Chapter 10: Data Recovery from CDs

How to Recover Unreadable CDs?

CDs are not extremely reliable media. You can t insure against their surfaces becoming scratched or contaminated, even provided that you store them very carefully (sometimes, the CD drive itself scratches the media, and there is nothing you can do in order to prevent this). Even a disc that appears to be OK at first glance may have hidden flaws that render it totally or partially unreadable on standard drives . This problem is especially urgent for CD-R/CD-RW discs, whose manufacturing quality is still far from perfect, and the process of burning can be affected by various errors.

But even a disc that has physical defects of the surface can be read successfully, thanks to the excessive redundancy of the information that it stores. With the gradual increase of the number of defects, however, the correcting capabilities of Reed-Solomon codes suddenly cease to be sufficient. When this happens, the disc becomes unreadable for no visible reason. The drive can even sometimes refuse to recognize it.

Fortunately, in the vast majority of cases it is possible to recover the information stored on the disc even under these conditions. The next section will explain how to do this.

General Recommendations on Recovery

Not every CD that is unreadable (or is unstable) is actually defective. Most frequently, it is the operating system or the CD drive that is to blame. Before jumping to any conclusions, try to read the disc on all drives available to you that are installed on computers with newly installed operating systems. Many drives, even expensive brands, become extremely capricious after a short period of use. They often refuse to read discs that can be read without any problem on other drives. With regard to the operating system, it tends to catch various strange bugs as you install, reinstall, or remove various software. Sometimes, these bugs manifest themselves in unpredictable ways. For example, the TEAC drive installed in a system with a CDR4_2K.SYS driver inherited from PHILIPS conflicts with the CD Player, refusing to display the contents of data discs when that application is active. After removing the CDR4_2K.SYS driver, everything works without any problems.

Also, it mustn t be forgotten that the corrective capabilities of different drive models are very different. For example, correcting codes of C ₁ , C ₂ , Q-, and P- levels can be accomplished effectively by all drives with which I am familiar. Their corrective capabilities is up to two errors for each of the C1 and C2 levels, and up to 86 and 52 errors for the Q- and P- levels, respectively. To be honest, the number of detectable, but mathematically irrecoverable, errors is as high as to 4 errors for C ₁ and C ₂ levels and up to 172/104 errors per Q/P levels. But only the position of erroneous bytes in a frame or sector, and not their values, can be determined with any guarantee. Knowing the position of erroneous bytes and having at your disposal the source HF signal (e.g., analog signal taken directly from the read head), however, it is possible to recover some odds and ends of information ”at least, theoretically. However, in my experience, the price of the drive is rather weakly correlated with its reading capabilities. For instance, relatively inexpensive ASUS drives read practically everything, while expensive PHILIPS drives recognize even their native discs with native drivers less consistently.

Another important characteristic is the available range of read speeds . In general, the lower the rotation speed of the disc, the lower the requirements on its quality. However, this relationship is not always linear. Most drives have one or more preferred rotation speeds, at which their reading capabilities are at a maximum. For instance, it is possible that a defective disc is readable without any problems at 8x, but is totally unreadable at any other speed. The optimal speed can be determined easily by experiment. All you have to do is test the full range of available speeds. When purchasing a CD-ROM drive, choose the device that has the widest range of speeds. For instance, the PHILIPS CDRW 2400 can operate only at the following speeds: 16x, 24x, 38x and 42x. The lack of speeds such as 4x and 8x limits the ration of this drive to high-quality discs only.

For some unknown reason, Windows built-in tools do not permit the control of disc-rotation speed. Therefore, to gain such control, it is necessary to use third-party utilities. Fortunately, there are a large number of these tools available. For example, you can use Slow CD, Ahead Nero Drive Speedy or any other similar tool. Generally speaking, most drives reduce the rotation speed on their own initiative if they encounter unreadable sectors. However, the quality of algorithms they implement for this purpose is still far from perfect, and user s control over the rotation speed provides considerably better results.

If the disc cannot be read on any drive available to you, it is possible to try polishing it with any polishing paste. The Internet provides a lot of information on polishing optical surfaces in general, and CD surfaces in particular. As a matter of fact, books on astronomy, especially those concentrating on the construction of telescopes, are the most useful in this respect. Because of the general availability of such information, I will cover this topic only briefly in this book. In the majority of cases, it is possible to polish a scratched disc. If done properly, it is highly likely that the information can be restored. However, a number of factors do limit the degree, to which this is true. First, polishing repairs only the scratches on the lower surface of the disc, and cannot deal with the destruction of the reflecting layer. Second, when removing scratches, you inevitably create new ones. Consequently, some CDs can become damaged even more badly after polishing. Third, it is impossible to acquire the skills of polishing CDs on your first try. To master this skill will take much time and a number of attempts. But we ll try another way!

What, in fact, is very good for your disc is sponging it with a special anti-static liquid (you can buy it in your local computer store). Before wiping the disc, blow off all dust particles from its surface, to avoid producing more scratches. Never wipe the disk with concentric circular motions ! It is necessary to wipe the surface with radial movements, from the center to the edges, replacing the cloth with each pass.

The Disc Cannot Be Recognized by the Drive

You insert the disc into the drive. The drive starts to rotate the disc, with the activity LED blinking wildly, making sure that at the specified speed, the disc cannot be read. It then starts to reduce the rotation speed until it comes to a full stop. The DISK IN indicator (if it is present on the front panel of the drive), sadly, blinks and goes off, thus signaling you that the piece of plastic that you have inserted, from its point of view, is anything you want it to be except for a CD. An attempt to access the disc results in an error message informing you that there is no disc in the drive and prompting you to insert one.

The inability of the drive to recognize the disc is, in most cases, evidence of a malfunction in the CD-ROM. Cases where the problem is caused by a damaged CD are much more rare. Even if this disc was easily recognized yesterday , and even if the drive successfully recognizes all other discs, don t rush to assume that the drive is working properly. Try to read this disc on another drive. Worse comes to the worst, reduce the disc rotation speed down to minimum. Be prepared for the fact that the drive may not obey you. Most drives automatically reduce the previous rotation speed settings when replacing the disc and do not allow you to change the speed until the disc has been recognized successfully (in particular, this is particularly characteristic of TEAC drives, while ASUS drives usually behave less temperamentally).

If the disc being tested cannot be recognized by any of the drives available to you, the most likely cause is that the drives cannot read the disc s TOC, which is stored in the Lead-in area. Remove the disc from the drive and take a careful look at the narrow glittering ring located near the internal edge of the disc. This ring is the Lead-in. Make sure that it has no deep scratches or is not contaminated. If it is dirty, remove the dirt using a clean cloth (most individuals forget about the Lead-in area when cleaning the disc, probably taking it for some kind of useless decoration). Overcoming scratches is significantly more difficult. Without enough experience at polishing CDs, it is not advisable to attempt this. If this is the case, bring the disc to a center specializing in information recovery. However, these aren t always easy to locate and, further, this doesn t guarantee success. Finally, it is wise to consider confidentiality and the cost of the service, among other factors.

Is it possible to reanimate such a disc on your own? It is! However, in order to do this, you ll have to purchase certain equipment, at a cost of around $30. You will need a spare CD-ROM drive, with which you can experiment as you please and the loss of which won t be a problem. Low-speed drives left over from previous upgrades of your computer are the most suitable for this purpose.

The trick is that the TOC is not a must for working with CDs at the sector level, so you can do without it. In fact, this is not a hardware problem ”it is the software problem. Having determined that irrecoverable errors have occurred in the course of reading the TOC, the drive firmware refuses to process the disc, despite the fact that TOC content is duplicated in the Q subcode channel and distributed over the entire spiral track. Furthermore, the drive actually only needs three main fields of the TOC: the address of the disc Lead-out area (in order to know, up to what position it is possible to move the head), the starting address of the first track (in order to know where to start reading the data) and the address of the next Lead-in area (only for multisession drives). The easiest task is dealing with the starting address of the first track ”it is equal to 00:02:00 (which corresponds to the zero LBA address). The Lead-out address, directly dependent on the CD capacity, does not necessarily need to be specified with high precision. It is enough to choose it so that it is no less than the address of the current Lead-out. Otherwise, all of the sectors located beyond this boundary will become unavailable. By setting the Lead-out address to 80 ” or even 90 ” minutes, we can guarantee that the entire disc surface will be available to the drive. Briefly speaking, if we can get access to the internal structures of the drive firmware, the recovery of the corrupted TOC would be a trifling matter. For this purpose, I use specially modified firmware of a quite ordinary, old CD-ROM drive (this is an old 8x, no- name brand), which allows for the manipulation of any service data. Therefore, it reads everything that is physically readable.

If hacking microprocessor programs is too difficult for you, it is possible to proceed using another method. Carefully disassemble the CD-ROM drive and extract its internals from the case (the cheaper the spare drive, the better). Now, unscrew the bolts that fasten the metallic strap, on each a plate is fastened, that is pressed to the upper edge of the CD, to prevent it from sliding. Instead of this assembly, you can use a metallic ring, on any other heavy object. The main idea is to get free access to the CD, thus providing the possibility of hot-swapping it on the fly, without opening the tray.

Now, connect the drive to the computer, power it on, and proceeding in a normal way, insert a specially prepared disc into the drive. The Lead-out area of the disc must be located somewhere around 80 to 90 minutes (it is possible to insert any CD with the video, with the size starting from 700 MB). Make sure that the disc has been correctly recognized, and, without shutting the system down, remove it from the drive without opening the tray. Now insert into this drive the disc that you want to restore. Since the TOC of the previous disc is already loaded into the cache, and the drive is unable to detect disc replacement carried out in such a barbaric method, it will work with the new disc just the same way it would have with the old one. The only thing that you should not try to do is read the disc contents using the built-in operating-system tools. This won t produce any result. After all, the operating system also caches data. Therefore, you can click Refresh until you go crazy, but Windows will continue to display the previous contents. Instead of this, take any grabber that can read the disc at the sector level and doesn t ask any extra questions (for instance, you can use the free cd_raw_read utility by the author of this book) and copy the entire contents of the disc from the first sector to the last into an image file. Then, using any suitable CD-burning utility, write this image to CD-R or CD-RW. Although you won t recover the disc itself, you will at least save its contents! This method can be used both for audio and data discs with equal success.

As an alternative, instead of removing the pressure plate, you can find the sensor for disc replacement and temporarily remove it, thus preventing the drive from noticing that the disc being recovered has not been replaced . Cheap drives use simple mechanical sensors that can be found easily. More expensive models do not contain a separate sensor at all. In these, the act of pressing the <EJECT> button is considered an indication of disc replacement. If this is the case, you can use the hole for emergency disc ejection. However, bear in mind that ejecting a disc using this method on a running drive can ruin the mechanism altogether.

By the way, there are some drives that manage to read the disc even if the TOC has been completely destroyed . For instance, the list of these drives includes some MSI models. The lucky owners of these drives don t need to disassemble their devices, since they are capable of reading corrupted drives even without this operation.

When recovering multisession discs, it is possible to try to color the Lead-in disc area with black marker. The contents of the first session will be lost. All of the other sessions, however, will be read successfully by most drives. To do so, just recall that the Lead-in disc area looks like a glittering ring encircling the internal edge of the disc.

The Disc is Recognized by the Drive, but Not by the Operating System

You insert the disc into the drive. The drive starts to rotate it, and the DISK IN indicator (if there is one) goes on. However, any attempt to view the disc contents using standard OS tools results in various error messages. Scanning of the disc surface using Ahead Nero CD Speed (or any other similar utility) discovers one or two damaged sectors.

This is an obvious symptom of file-system corruption. To be more precise, this is the corruption of the root directory of the file system. If this happens, do not panic. Recovery of the CD root directory, in contrast to recovery of the root directories of hard disks and diskettes, doesn t present a serious problem. Most CDs contain not one, but two duplicate file systems ”ISO 9660 and Joliet (this is true for all discs manufactured after 1995). The simultaneous corruption of both root directories is extremely unlikely . Besides this, due to the lack of fragmentation, subdirectories are not scattered over the entire CD surface. On the contrary, they are concentrated in a single location. Thanks to this, even if the root directory is totally destroyed, they can be recovered quite easily. Finally, each next session of a multisession disc includes the contents of the file systems of all previous sessions (excluding, naturally, the deleted files). Therefore, if the file system of the last session is destroyed, we can easily recover the contents of all of the previous sessions.

Unfortunately, built-in Windows tools do not allow you to selectively mount either the preferred file system or the preferred session. Instead, they force you to use the root directory of the Joliet file system of the last session. The simplest idea that comes to mind is to try to read the disc under pure MS-DOS with the MSCDEX driver, working exclusively with ISO 9660 and ignoring the existence of Joliet. Another variant is to use the ISO 9660. dir utility developed by the author specially for working with destroyed file systems and restoring practically everything that can be restored.

Naturally, since the maximum length of the file identifiers in the ISO 9660 systems is only 11 characters , long file names become irreversibly corrupted. However, you will probably agree, this is still better than nothing.

The Computer Freezes When Inserting the Disc into the Drive

You insert the disc into the drive, the drive starts rotating it, the activity indicator blinking intensely, and then it freezes. Quite often, it also freezes the operating system. In the best cases, the situation can be resolved by pressing the <EJECT> button. In the worst case, you ll have to press <RESET>.

Such behavior is typical for protected discs, the protection of which is based on a corrupted TOC. Most drives are loyal to corrupted TOC (although that depends on what exactly has been corrupted). However, one can run into devices that simply freeze after such a situation occurs. If it is still necessary to read the protected disc, try to change the drive.

Another possible variant is a looped file system. This happens frequently when burning CD-R/CD-RW discs using incorrectly written software. If this is the case, press and hold the <Shift> button during disc loading to prevent the operating system from reading its contents (or temporarily disable the AutoRun feature). Then, using utilities like ISO 9660.dir, copy everything that can be copied from the disc.

The Disc Is Read with Errors

If, despite all your efforts at decreasing the rotation speed or cleaning the disc surface, the disc is still read with errors, and corrupted sectors fall exactly to the area taken up by the most valuable files, then things have taken the worst turn . However, there is still some chance of successful data recovery, although very slim.

First and foremost, there are all kinds of errors. Cases where the entire sector is unreadable are rare. As a rule, the case in point is that one or two bytes belonging to that sector are unreadable. At the same time, the correcting capabilities of redundant codes are such that up to 392 corrupted bytes are corrected already in the first-level decoder (CIRC-decoder). P-codes are capable of correcting up to another 86 errors, and Q-codes can correct up to 52 errors. This means that, under favorable error distribution, it will be possible to recover up to 530 errors, or up to 25 percent of the total sector capacity. Only the horrible unreliability of optical media causes situations where even this vast data redundancy is sometimes unable to withstand failures.

Depending on the setup parameters, the drive, having detected an irrecoverable error, either returns the sector in the form, in which it managed to read it, or simply reports an error, leaving the contents of the output buffer in an undefined state. The idea of data recovery consists of making the drive return anything that it is capable to read. Naturally, corrupted bytes cannot be recovered. However, many file formats are quite tolerant to minor corruption. Music in the MP3/WMA format, video films and graphic images will be successfully recovered. Only in the exact place of the error will a click ” sometimes louder than others ”be audible, or some artifact will be noticeable. The case with archives is significantly worse. However, in most cases, only a single file will be lost, while the remaining contents of the archive will be unpacked normally (by the way, some archivers, such as RAR, support their own correcting codes, which provides for the restoration of corrupted archives with the expense of minimum redundancy).

Wait! ”some readers will cry, ” This is not right! In fact, we tried to recover unreadable discs with one or another utility. And what was the effect? The system refused to consider reanimated MPG or AVI as video files! However, I would object that these utilities simply discarded all of the sectors that they could not read, and as a result, the file size and relative offsets of all its structures have changed. Consequently, it is no wonder that the file cannot be played after such a recovery.

Use any copier of protected discs that provides selective control over the error-processing mode and choose the 24h mode (the maximum possible error correction without interrupting data transmission in the case that an error encountered proves to be irrecoverable). Among all of the utilities suitable for this purpose, I recommend that you use the cd_raw_read utility developed by the author. As an alternative, it is possible to use Alcohol 120% and/or Clone CD.

What reasons can be behind the fact that the sector is unreadable? First of all, there could be deep and wide radial scratches on the upper part. After penetrating a thin barrier of protective coating, the scratches damage the reflective material directly, and with it, useful data are damaged.

Narrow scratches that are not too numerous generally are not too dangerous, since the sector contents are distributed along the spiral track. Therefore, the loss of several bytes can be easily compensated for by data redundancy. However, there is one but. How does the drive get to know how many pits and lands have been omitted? Since pits and lands do not correspond to binary zeroes and ones directly, and a binary one is encoded by the transition from pit to land and vice versa, while zero is encoded by the lack of transitions at the given point, it becomes obvious that the disappearance of an odd number of pits/lands seems to reverse the entire frame trailer. In other words, it simply ruins it. Hence, even a single scratch can generate an entire cascade of errors that cannot be recovered using standard correcting codes. However, theoretically, such errors can be repaired manually. Manually? Well, not quite, since for this purpose, you ll need a special utility. An example of such a utility has already been written by the author and is now undergoing alpha-testing. It can successfully read discs that were not readable in a standard way. I hope that by the time that this book is published it will be ready for beta-testing and become available for free for all users who need it. However, since the length of one frame is only 24 bytes, the destruction of several sequential frames can be repaired even by standard correcting codes. Therefore, my utility will be needed only for recovering badly damaged discs with lots of scratches.

Wide scratches are a different matter. Not only do they eat up several frames entirely, they also send the optical head astray from the track. When the head falls into the hole created by a scratch, it becomes totally disoriented, since it simply has nothing to rely upon. After that, the head flies out and lands on one of the neighboring tracks. Intellectual drives detect such a situation and position the head to the required place. The drives that have no such intellectual functions, which, by the way, are the vast majority, self-confidently try to continue reading as if nothing has happened . As a result, the header of one sector is combined with the trailer of another one, and, naturally, any attempt to restore such a sector using standard correcting code will produce nothing but garbage. Consequently, the drive will report an irrecoverable error. The only way out is to read such a sector until the head falls into the same track, from which the reading of the sector was begun. The number of reading attempts in this case must be large enough (100 or more). After all, it is much easier to divert from the narrow spiral track than keep on following it!

Concentric scratches represent the most destructive type of damage that can only exist on CDs. The distribution of information along the spiral track is now unable to withstand the failure, since corruption influences the entire sector (in contrast to radial scratches that damage only a small part of the sector). Besides this, concentric scratches disorient the tracking system, since sensor lasers are slightly defocused, and, therefore, are very sensitive to such surface defects.

The scratches located at the lower side of the disc can in most cases be eliminated by polishing, while scratches that involve the working layer are impossible to eliminate.