite readable, and if you understood the preceding
sections you should have no trouble sorting it out.
If you want source code for the CIRC, RSPC, EDC, and scramble functions,
look for Heiko Eissfeldt's edc_ecc.c (and related files). The code is
part of Mode2CDMaker, CDRDAO, and possibly others.
If you want an explanation of DSV and the problems associated with it,
read the Philips patent on the sector scrambler (US4603413), or one of
the associated patents on removal of DC content from a digital signal.
The full text of the patent can be found at http://www.uspto.gov/. In brief:
"[...] If the frequency of such oscillation is comparatively high,
during the read operation the decision level for detection of the
channel bit signals may be rendered inaccurate. As a result, read-out
of the information will be disturbed to such an extent that even the
error-correction measures cannot prevent errors. Moreover, the tracking
system for controlling the read laser which reads the channel bits may
become incapable of keeping the laser beam accurately positioned on
It appears that, when the DC offset in the signal becomes too large, the read
head has trouble "seeing" the disc. The voltage level in the photodetector
has pegged, so the difference between a pit and a land is unnoticeable.
An article at http://www.digit-life.com/articles2/magia-chisel/index.html
examines why one specific file failed to record properly. It turns out
that, after passing through the scrambler, a piece of the file has a
section that matches the sector header sync pattern.
For some technical information on how CD-Rs are constructed, look through
the uspto.gov site for relevant patents. For example, US5348841 describes
"Organic dye-in-polymer (DIP) medium for write-once-read-many (WORM)