oreilly.com: Coaster-Free Burning with IDE CD Writers: Part 3

Published on Windows DevCenter (http://www.windowsdevcenter.com/)
See this if you're having trouble printing code examples

Coaster-Free Burning with IDE CD Writers: Part 3

by Robert Bruce Thompson, Barbara Fritchman Thompson
10/30/2000

Part 1 and Part 2 of this article examine choosing the best hardware, software, and discs for dependable CD burning and configuring your system for maximum reliability. This concluding section describes the steps you should take before and during a burning session to minimize the chance of coasters.

Disable Nonessential Functions

If your burns usually succeed, but occasionally fail for no apparent reason, a background process is a likely culprit. This is especially true if you run Windows 9x rather than Windows NT/2000, and if your computer is relatively slow or has limited memory.

As a matter of good general practice, disable all nonessential background processes and functions before you burn a CD. Turn off your screensaver and disable power-management features. If you share your disk or printer with other users on a network, disable sharing while you burn the CD. Make sure that RAS/DUN, Internet Connection Sharing, fax servers, and similar background processes running on the local machine are disabled while the CD is being written.

In particular, two applications are notorious for interfering with burns:

Find Fast. A Microsoft indexing utility that installs by default with Microsoft Office, Find Fast periodically indexes the contents of your hard disk. If Find Fast indexing kicks in while you're burning, a coaster is likely. Disable Find Fast before you burn. To do so, display Control Panel. If the Find Fast applet is visible, Find Fast is installed on your system and is probably active. Double-click Find Fast and choose Index -> Pause Indexing to disable Find Fast for the duration of your burning session.
Antivirus software. Sometimes antivirus software causes failed burns. If your software initiates a scheduled scan during the burn, a coaster is likely. But even disabling scheduled scans is no guarantee of success. Antivirus software often runs as a background process, watching all system activity for "virus-like" behavior. Such background activity may sometimes be sufficient to cause a failed burn. If your antivirus software permits doing so, disable both scheduled scans and background checking for the duration of your burning session.

Use Image Files for Reliable Burns

Broadly speaking, there are three ways to burn a CD, whether the source data is another CD or a random collection of files on your hard disk:

Burning on the fly. With this method, data is streamed from the source CD or hard drive, formatting and error-correction data is added in real time, and the resulting data stream is burned to the CD. The advantages of on-the-fly burning are that it is faster than other methods and that it requires no extra disk space. The drawback is that on-the-fly burning is the method most likely to create coasters. Most newer systems are fast enough to successfully dupe audio or data CDs on the fly, but you may have problems if you attempt to write hundreds or thousands of relatively small files to a CD, when, for example, you use your CD writer to back up your hard disk.
Burning true image files. This method uses a two-step process. Data to be written to the CD is first read and processed to add formatting and error-correction data. That formatted data is then written out to the hard disk as an ISO image file, which is an exact binary representation of the data as it will be written to the CD. The drawbacks to using true image files are that it takes longer and that you must have enough free disk space to accommodate the image file, which can be 800MB or more, when you are copying audio data to an 80-minute blank. Despite these disadvantages, burning a true image file is by far the most reliable method.
Burning virtual image files. This method is similar to using true image files, with the exception that an actual image file is not written to the hard disk. Instead, a virtual image file is created, which contains pointers to the locations of the files to be written to the CD. Because formatting, adding error correction, and all other preprocessing is completed before the actual burn starts, using a virtual image file is more reliable than burning on the fly. Conversely, because the files to be written must be retrieved from random locations on the hard disk during the burn, using a virtual image file is less reliable than using a true image file. Using virtual image files is slower than burning on the fly, but faster than using true image files.

The best method to use depends on the capabilities of your system, your CD writer, and your software as well as the type of data you want to burn to CD. On-the-fly burns usually work well for duping audio or data CDs, and (assuming that you have enough free disk space) using a true image file is best for doing backups and similar operations that require writing many small files to disc. As always, the best way to judge is to try each method and use the one that works fast and reliably for you.

Note: If you are building a new system or installing a new hard disk on a CD-writer system, we recommend creating a dedicated hard-disk partition to be used as a "staging area" for ISO image files. This partition needn't be large--a gigabyte or so is sufficient--but it should be on your fastest hard disk if there's a choice (and certainly on a hard disk that is on a different channel than the CD writer). Read Part 2: Configuring for Reliability for more on using different channels.
Configure your burning software to write the ISO image file to the dedicated partition. After you complete each burn, you can delete the ISO image file or move it elsewhere, freeing up the partition for the next burn. We've never had a problem just deleting the old ISO image file, but some belt-and-suspenders folks we know do a Quick Format of the partition each time to ensure the ISO image file is written sequentially to the partition.

Defrag your Hard Drive

If you burn CDs from an image file on your hard drive--as you should for the most reliable burns--defrag that hard disk regularly unless you have created a dedicated partition for ISO image files. Except in the worst cases, it's not necessary to defrag your hard disk before every burning session, but defragging every week or so will help ensure reliable image burns. The goal is to have a free space area at least as large as the ISO image you will write to the hard disk.

Windows 9x and Windows 2000 both include adequate defragging utilities. For either, run the defragmenter by choosing Start -> Programs -> Accessories -> System Tools -> Disk Defragmenter. Although Windows NT 4 does not include a defragmenter, Diskeeper Lite for Windows NT is a free download from Executive Software, and does the job well.

Figure 1. Using Diskeeper Lite to defrag an NT4 volume

Don't Bother Doing Test Burns

A test burn performs the entire CD-writing process, but without actually writing the CD. In theory, a test burn lets you do a rehearsal, without wasting a blank disc if something turns out to be wrong with the setup. The whole idea of test burns is left over from the 'Bad Old Days' when a blank disc cost $25 or $50. At that price, taking pains to avoid burning coasters was worthwhile, so everyone did test burns.

The problem with test burns, though, is that they aren't a very good predictor of success during an actual burn. Oh, if a test burn fails, you can be pretty certain that a real burn would also have failed, but if the test burn succeeds, that's no guarantee that the real burn will also succeed. It is quite possible to do a successful test burn immediately followed by a failed actual burn, because the test burn doesn't test the actual writing process, which is by far the touchiest aspect of all. A successful test followed by a failed burn is particularly common on systems with marginal power supplies, which are adequate during the testing phase, but cannot provide sufficient power when the CD-writer laser is running at full burning power.

A test burn is now just a good way to waste time. All CD-writer software advocates doing test burns, but we suggest you don't bother. The only real way to test is to burn a CD. If it succeeds, you can get on with your life. If it fails, you're out a buck or so, but you've been given some useful information.

Burn at the Optimum Speed for Your Drive and Media

Don't assume that you can use your CD writer's fastest speed, even if your burning software tests a disc and claims that it is writable at the highest speed. In general, burning at higher speeds is less reliable than burning at lower speeds, both because faster burning is more likely to generate errors while writing, particularly with marginal discs, and because the CD writer's buffer, whatever its size, empties faster at higher burning speeds. For example, when writing at 12X (1,800KBs), a 512KB buffer stores only about a quarter of a second's worth of data. Any interruption in the data stream longer than that generates a coaster (unless, of course, your drive supports BURN-proof). Larger buffers and lower write speeds minimize the chance of that happening.

However, "slow equals reliable" is by no means a universal truth. Burning at a slower speed is not always more reliable. For example, we have one 8X CD writer that writes most discs reliably at 8X, some discs reliably at 4X, and very few discs reliably at 2X or 1X. The optimal burning speed depends on numerous factors, particularly the combination of drive, firmware revision, and disc.

In general, when we start with a new batch of media on a given CD writer, we first attempt burns at the highest-rated speed of the drive, regardless of the speed for which the media is certified. For example, we have a spindle of no-name 4X certified media that works without problems at 8X on all three of our Plextor 8X burners, both SCSI and IDE. Those same discs, however, generate about 10 percent coasters in our Smart and Friendly 4X burner, when burned at 4X or 1X, but nearly 50 percent coasters when burned at 2X. Go figure.

Also note that the type of data you are burning and your staging method can make a difference. For example, we have some no-name 8X certified discs that work fine for on-the-fly copies of data or audio CDs at 8X. But when we attempt to use those discs to do an 8X on-the-fly backup of our hard disk, coasters generally result. If we instead use a true image file as the source, the backup runs fine at 8X to those discs. Conversely, if we back up to Kodak 8X certified discs, both the on-the-fly and the ISO image backup work reliably. It shouldn't make a difference--8X is 8X--but it does.

As always, the best solution is to test in your own environment.

Conclusion

In these articles, we've explained what our experience tells us are the most important factors for burning CDs reliably with an IDE CD writer. If you follow our advice, there's no guarantee you'll burn perfect CDs every time, but your success rate should approach 100 percent. And that, after all, is what it's all about.

Robert Bruce Thompson is coauthor of PC Hardware in a Nutshell. He built his first computer in 1976 from discrete chips. It had 256 bytes (not kilobytes) of memory, used toggle switches and LEDs for I/O, ran at less than 1MHz, and had no operating system. Since then, Robert has bought, built, upgraded, and repaired hundreds of PCs for himself, employers, customers, friends, and clients. He is the author or coauthor of many online training courses and computer books. Robert maintains a personal online daily journal page, as well as a Web site devoted to PC Hardware in a Nutshell.

Barbara Fritchman Thompson is the coauthor of PC Hardware in a Nutshell. She worked for twenty years as a librarian before starting her own home-based consulting practice, Research for Authors. Barbara, who has been a PC power user for fifteen years, researched and tested much of the hardware reviewed for the book. Barbara spends her working hours doing research for authors and her leisure hours reading, working out, and playing golf.

Robert and Barbara Thompson's Web site is HardwareGuys.com.

O'Reilly & Associates recently released (October 2000) PC Hardware in a Nutshell.

Sample Chapter 25, Designing a PC, is available free online.
You can also look at the Table of Contents, the Index, and the Full Description of the book.
For more information, or to order the book, click here.