A more detailed approach about CDROM media & devices

30-7-97

Updated on 25-8-99


All you wanted to know about CDs and CDr media (and more)

This document is intended to introduce the user to the world of CD-Recordables and to clear some confusion on how the CD-Recordable works and how it evolved. It will also attempt to cover what type of media to use and which one is better specifically and for what reasons. There are also explanations to why CDrs last less in some ambients and longer in others. The data gathered here is recovered from personal tests and experience over the years of my CD burning life since I started in 1993 with my first Philips CDD521 then switched for a CDD522 together with a very nice and complete manual from which I studied the CD structure in detail. I still think this is one of the best CD Recorders produced in terms of head mechanism and stability that I have ever used and will continue using even if only at 2x speed. Current recorders still can not out perform it in terms of laser accuracy and bias (pit depth/slope) at equal writing speed (2x) (no, not even Plextor or Yamaha can match this device yet). We must all be very thankful to Philips for this great invention after the digital Audio CD itself which opened a door to a new era in audio and consequently storage.

DISCLAIMER

As usual, the information provided here is for educational purposes only and I may not be held responsible for the damages you may inflict to your poor CDR devices directly or indirectly or to yourself or any other person. What you do in your own time and space is your responsibility!

Now let's start from the very beginning.

The CD-ROM Standard

Physical Appearance

So what is a CDROM? It is a circular 1 mm thick 12cm in diameter shaped disc created by Philips after the 80s on which you can store up to 700MBs of any kind of data (which includes audio which is digital audio data). Formerly known to all of you as an Audio CD and born in the beginning for this purpose, it has evolved in steps meeting the demand of compact data storage thus opening our way to the multimedia world in computing and informatics. A standard was already existing for the known non recordable Audio CD called the "Red Book" which contained the data encoding rules for this type and actually the World standard to which ALL Compact disks must comply with still today (CDDA and Data disks). In Audio CDs, data is encoded using a 2 layered EDC scheme (C1/C2) together with data scrambling and interleaving. Unfortunately this wasn't enough for DATA CDROMs so a more  reliable Error Detection/Correction scheme was added called C3. These data CDs were introduced with the Yellow (CDROM/CDROM-XA) and Green book (CDI and OSs). Note that we are still in the NON RECORDABLE world here (and much confusion is made around this) and we are considering pre-printed glass CDs and no CD recordables at this point.

The Orange Book

Here came the famous Orange Book for our current CDrs around the 90s then modified around 1992 to add the multisession possibility. It is important to understand that the Orange Book is not considered a part of any other as it is a blank CD containing no data structure to identify it before you write to it for the first time. The only pseudo data a blank Orange Book CDr contains are some ATIP codes and a pregroove for laser tracking and alignment. This pregroove is a very small contiguous undeep gutter laid on the disk in a spiral form. This spiral exhibits a small jitter called "wobble" that has a frequency of 22.05KHz if the disk is spun at the correct speed. The Wobble is used for speed control and pregroove tracking - but this is not enough for the CDr to know where it is so the wobble is FM modulated with absolute time information or the ATIP (Absolute Time in Pregroove). This signal is present from the beginning right to the end of the pregroove. So as you see, the blank orange book CDr has still to become part of some  sort of book and as Philips itself states, one can consider it as a sort of "Chamaleon" that will sooner or later become either Red,Yellow or Green after finalisation.


Physical Layers - CDROM Construction

The CD-Rom is physically structured into 6 layers (Physical, EFM, Subcode, Sector, Track and Session) all placed between 2 plastic layers, one much thicker than the other, that actually form the CD. If you look carefully at any CDr, you will notice that the record dye is on the label side, not on the bottom side. Contradictorily this makes the label side the most delicate in terms of scratching and thus any mechanical mistreatment and must be handled with care. This explains why the "overcoating" or Kodak's "InfoGuard" patent is important to protect your data physically. The reason the physical structure has this layout is cleared when we come to the pit focusing mechanism. The data structure is so behind the thicker transparent plastic form and due to light refraction through the 1mm thick plastic some dirt or light scratches that lay on this surface are actually out of the laser's focus plane that is generated from underneath (on the top there is the dye then thicker plastic then the laser underneath facing upwards). The next paragraphs will describe the physical layers and are numbered in order of the layering, not paragraph indexing as I preferred placing the EFM and Subcode at the
end for a more direct comprehension.

1. Physical Layer

Now comes the physical bit reading part. The laser runs along the track and when it encounters a pit or bump the position of where the laser hits the dye is further away than where it was out of the pit so due to the refraction angle of the light due to the transparent plastic medium less light returns to the laser feedback couple and this is translated into an electronical bit. So a train of bumps or pits will generate a series of bits that form the data and data structure (we are talking low level here). Just to give you an idea of the dimensions we are talking about, a pit is around 0.2micrometers in depth, 0.4 wide and 2 pits are 1.66micrometers apart from each other (the track separation). These pits or bumps are layed out in a spiral from the center of the disc to the outer part. Now you know why laser positioning is so critical and important - but that's not all. In order to "burn" this pit, the laser has a time and accuracy that is indirectly proportional to the writing speed: at higher speeds the same CDr is less accurate in burning the pits and having less time it needs to increase it's laser "beam" power (Laser LED Bias or Optimum Power Calibration-OPC areas) in order to achieve the same pit depth in the same amount of time - this can be compared similarly to the more professional tape decks' head Bias calibration. Having considered this, we also must consider the quality of the CD dye we are using. If we have a cheap unbranded CD with a greenish pale dye and we are burning at 4x or even 8x normal speed, we will be pretty certain that the finished CD will be having a lot of trouble being read by regular CD readers. 
Infact if the used dye does not meet "melting time" specifications because it is of poor quality or contains impurities (so it will also not achieve a high contrast) then the pits will not only be too "shallow" but due to the low reflectiveness of that dye, the light contrast between no pit and pit will be too little and the reader will need to EDC or re-read often the same sectors to recover the correct data giving a very slow transfer speed or even total read failure. This explains why a low quality high speed burned CD put in a 40x CD reader will be read in even more time than a older 4x model as the 40xMax drive will have to continue slowing down until it can reach a decent read speed - not to talk about fragmented data over the low quality CD which will drastically increase readback time due to incorrect positioning and localisation of the sectors. Viceversa, if we have a good CD (Kodak Gold for ex.) but our CD Recorder has it's laser LED and tracking mechanics warn out (yes they have a short MTBF = Medium Time Before Failure) then we will obtain a poorly mastered CD even if we have a good dye. It is important to more or less calculate the burning hours of your CD recorder and see if they reach specifications limit. Most important of all is the ambient in which the CDr is positioned and used. 

-Performance degradation caused by dirt particles and smoke

Dusty rooms and places where people smoke a lot are a killers combination for your CDr's Lense positioned in front of the beam. The nicotine and smoke particles exhaled by you and emanated by your lighted cigarette will saturate the room and be sucked in by your PC's venting or your CDr's external housing or tray. Even worse, if you have a spinning CDROM this will create a little whirl wind around the tray and the laser lense. All this brings dust and vaporised nicotine particles in the unit and so onto the lense and under (!) it. The condensation of the nicotine on your laser's lense creates a very sticky and opache coating which is not removed by commercial lense cleaning CDs and can only removed by contact - not a very easy and safe thing to do. Unfortunately the lense is composed (like any ordinary lense) by an exposed and hidden surface which is also open to air. The exposed surface is the one you see directly on the head mechanism (if you've ever opened any CD unit) but the other side is underneath facing the Laser LED. This is all mounted on a floating solenoidal platform electromagnetically controlled in order to always keep the above CD groove perfectly in focus. This also becomes practically impossible when the lense has been nearly obscured by the nicotine. If you want to know what the laser sees through that sticky mould, try exhaling against a cold window and then try looking through it! The only difference is your exhaled water vapour will eventually evaporate from the glass into the air whilst the nicotine will never! Remember "Smoking damages your health and all your optical devices and more!"


4. The Sector Layer (2 and 3 have been jumped)

We'll hover over the EFM (Eight to Fourteen Modulation = 2nd Layer) and Subcode structure (=3rd layer) and jump to the Sector structure description as it is more important to understand the different types to get a better picture of why the same empty size Orange Book CD can contain more or less data (most of the people still think it is a special trick or compression!). I'm not referring to the so called"overburning" (a method where a CDr can be forced over it's reported max capacity using a software trick if your CDr's Bios allows it) - this is something else which will be discussed maybe later on. I am referring to an increase of total user data space in MBs due to different used sector modes.
    In Audio CDs, we only need to localise the beginning of each track, there is no way to jump to a particular part of that track directly if no sub indexing is used - all you can do is FFWD/RWD/skipping to that position unless you have priorly stored a marker when listening to that CD - but this is an optional of your CD player not pertaning to the Red Book CD itself. This is ok until we talk about DATA. As you now, on any HD data can be contiguous or fragmented around the disk, but the sectors are still read back in the right order as they are "chain listed". Similarly this is what the data sectors in the CDROM do. These data sectors contain the user data AND additional EDC and positioning data giving us a 2352 byte Data Sector. What exactly these sectors contain depends on the mode of the sectors. For computer/data applications these are contained in the Yellow Book Mode 1 and 2 and also CDROM-XA/Green Book Mode 2 Form 1 and Form 2 (quite rare) for compressed audio/video. We can have 5 different sector modes (Sector modes not Disc Types!):

1. Red book 
[2352]Audio Sample Bytes (=75 sectors per second for 44.1Khz 16Bit Stereo = 333000 in 74min)

2. Yellow Book Data sector Mode 1
Sync[12] Header[4] User data[2048] EDC[4] Zero[8] ECC[276] {=[2352]}

3. Yellow Book Data sector Mode 2
Sync[12] Header[4] User data[2336] {=[2352]}

4. CDROM-XA and Green Book Data sector Mode 2 - Form 1 (Computer Data)
Sync[12] Header[4] SubHeader[8] User data[2048] EDC[4] ECC[276] {=[2352]}

5. CDROM-XA and Green Book Data sector Mode 2 - Form 2 (compressed A/V)
Sync[12] Header[4] SubHeader[8] User data[2324] EDC[4] {=[2352]}

As you can see type 2 and 4 have the least User Data space (2KBs) but the most data security/integrity (276 ECC bytes + 4 EDC = Third Layer Error Correction /C3). The Sector Sync is used as padding bytes to synchronise the beginning of the sector correctly and is always 00 FF FF FF FF FF FF FF FF FF FF 00. This explains why the same CD media can contain slightly more data than it's twin as it can be recorder using different modes. The disc containing more user data is ofcourse less reliable and if heavily scratched may become unusable in some points where the one with type 2 sectors may be recovered due to C3 error correction. Considering that there are around 336000 sectors on a CD (depending on the brand - they are not always the same!) we gain 288bytes every type 3 sector in comparison to the type 2 one. This gives us a total of nearly 95MBs of more user data space but a major data integrity loss hazard (650MB with type 2 and 750 with type 3). Current CDr applications default to or force Yellow Book mode 2 but with some obsolete older ones it was possible to select Yellow Book mode 1 thus giving more user space.

5. The Track Layer

Let's go one step further again into the complete Track structure. At this point, we have all our data sectors together with all the user data in whatever mode which all form 1 track. For ex. in CDDA, 1 track is 1 song, usually containing a pre-gap of 2 seconds which is not part of the track. In data CDROMs however, 1 track can be one complete session, or there can be one 1 Data track and the rest audio tracks all in one session depending on the TYPE OF CDROM/DISC you burned. Every track is listed in the so called "Table Of Contents" or TOC in the beginning part of the CDROM. This TOC is placed in the Q channel at Lead In and containd the list of all the Absolute Time positions of the beginning of the tracks on the current CD. Think of the Q channel as data extracted from the Subcode after the EFM demodulating process (part which we skipped). You see, data bytes are not present on the disc as listed above, but are present in EFM frames which conatin 588bits each (24sync bits, 14bits per data byte (33 databytes), 3 merging bit per databyte and 3 closing merging bits). From these EFM frames the bits are demodulated and sectors and subcode channels are extracted (there are 8). The Q subcode channel contains different information depending on where the laser is positioned, if in the beginning or Lead in in the Program Area where User data is found or the Lead Out. For Example if I am listening to an Audio CD and am on Track 4, the display of my CD Player will show me the relative time from the beginning of the track. This info is extracted from the Q channel frame in mode 1. Also if I want to search forward or backwards, the same Q channel info is used to run along the sectors.

6. The Session layer

At this point, we are ready to write session information. After the track(s) have been written, the session must be closed. This is achieved by writing the Lead In (LI) and Lead Out (LO). The Lead In containins the TOC in the Lead In Q channel (as stated above) listing all the tracks in the session, Disk Label, Unique Disk ID Number all stored in the subcode channels; the data sectors contain digital silence). The Lead Out is written at the end of the last track and contains the LO code in it's subcode area (TNO=AA). The Lead In and Lead Out and the information between these blocks is so called a Session.

Disc Types

On CDrs we have the "liberty" of writing different kinds of discs due to the fact that there are different kinds of track modes - we can have a mixed mode CD for example where we have a data track at the beginning and the rest are audio tracks. We can have a multisession Disc, where we have written 2 or more sessions/tracks in different lapses of time. Recently some new types of CDROMs have surfaced - one which I created in one of my first tries in experimenting with more user powerful programs some years ago for example where the audio tracks are at the beginning of the CD and the data track is at the end(!). Nowadays there are more than 7 different Disc types that can be found on the market or burned. Among these are: 

- CDDA Audio (Red Book)
- CDROM Mode 1 (type 2 data sectors) or 2 (type 3 sectors) Mode 2 are usually XA though (type 4 or 5 data sectors see CD-XA)
- Mixed Mode (Data track + Audio)
- CDROM-XA Mode 2 (Form 1 and Form 2) for multimedia (audio/video interleaving)
- CD-I (Green Book - Same as CDROM-XA sectors Type 4 and 5)
- CD-I ready Type I (special CDDA disks with additional song info in pregap)
- CD-I ready Type II (Multisession CDI with Audio track in session 1 and CDI in the next session)
- CD-Bridge (in order to Play a real CDI on a CDROM-XA unit as data starts in different positions for the 2 standards: for CDXA disk label points to 00m02s and 16sectors with byte offset 1024 while the CDI byte offset is null) An example is the Photo CD (Data tracks all mode 2 no mode 1)
- CDROM Multisession (more sessions written in different laspes of time)
- CD Bridge Hybrid (different kinds of sessions on same disc)
- (*)One Session Fixated Disk (Orange Book - one session is fixated and another is present but not final fixated so more can be added)
- Fixated Disk with sessions (becoming Yellow or Green Book - more sessions can still be added)
- Final Fixated Disk (Yellow or Green Book - No more sessions can be added as in the last written session LI space there is no pointer to the next)

(*) note how the NON FINAL FIXATED DISK is NOT part of ANY other book other than Orange Book. Only final fixated CDs become part of a book according to the type of data and tracks you have written on them. So called "Open Session" CDs are not Final Fixated thus remain in the Orange Book.


Track At Once Vs. Disk At Once

Your CD recorder has the capacity to write tracks on your blank CD - but this can happen basically in 2 ways: write each track in seperate instances or writing the whole CD in one go from beginning to end. But where's the big deal? Well, when you write your CD using track at once, the laser turns on and off every time a new track is written, even if you write it in one instance but you do not select DAO mode, Track at Once is used. The Lead In and Lead out are also written after the track - so first the user data is written to the track and then the laser turns off goes to the beginning of the CD and writes the Lead In then to the end for the Lead Out. This creates some "run in and out" sector gaps between the last written sector and the first written one in the next write instance. This isn't usually a problem for PC CDROM's but can be for other non standard devices so it is usually a good habit to master in Disk At Once mode all CDs to be passed into small distribution like demo CDs, Multimedia catalogues and similar data to ensure better read compatibility between platforms and CDROM readers. Ofcourse in order to write in DAO mode you have to have all the data ready on your HD, hopefully defragmented if not fast enough (difficult nowadays!) to avoid buffer underruns. Also, DAO mode prevents any further writes to the disk, so you cannot create a multisession CDROM in DAO mode - it would be paradoxal. Basically DAO mode is used for duping CDs and mastering them - this ensures the best reliability. DAO mode also enables you to eliminate inter track sector gaps like the 2 seconds pre-gap on Audio CDs before each track - ofcourse Track 1 always must have these 2 seconds which are inserted automatically by the CDr or software. 
    A typical problematic example is mixed mode playstation CDs. These usually contain one data track and the rest audio. If you try copying a psx CD not in DAO mode the CDr software will write the first data track then turn off the laser find the audio track and start recording again - this for all tracks. Finally it writes the LI and LO. The PSX does not like these small gaps which create data sector "gaps" and they usually do not work as the absolute sector tracking fails.


CDr Media Manufacturers

The 2 main manufacturers of CDr media are Kodak (the only genuine "gold" which is found in OEM forms as Traxdata gold and a few other) and Philips for the "green" type patented by Taiyoiuden (the inventor). A third variation has been spread over the years which is the "blue" type (Verbatim) and "platinum" (Philips). Each one of these moulds have a different characteristic and target use actually. Basically the Kodak Gold is optimum for data storage and functioning over a wide variety of CD readers - even problematic ones. It can be also used for audio even if it is not the best for this specific use. Domestic Audio CD players have a different led calibration (less bias) as tuned for "glass CDs" or rather the standard printed audio CD one buys. These "transfer printed" CDs yield the best performance on any CD reader of course as they offer the most contrast due to the material with which they are made of. This is why more specific "audio CDrs" appeared on the market - their dye attempts to reflect the glass CD's characteristics to meet as much as possible it's response so domestic CD players would not reproduce jitter or even skate. These CDs are of the "platinum" or "dark blue" series. If you're looking for audio CD creation for demos/promos, then you should try buying "Platinum" types which usually cost as much as gold ones. Specific "Audio CDrs" are also sold as stated above which are actually a "platinum" type but the cost is way too much (up to 10 times a normal goldie) for what they offer. Also remember the above relationship between writing speed and bit burning time which are indirectly proportional. The slower you burn a CD, the better quality your pits will appear to any reader, and the easier it will be to readback the data - this is especially for audio CDs. Contrarily, this rule can be used to test an unknown manufacturer's CD dye quality. Buy a few different brands of CDrs and try duping a FULL audio CD (the bigger the better) at maximum record speed. Next, put the burned CDs in your domestic Audio CD player (not your PC's CDROM reader which has a much higher Bias!) and skip around the tracks going from the last ones to the first ones and searching backwards and forwards. If the CD is of bad quality your reader will have a really hard time in positioning it's laser along the groove and you will probably notice some delays in response or even radical skipping even of one complete track or more (total loss of positioning). Another test is to listen carefully at low volume parts from classical movements and listen carefully for jitter noise (slightly cracking or dusty noise, clipping etc). This noise is created due to the very bad audio byte readout from the CD (there is no error recovery here) and is translated into awful an sound. You can also try tapping on the CD Reader's side or front and see if it starts skipping too easily or try with 0 volume to listen to the head track/focus mechanism and see if you hear a too loud "squeaking" sound which is produced by the solenoid in the laser when having difficulty tracking the groove. I can think of more, but I think this is more than enough to test the quality of the CDs to the limit.

I still feel this document is not complete as it lacks some more problematics related to current fast CD readers (vibration problems) and problematics related to printed CDs in this case. Also I would like to add a small guide to clean and try to recover a bad CD Reader/Writer with some cleaning tips for the more expert. If I have the time and will I may add this to the document to make it even more complete and useful so the final word has still to come.

In the mean time,

Happy CD recordings!