A "standard" CD is 74 minutes long.
Each minute has 60 seconds.
Each second consists of 75 frames (aka sectors).
Each frame holds 2048 bytes of data (in Mode 1 format).
74 * 60 * 75 * 2048 = 681,984,000 bytes total data.

https://en.wikipedia.org/wiki/Rainbow_Books

I'm certain those books contain all the info you need and more.
Have fun going down the rabbit hole that is CD capacity.

The original standard audio cd had like 74 minutes of audio playtime. this is all described in the red book
with 16 bits per channel and a 44100/s sample rate
74 minutes * 60 seconds * 2 bytes * 2 channels = 783216000 bytes and that's only the roughest of capacity calculation
now add to that some error correction, some lead in/lead out, spacing etc...
then came the yellow book for cdrom

oh well. enjoy !

Most newer CDR are "80 minute" which are actually 79m 59s (~703Mb). Some "80 minute" CDR can overburn as much as 97 minutes depending on the drive and media:
https://goughlui.com/2021/05/29/tech-flashbac … sults-database/

The old scene standard for DVD rips to be burned to CD-R was 700MB.

https://slate.com/culture/2008/11/the-secrets … vie-pirate.html

Finally, the files are a predictable size: right around 700 megabytes, the amount of data that fits on a single CD-R.

700MB 80-minute CD-Rs came later but relatively early in the mainstreamization of CD-R, maybe around 2000-2001. Before then it was 650MB on 74-minute discs.

I have a vague recollection it may have been around the same time that CD-Rs (in quantities too small for spindles) switched to being sold in thin cases instead of the thicker jewel cases that audio CDs had used since the 1980s.

Remember, the CD was invented as an audio-only format (which, despite being without doubt digital, is in many ways closer to a cassette or LP than a bunch of files in the codec of your choice), so read the Red Book first (you can skip every 2nd page as that's the French translation, and the many pages about CD-Text and CD+G extensions 😉 - but do read about the Q subcode (which has a different official name), that's part of the "low level formatting"/sector addresses)

There you'll find the 74 minute nominal maximum capacity, and the manufacturing tolerances which, if fully exploited, give the 80 minute capacity;

Then you might want to read the Yellow Book (data CDs), where limitations caused by the "analog characteristics" I hinted to are worked around at the expense of sector capacity (spoiler: exact seeking to a specific sector was not in the original specification)

After that, I think (having not read that one) you should proceed to the Orange Book (writable CDs), which should repeat the 80 minute limit for ATIP (which includes the nominal capacity reported to the drive) 😀

Then go to Andy McFadden's CD guide, with a bunch of real-world experience (which trumps any wishful engineer thinking they try to sell for $$$) to see how the first CD-Rs were actually quite smaller than 74 minutes, probably to maximize then-unknown reliability 😀

Technology Connections has a good video on this topic: https://youtu.be/TCTWyNstpD0

jmarsh wrote on 2023-07-08, 07:44:

Each second consists of 75 frames (aka sectors).

That's what it is usually called by most people, but frames on CDs are entirely different thing and are 25 bytes long. I prefer to decode MSF to minutes, seconds and fractions (of a second) but then again I tend to think of CDs at a lower level and I need frames to mean frames.

And in general the answers given in this thread are correct but stricly speaking a CD-R has no set capacity. It has a nominal one (that is, a 74 minute media must be able to fit 74 minutes of audio) but since it is recordable you can't know in advance how long the lead-in is going to be. That depends on the writer and perhaps also the software in some cases. Then there's the outer edge and manufacturing tolerances, you are going to run out of reflective layer or recordable media and operating in this area is dangerous to the lens - so the lead-out must also have certain length but can be longer and you don't control that either.

Lastly one could do what SEGA did and just increase bit density, this would be compatible with many drives if the difference wasn't big, after all the disc is read at CLV so the spindle is already capable of variable rpms.

Deunan wrote on 2023-07-08, 21:03:

Lastly one could do what SEGA did and just increase bit density, this would be compatible with many drives if the difference wasn't big, after all the disc is read at CLV so the spindle is already capable of variable rpms.

Is that the only thing that constitutes a "GD"ROM

I think GD ROMs were CAV disks (and think that is what Deunan was implying) which modern burners can do, versus CLV which old cd burners did but can they read CAV burned disks.
or something like that. Sega was one of the first to put it to good use 😀

maxtherabbit wrote on 2023-07-09, 01:03:

Is that the only thing that constitutes a "GD"ROM

Higher bit density? Yes, and no. Yes because everything else is the same and in fact some standard PC CD drives can read the hi-density area once you fool them into thinking the loaded disk is just one long audio track - which is done with a special CD-R that is then swapped for a GD-ROM but without tripping the caddy sensor. In fact REDUMP insists this is the only proper way to do it, and anything else (including using the original GD-ROM drive in Dreamcast) is a heresy for which you'll be burned at a stake. And then they broke GDI format because reasons ("we fixed it" was their excuse) and then they switched to CUE format which can't even deal with mixed-mode tracks and multiple sessions on a standard CD, much less with anything else, so that too was "fixed" to make it work. But what do I know.

No, becaue there is also a ring with protection code in the middle between low and high density areas - but SEGA never used this properly and it's easy to bypass, so while it cannot be read by CD-ROMs, you also don't need it. GD-ROMs somewhat resemble multi-session CDs, except there is no link in the TOC of low density area, this is on purpose to make sure no drive would try to access the high density one.

The main issues in reading the high-density area on a standard CD drive are:
a) be able to adjust the spindle rpms pretty low - many drives will struggle to properly lock onto the stream
b) have a robust seeking self-correction - since the usual cheat tables that most drives use will put the head so far off that the drive can't deal with it
c) not trip on address data that does not match TOC (track type, track number, index)
d) ability to read hi-density area TOC - but that is optional, TOC data is also stored in the info/loader area of the first data track

Hope that helps and sorry for OT.

Deunan wrote on 2023-07-09, 17:27:

Higher bit density? Yes, and no. Yes because everything else is the same and in fact some standard PC CD drives can read the hi- […]
Show full quote

maxtherabbit wrote on 2023-07-09, 01:03:

Is that the only thing that constitutes a "GD"ROM

Higher bit density? Yes, and no. Yes because everything else is the same and in fact some standard PC CD drives can read the hi-density area once you fool them into thinking the loaded disk is just one long audio track - which is done with a special CD-R that is then swapped for a GD-ROM but without tripping the caddy sensor. In fact REDUMP insists this is the only proper way to do it, and anything else (including using the original GD-ROM drive in Dreamcast) is a heresy for which you'll be burned at a stake. And then they broke GDI format because reasons ("we fixed it" was their excuse) and then they switched to CUE format which can't even deal with mixed-mode tracks and multiple sessions on a standard CD, much less with anything else, so that too was "fixed" to make it work. But what do I know.

No, becaue there is also a ring with protection code in the middle between low and high density areas - but SEGA never used this properly and it's easy to bypass, so while it cannot be read by CD-ROMs, you also don't need it. GD-ROMs somewhat resemble multi-session CDs, except there is no link in the TOC of low density area, this is on purpose to make sure no drive would try to access the high density one.

The main issues in reading the high-density area on a standard CD drive are:
a) be able to adjust the spindle rpms pretty low - many drives will struggle to properly lock onto the stream
b) have a robust seeking self-correction - since the usual cheat tables that most drives use will put the head so far off that the drive can't deal with it
c) not trip on address data that does not match TOC (track type, track number, index)
d) ability to read hi-density area TOC - but that is optional, TOC data is also stored in the info/loader area of the first data track

Hope that helps and sorry for OT.

Thanks for the explanation. I come from the era of dcload IP dumping over the BBA, the concept of reading the disc in a PC drive is new to me. The physical attributes of the high density area were always a black box to me.

maxtherabbit wrote on 2023-07-09, 21:07:

Thanks for the explanation. I come from the era of dcload IP dumping over the BBA, the concept of reading the disc in a PC drive is new to me. The physical attributes of the high density area were always a black box to me.

Never had a BBA, too expensive, so I used the serial port with my own dumper (loaded via dcload CD-R) to transmit the GD-ROM data back to a PC. Took a tad bit longer than BBA though. Then someone made an SD card dumper - same idea except there's an SD card in SPI mode connected to the serial port, that's faster and can be used without a PC. The PC CD-ROM disc swapping worked for me too but the drives I had struggled a lot with GDs so I only really used it a few times.

BTW some people think that the changes SEGA made to CD drives to turn them into GD drives is what actually made them rather unreliable at reading normal CD-ROMs, especially the near the center of the disc - and that goes double for recordable media which offer weaker signal. Which is why GD drives struggled with "scene" game rips.