Last-modified: 1997/09/14
Version: 1.11
The CD writing process can't be interrupted in mid-session. To prevent this from happening, the makers of CD recorders put a write buffer in the drive, usually 512K or 1MB. Data read from the hard drive, tape, or another CD is stored in the buffer, and pulled out as needed by the recorder.
If the recorder requests data from the write buffer, but there's none there, it's called a buffer underrun. The disc is still spinning, but there's no data to write, so the recording process aborts.
You can still use the disc with multisession CDROM drives by closing the session and starting another, assuming there's enough space left on the CD, and assuming your pre-mastering software didn't choose to finalize the disc for you.
Advice for preventing buffer underruns is scattered throughout this FAQ. A brief summary:
Also watch out for things like anti-virus programs that wake up, virtual memory settings that cause swapping, screen savers that activate during the CD creation process, unusual network activity, and background downloads of data or faxes. One way to check is to run the HD defragmenter in Win95. If it restarts every few seconds, it's because something is hitting the drive.
Most Win95-based software recommends turning off Auto Insert Notification. Having this on can interfere with closing sessions or even just inserting discs into the drive. You can disable it by opening the "System" icon in the Control Panel, and selecting "Device Manager". For each item under CDROM, select the device, click on the "Settings" tab, and make sure the "Auto Insert Notification" checkbox is unchecked. [With my setup I get SCSI errors when AIN is off for my CD-R but on for my CDROM, even if the CDROM drive isn't in use at the time.]
If you're using WinNT, you can turn it off with the "TweakUI" program available in PowerToys, or by modifying a registry key with Regedit32 (0=disabled, 1=enabled):
HKEY_LOCAL_MACHINE \ SYSTEM \ CurrentControlSet \ Services \ Cdrom \ Autorun
Another helpful tip for Win95 is to limit the size of the file cache, which by default is unrestricted. The procedure is simple:
[vcache]".
[vcache]" line:
minfilecache = 512
maxfilecache = 4096
http://www.adaptec.com/support/cdrec/configcdr.html is interesting reading for users with CD-Rs attached to Adaptec SCSI cards. They're pretty far on the conservative side, but if you're having trouble this may help you.
An article by Dana Parker entitled "CD-R on the Safe Side: Seven Rules of
Successful CD Recording" in the April 1997 issue of Emedia Professional
(http://www.onlineinc.com/emedia/EMtocs/emtocapr.html) listed the Seven
Habits of Successful CD-R Users:
It appears that Win95 and DOS 7.x are simply screwed up.
(This problem was reported with an HP4020i and a Buslogic BT946C controller; if you have an HP drive you should get the c4324hlp.vxd driver from the HP web site. See section (6) for the address.)
One possible cause of this problem is writing a multisession disc in MODE-1 format. Some older CDROM drives incorrectly assume that a MODE-1 disc can't be multisession, so they don't look for additional sessions unless it's written in MODE-2 (CD-ROM XA) format.
Also, if the final session on the CD isn't closed, standard CD players may become confused (the NEC 6Xi certainly does under Win95). This doesn't mean that the *disc* must be closed, just that the *session* must be closed. (Actually, the NEC 6Xi doesn't like open discs either... sigh.)
If that fails, a number of people have discovered that the problem is a hosed CD-R unit (similar behavior has been reported on Sony and HP units, which have different mechanisms). You should try 1x writing from a fast source and with different sets of data before contacting the manufacturer, since they will likely tell you to do exactly that anyway.
Be sure that there aren't environmental factors creating difficulties. CD-R units are usually built to handle small shocks, but having a set of speakers playing loud music on the same table as a CD-R may cause it to skip, resulting in a failed write. Sonic booms, heavy construction equipment, and nuclear detonations may have similar effects.
It's also possible that you simply have a bad batch of media. Try a different type and brand of disc. Some distributors (e.g. dataDisc) will exchange media that's provably defective.
Some CD-R packages allow you to start the real write pass a few seconds after the test pass has succeeded. They may not disable this even though they know that the disc will have to be ejected. Make sure the option is set to "wait until told to continue."
If it fails with both green and gold media, chances are your drive simply doesn't work with CD-R discs. This is rare but not unheard of. In one case, returning the drive for an identical unit solved the problem.
While there are stringent specifications for CDs, there are no such specifications for CD audio and CDROM devices. They just have to play the discs. If the disc and the drive are both marginal, you lose.
(For the Mac, look at "ISO 9660 File Access" in the System:Extensions folder with Command-I. If the version shown is 5.0 or greater, your system should handle the version numbers just fine. If not, you should update your system software.)
"mkisofs" has an option to omit the version number when constructing an ISO-9660 image.
Another user with the same setup found that doing power-up diagnostics and device reset right before burning the CD helped.
The ways of dealing with this range from minor system changes to the placement of chicken entrails on selected components. Reducing the DMA rate on the AdvanSys SCSI card may help, buying better SCSI cables and checking for proper termination may make a difference, or even powering off and on again right before the burn.
One user was told by Philips tech support that if error 50h (write append) occurs, it means the drive has to be returned to the repair center. Other users have been told that the error can occur when attempting to write an empty directory or zero-length file. Under Easy-CD Pro '95, this is reported as error 171-00-50-00 (see the Adaptec web site for a complete list of error codes).
If the fault is caused by the worn spring, it may be possible to fix the problem by replacing the spring. This will definitely void your warranty, and you shouldn't even think about trying this unless the only alternative is to throw the drive away. Jonathan Oei posted some details about the process (search for comp.publish.cdrom.hardware, subject "CDD2000 & Spring Fix", on http://www.dejanews.com/). This procedure requires some special tools (mini torx drivers and really fine jeweller's pliers), and involves disassembling much of the drive. If you open up the drive and remove the circuit boards, you will see that the laser writing assembly is moved by a DC stepper motor. The motor has a plastic drive gear that is meshed with a plastic "rack" on the laser. The spring in question is a piece of wire that pushes the rack against the drive gear, so when it weakens the gear slips and the write fails. Replacing the 0.012" wire with a 0.02" diameter wire solves the problem.
This question is also covered in the HP4020i FAQ, available at http://www.cd-info.com/CDIC/Technology/CD-R/HP-FAQ.html.
It may be that the system is looking at the disc, not finding a TOC (table of contents), and ejecting it as useless. You need to disable this.
Under Win95, disable auto insertion for all CDROM devices (see section (4-1)). One user found that reinstalling Win95 helped. On the Mac, you may just need more recent drivers.
If that doesn't work, make sure the CD-R drive is perfectly level. Apparently some units are sensitive to being tilted at an angle. Some users have had trouble when a CD-R has been on for a while and has overheated, so if you only have trouble when the machine has been powered on for a while, try putting a small fan above the unit to blow air over it.
If nothing helps, there's a strong possibility that the drive is mis- aligned and needs to be serviced. This has been known to happen to drives during shipping.
One user reported problems when using the wrong type of caddy. It has to be a Sony-type caddy, which is the kind most commonly found in stores.
The error indicates that the CD-R drive is having trouble calibrating the power setting. Try different brands of media, and if that doesn't work you should have the drive checked.
Some versions of the firmware for the Philips CDD2000 and HP 4020i will report a power calibration error if you try to do a 1x write after a 4x read.
Out-of-date versions of MSCDEX have been known to "forget" certain files when browsing a disc. If you're using DOS or are using the "real mode" drivers from within Win95, make sure you're using the most recent version of MSCDEX.
Old versions of certain CD creation programs would occasionally omit things when asked to burn a large number of files. These problems haven't been reported for some time, however.
If you were burning a multi-session CD, you're on your own.
If you're using Easy-CD Pro for Win31 or Corel CD Creator, you can load the contents of all the previous sessions, and burn a new session that has all the files you want. This feature isn't currently available in Easy-CD Pro 95, which only allows you to link to one previous session.
The files themselves aren't lost forever though: most packages will allow you to extract a track as an ISO-9660 image, and you can use WinImage to pull individual files out of it.
One caution: without something like Corel's Session Selector, you may not see the last session on the disc anyway. Some CDROM drives stop looking for sessions after a certain point.
Bertel Schmitt wrote an excellent article on the ins and outs of proper cabling and termination. The article can be found in text form at ftp://ftp.netcom.com/pub/fa/fadden/scsi-trm.txt. Granite Digital, a company that makes high-quality cables and terminators, can be found at http://www.scsipro.com/.
If you're using an HP 4020i with the AdvanSys SCSI card, reducing the DMA transfer rate may help.
Another thing to try is to do multiple large reads from a CD-ROM disc, and compare the results of each. If the SCSI cable is flaky or the termination isn't done correctly, the problem might affect reads as well as writes.
If a file appears to be getting corrupted on the CD-R, try copying it back to the hard drive and then comparing it to the original. If possible, see if the file is missing large chunks or just has sporadic damage throughout.
It's possible, though less likely, that it's a defective CD recorder.
The most fundamental problem is that, if the CD is dirty, the error correction may not be able to correct all of the errors. Some drives will interpolate the missing samples, some won't.
Another problem most CD-ROM drives face is "jitter". See section (2-15) for details.
See also section section (3-3) on avoiding clicks in extracted audio, and section (5-5) on which CD-ROM drives are recommended.
Suppose you extract the audio track from the copy, and it's an exact binary match of the track you wrote from your hard drive, but the CDs don't sound quite the same. What then?
Most people don't notice any difference between originals and duplicates. Some people notice subtle differences, some people notice huge differences. While it's true that "bits are bits", there *are* reasons why CD-Rs may sound different even when the data matches exactly.
The manual for the CDD-2000 reportedly states that the drive uses 4x oversampling when playing back pressed CDs, but switches to 1x for CD-R. This affects the quality of the D/A conversion, and can make an audible difference.
It has been suggested that the D/A conversion process in the CD player is more susceptible to "jitter" when reading CD-Rs, because the clocking of the bits isn't as precise. A quality CD player will compensate for this automatically. (Note: this kind of jitter is different from the DAE kind of jitter described in section (2-15).)
Others have asserted that *any* two CDs, pressed or otherwise, will sound slightly different.
Some people believe that audio CDs should be recorded at 1x, while others have asserted that, for various technical reasons, 2x is better.
An extremely technical introduction to CD reading is available at http://www.tc.umn.edu/nlhome/g496/eric0139/Papers/paper.html. This may shed some light on why reading audio CDs is difficult.
If you are finding noticeable differences, try different media, a different player, and a different recorder. There is some evidence that different brands of media and recorders may work better for audio, but in the end it's a highly subjective matter.
CDs that start at 00:02:32 (0 minutes, 2 seconds, and 32 frames) are surprisingly common. The problem can be worked around manually, by looking at the output of the Jeff Arnold's (freeware) TOC program (available from http://www.goldenhawk.com/), and supplying "/start=" and "/end=" parameters that adjust backward by the number of frames in excess of two seconds.
For example, if the first track started at 00:02:32, you would subtract 32 from the starting and ending Logical Block Addresses.
A better solution is to use a CDROM drive that doesn't have this problem (and most likely can extract audio more quickly than the CD-R can).
The Yamaha CDR-100/102 and the Philips CDD2600 are known to have this problem, though it may get fixed by a firmware update. The Ricoh 6200S reportedly does not return the disc's table of contents correctly for these sorts of discs.
Home | Applications | Bibliography | CD-Lite
History | Industry | Sponsors | Technology
FAQ Copyright 1997 Andy McFadden
HTML Copyright 1997 The CD-Info Company, Inc.
970819