The King of all CD knowledge
2002-11-06 by Paul Schreiber
OK, I'm taking a break, beating PIC code to death. And pulling my hair out over "Configuration
Bits"...ARGH........
I spent 3 YEARS designing CD players. I am the King of All CD Knowledge :)
a) only 38% of data on a CD is audio data. The rest is error correct and servo data. The CD is a
constant linear data rate, which means the SPEED must be changed at about 7 times a SECOND to
keep the data rate at the laser constant. The rotational rate varies 2:1 (600 to 300 RPM).
b) there are standard discs made in France (at $750 EACH) that are "controlled error" discs. Yep:
$750 EACH and there are 5 in a set.
c) All CD decoder ICs have a pin for /ERROR (active low). You connect this to a pulse counter.
You run the test discs. You look at the count. ZZZzzzzzzzz.....
d) Even the lowest, crappy Sony or Sanyo decoder chips will pass the standard test discs with
less than 5 errors. An error is defined as data not corrected. Different chips handle the error
data in different ways (depends on your patents). Philips uses the last 9 samples to "guess" what
the next sample *should* be. Sony uses a 'zero order hold' that simply outputs the previous
value.
e) Toshi Doi, who did the mathematical proof of the error correction, sat alone for 7 YEARS
writing out on over 1000 legal pads until he figured it out. No computer simulations, just a
calculator. Repeat: 7 YEARS. The foundation of the work is in a book called "Algebraic Coding
Theory" which was originally a Master's Thesis but it won all sorts of awards because you could
"prove" a FEC (forward error correction) would work mathematically. I dare anyone to get past
page 6. It's like reading Martian.
Paul S.
Bits"...ARGH........
I spent 3 YEARS designing CD players. I am the King of All CD Knowledge :)
a) only 38% of data on a CD is audio data. The rest is error correct and servo data. The CD is a
constant linear data rate, which means the SPEED must be changed at about 7 times a SECOND to
keep the data rate at the laser constant. The rotational rate varies 2:1 (600 to 300 RPM).
b) there are standard discs made in France (at $750 EACH) that are "controlled error" discs. Yep:
$750 EACH and there are 5 in a set.
c) All CD decoder ICs have a pin for /ERROR (active low). You connect this to a pulse counter.
You run the test discs. You look at the count. ZZZzzzzzzzz.....
d) Even the lowest, crappy Sony or Sanyo decoder chips will pass the standard test discs with
less than 5 errors. An error is defined as data not corrected. Different chips handle the error
data in different ways (depends on your patents). Philips uses the last 9 samples to "guess" what
the next sample *should* be. Sony uses a 'zero order hold' that simply outputs the previous
value.
e) Toshi Doi, who did the mathematical proof of the error correction, sat alone for 7 YEARS
writing out on over 1000 legal pads until he figured it out. No computer simulations, just a
calculator. Repeat: 7 YEARS. The foundation of the work is in a book called "Algebraic Coding
Theory" which was originally a Master's Thesis but it won all sorts of awards because you could
"prove" a FEC (forward error correction) would work mathematically. I dare anyone to get past
page 6. It's like reading Martian.
Paul S.