Reviews Leaderboard Database Reference Search StorageReview Discussion Reliability Survey Search About Contents

ATAPI CD-ROM Drive Roundup

  May 30, 2000 Author: Tim Zakharov  
Special thanks to Thortek for providing evaluation units.

Part 1: ATAPI CD-ROM Drives


In preparing for's first-ever CD-ROM Roundup article, we obtained nine of the fastest currently available CD-ROM drives for evaluation, five ATAPI and four SCSI. We debated the merits of including all in the same article, or splitting the ATAPI and SCSI models, as we do with hard drives. In the end, we chose to keep the two interfaces separate. As is the case with hard drives, SCSI CD-ROM drives are largely intended for a different customer-base than ATAPI models. They are generally priced higher than their ATAPI counterparts and are more often used in a business/office environment where connectivity and reputation for reliability are more desired. ATAPI units are generally much less expensive and are more often used in name-brand systems as well as by typical home computer users. It was a toss-up as to which we'd publish first. Since the ATAPI units are generally more popular with the masses, we will start with them and follow with a SCSI Roundup article shortly.

As many of you may already know, optical drives generally come in two flavors of spindle motor spin rate: Constant Angular Velocity (CAV), and Constant Linear Velocity (CLV). In a nutshell, CAV drives spin at a constant rate. Because the inner tracks of a CD are much smaller than the outer tracks, a CAV drive will have slower read rates at the inside of a CD than at the outer edge. CLV drives, however, vary the spindle motor's RPM according to where on the CD the information is being read. The motor will speed up at the inner tracks and slow down at the outer tracks. This creates a linear read rate across the CD. Most of today's CD-ROM drives are CAV. However, under certain conditions a CAV reader may operate in a CLV-fashion. For example, many drives will read CD-RW media or extract digital audio at CLV. This could be for compatibility, error-correction, or performance reasons. We will specify when CAV units revert to CLV. Finally, it should be mentioned that there is what is called P-CAV. This is partial constant angular velocity. This is when a drive starts reading a CD in a CAV-fashion at the inner tracks, then when it reaches a specified speed, reverts to CLV for the remainder of the CD.

Since the days of 1X, CD-ROM drives have been capable of playing audio CD's. However, in order for the audio tracks to be heard (ignoring any front headphone jacks ), one must use an audio cable to connect the drive to the sound card. As sound card technology has improved, analog audio has begun to share the stage with digital audio. Sound cards such as the SoundBlaster Live! now have inputs for digital as well as analog CD audio. To those who own such sound cards, it is important information to know which optical drives support digital audio connections.

In order to save article space and avoid boring many of our readers, we will link each drive's model number below to the manufacturer's spec sheet, rather than regurgitate pages of information here. Occasionally (as in the Creative unit below) a drive manufacturer will have no info on a tested drive on their site. We apologize for not being able to provide links to spec sheets for these units.

Now, on to the contenders!

The Drives

Creative Labs CD5230E - Firmware revision 1.01:
Creative Labs' entry into our roundup is their top-of-the-line 52X CAV unit, sporting an 80ms access time, 128kb buffer, and a standard 1-year warranty. A bit of an enigma, the 52X unit is not even listed on Creative Labs' website (48X is the highest speed drive they list), hence we cannot link you to any specifications. Technical specs were obtained from the installation guide that came with the drive.

Although one of our speediest performers overall, the CD5230E was also one of the loudest. At full-RPM, the drive was clearly audible not only over all other testbed system sounds, but over the sounds of my personal system next to it as well as an audio CD playing at moderate levels. You'll need to crank up your PC speakers to completely drown out the sounds this drive makes when it's in high gear. Fortunately its heat levels do not come near matching its sound levels. The drive does become a bit warm to the touch after continuous extended use, but well within normal ranges. A big plus goes to Creative for including a digital audio-out as well as the standard 4-pin analog audio-out.

ESP: $65

Delta OIP-CD4800A - Firmware revision 3.30:
A Taiwan-based company, Delta Electronics was founded in 1971 and produces primarily in the OEM segment. Their second-fastest current CD-ROM drive at 48X (they also have a 52X unit we were unable to attain), this CAV CD-ROM reader is the least expensive of this roundup, while turning out some of the highest levels of performance. Its access time is rated at 85ms, with a 128kb buffer and a standard 1-year warranty.

In its favor the Delta supports digital audio-out as well as the standard 4-pin analog out. Unfortunately, however, it takes after the Creative unit in sound levels, matching it at full-RPM, whine-for-whine. Heat levels were moderate and entirely acceptable.

ESP: $40

Kenwood UCR-421 - Firmware revision 217G:
Kenwood Corporation's fastest True-X drive, the 72X uses Zen Research's now legendary 7-beam pickup to split the read laser into multiple beams, allowing the drive to read 7 tracks in parallel for somewhat P-CAV 72X performance across the majority of the disc. An added benefit of this read method is a drastically reduced RPM, leading to much quieter operation. Kenwood specifies an access time of <100ms (which our drive could not duplicate), with a beefy 2048kb buffer and a standard 1-year warranty.

Despite the Kenwood's quiet levels, there was a noticeable shaking during operation, which was isolated to random accesses. This caused the entire testbed case to shake (not quite visibly, but definitely noticeable by placing your hand anywhere on the system case) during drive operation. Any drive defects were ruled out when I noticed the same phenomenon in Kenwood's 52X SCSI unit (reviewed in our upcoming SCSI roundup). Considering this occurs during random access testing but not sequential transfer testing, this seems to be somehow related to how smoothly their head actuator mechanism tracks along the rails as it changes directions when seeking.

On the heat side of things, the 72X can get very warm during extended use. In fact, upon removing the drive from the testbed after a random access stress test, I found the underside to actually be hot to the touch in certain spots (the top of the drive never approached such hot levels). Because CD-ROM drives are seldom, if ever stressed as extensively and consistently as the access time loop I run the drives through when checking for noise and heat levels, it is highly unlikely that heat would ever be a concern during normal usage.

While the 72X has what looks like a digital audio-out next to the standard analog out, it is not functional. Kenwood does not advertise digital audio-out, so no harm, no foul.

ESP: $125

Mitsumi CRMC-FX4820T - Firmware revision D03A:
From Mitsumi Corporation comes the CRMC-FX4820T, currently their fastest CD-ROM drive at 48X CAV. Featuring an access time of 75ms, a 128kb buffer and a standard 1-year warranty, the Mitsumi appears on the surface to be a contender. Unfortunately, we encountered serious problems with our test unit, to the best of our knowledge isolated to the drive's head assembly. Any test we threw at the unit that involved random accesses caused the drive to either fail completely or perform extremely poorly. However, all tests involving sequential transfers yielded very competitive performance. Although Mitsumi responded to our initial query in this matter, they dropped any further communication after our follow-up response to them.

Like the Creative and Delta units, the Mitsumi's sound levels during full-RPM were extremely high. It was the coolest-running unit of the roundup, however, staying cool to the touch even during extended use. Finally, it should be noted that the Mitsumi has a functional digital audio-out connector.

ESP: $53

Toshiba XM-6702B - Firmware revision 1005:
Longtime CD-ROM manufacturer Toshiba enters the fray with their CAV 48X drive. The 6702B's access time is specified at a speedy 78ms, with a 128kb buffer and a standard 1-year warranty.

An interesting contrast to the other 48X+ CAV units, the Toshiba's noise levels were relatively quiet. Only the Kenwood was stealthier. Also impressive were its heat levels. However, like the Kenwood, the Toshiba had what looked like a digital audio-out connector that did not work. Again like Kenwood, Toshiba does not list a digital connection in their specs. In my personal experience with Toshiba's DVD-ROM drives, however, the digital connections worked properly despite not being mentioned in the spec sheets. This leads me to wonder if the 6702B has a digital audio-out that is defective, or is purposely non-functional. Either way, Toshiba is safe from scrutiny thanks to some prudent spec-sheet writing.

ESP: $59

For an overview on methodology, click here.

Performance Results

Low-Level Measurements

 Optical Testbed I Low-Level MeasurementsDetails... 
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 7520|
Creative CD5230E (ATA-33) - 3693|
Delta OIP-CD4800A (ATA-33) - 3577|
Toshiba XM-6702B (ATA-33) - 3480|
Mitsumi CRMC-FX4820T (ATA-33) - 3420|
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 11267|
Creative CD5230E (ATA-33) - 7700|
Delta OIP-CD4800A (ATA-33) - 7433|
Toshiba XM-6702B (ATA-33) - 7150|
Mitsumi CRMC-FX4820T (ATA-33) - 7037|
Windows 98 SE using FAT 32
Toshiba XM-6702B (ATA-33) - 74.3|
Creative CD5230E (ATA-33) - 76.0|
Delta OIP-CD4800A (ATA-33) - 84.3|
Kenwood UCR-421 (ATA-33) - 107.7|

* Although we measure interface burst speed for all optical drives, the usefulness beyond confirming manufacturer specs and interface functionality is questionable at best. Given the start and stop nature of optical drives, chances are slim to none that one would ever find a situation where the maximum burst speed is realized.

 Optical Testbed I CD Tach 98 - Burst Speed 16k *Details... 
Windows 98 SE using FAT 32
Toshiba XM-6702B (ATA-33) - 15905|
Mitsumi CRMC-FX4820T (ATA-33) - 15709|
Delta OIP-CD4800A (ATA-33) - 15631|
Creative CD5230E (ATA-33) - 11670|
Kenwood UCR-421 (ATA-33) - 10006|

When it comes to transfer rates, Kenwood's 72X monster rules the roost! Looking at its TR graph, we see a fuzziness that the others don't share-most likely related to the split-beam technology reading 7 tracks in parallel. Notice the P-CAV curvature as the Kenwood reads the first 1/3 of the disc at a variable rate of 49X-72X, then stays right around 72X-75X for the remaining 2/3. Compared to the Creative, Kenwood is twice as fast at the inner tracks and over 46% faster at the outer tracks. The Creative 52X is rightly faster than any of the 48X units, while among the 48X units; the least expensive drive in the roundup is also the fastest. Delta over-achieves at the outer edge, finishing nearly 2X faster than its rated 48X.

Looking at the access times, the Toshiba swings ahead with a blistering 74.3ms measured. That's nearly 4ms faster than its rated speed! Creative is a close 2nd at 76ms (also 4ms better than its rating), while the Delta barely ducks under its rated 85ms access time. Kenwood, specifying "<100ms" access times, strolls in 8ms later than expected. Finally, the Mitsumi, as noted earlier, could not complete our access time test without repeatedly erroring out, hence receiving no score.

How do these results relate to CD-ROM Winmark scores? You may be surprised!

CD-ROM Winmark 99

 Optical Testbed I CD WinBench 99Details... 
Windows 98 SE using FAT 32
Toshiba XM-6702B (ATA-33) - 1775|
Creative CD5230E (ATA-33) - 1658|
Delta OIP-CD4800A (ATA-33) - 1615|
Kenwood UCR-421 (ATA-33) - 1503|
Mitsumi CRMC-FX4820T (ATA-33) - 879|

In comparing the Winmark results with the access time scores, it becomes immediately apparent that low access times are very important to high Winmark scores. The Toshiba, with its 74.3ms access time, manages to best the Creative unit by 7%, despite giving up 6.1-7.6% in transfer rates to the 52X. The Delta, despite outracing the Toshiba in transfer rates, falls nearly 9% back of the Toshiba in Winmarks because of its 10ms access time disadvantage. And finally, the mighty Kenwood, while destroying the competition in transfer rates, gives up 33.4ms to the Toshiba in access times and pays for it in the Winmarks by dropping into 4th place, over 15% back of the leader. Predictably, the Mitsumi unit, which failed the very important access time test, finished dismally in the Winmark test, with a score less than half that of the Toshiba.

Is access time really so important to real-world CD-ROM performance? Consider for a moment how you use your CD-ROM drive. Do you typically select the maximum hard disk install for your applications, or do you select the minimum install and run your apps mostly from CD? The answer to this question may help you decide how important access times are to your computing experience. The more you utilize your CD-ROM drive as a main source to your applications (perhaps most importantly in database situations), the more important low access times become for you. If you use your CD-ROM drive mostly to install/copy the entire app to hard drive, you may benefit more from the brute transfer rates the Kenwood has to offer.

If you're wondering how big of an effect these "brute transfer rates" have in file and disc copies, wonder no longer! Read on to find out how each of these drives did in our file and disc copy tests.

File and Disc Copy

 Optical Testbed I CopyingDetails... 
File Copy
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 02:02.1|
Creative CD5230E (ATA-33) - 02:13.9|
Delta OIP-CD4800A (ATA-33) - 02:14.5|
Mitsumi CRMC-FX4820T (ATA-33) - 02:16.7|
Toshiba XM-6702B (ATA-33) - 02:41.4|
Disc Copy
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 02:08.8|
Toshiba XM-6702B (ATA-33) - 02:42.3|
Creative CD5230E (ATA-33) - 02:44.4|
Delta OIP-CD4800A (ATA-33) - 02:49.6|
Mitsumi CRMC-FX4820T (ATA-33) - 05:24.6|

In our file copy test we time how long it takes to copy a single 634MB file from the CDTach98 CD to the test partition of our testbed's hard drive. Being a single file of such large size, it makes sense that sequential transfer rates would take precedence over random accesses. Indeed, looking at the results of our five contenders, they are all ordered by their STR's except for the Toshiba (which should have been nestled between the Delta and Mitsumi). During the test I noticed the Toshiba having trouble with the outer edge of the CDTach disc, with audible fluctuations in spindle motor spin rates. This situation replayed itself for all three test runs, with only a 2.36% deviation in test scores. Whatever caused the slowdown, it was consistent between all runs. Perhaps the Toshiba is not as accurate as the other drives at decoding info at such high RPMs on a less-then-perfect disc. At any rate, the Kenwood blew by the competition with an average time of 2:02.1. That comes out to about 35X, less than half of the drive's rated speed. However, it's still nearly 10% faster than the Creative 52X unit. Interestingly, the Mitsumi performs respectably in this test. Given the lack of any random accesses in a single file copy, it makes sense that the Mitsumi, which performed well in our low-level STR tests, would put up good figures here.

An interesting contrast to our file copy test, the disc copy test introduces some random access to the mix, with 3000+ files nested in a multitude of folders. Looking at the results, the Kenwood again sits on top, blowing away the next closest drive, the Toshiba, by 26%. So much for the importance of random access! However, taking a step back, the disc copy test is still mostly sequential in nature. After all, CD-ROMs do not become fragmented like hard drives do, and always read from inner to outer tracks. But the fact that there are multiple files in multiple folders does still throw a bit of a random access wrench into the spokes: Every drive completed the disc copy slower than the file copy, despite 15 fewer megabytes to copy. This points to the disc copy, with its multiple files, being a bit more of a challenge to read then the single file copy. Excepting the Kenwood, the remaining four drives all place in the same order they placed in the random access tests-yet more evidence that random accesses still hold some weight in this test. The final proof is the abysmal showing by the Mitsumi. Where it breezed through the file copy test at about the same clip as the other 48X CAV units, once even a bit of random access was thrown into the mix, it fell flat on its face, finishing in roughly double the time as the other 48X drives. Something is definitely not right with the Mitsumi. It is unfortunate that Mitsumi's tech support chose to end the dialog with us before the issue was resolved.

Now that we have finished looking at traditional CD-ROM performance figures, let's switch tracks and take a look at a comparatively new area of CD-ROM performance: digital audio extraction.

Digital Audio Extraction

 Optical Testbed I CD Speed 99 - DAE Transfer RateDetails... 
DAE Transfer Rate - Average
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 46.25|
Toshiba XM-6702B (ATA-33) - 11.11|
Delta OIP-CD4800A (ATA-33) - 8.03|
Creative CD5230E (ATA-33) - 7.94|
Mitsumi CRMC-FX4820T (ATA-33) - 7.84|
DAE Transfer Rate - Start
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 30.75|
Delta OIP-CD4800A (ATA-33) - 8.01|
Creative CD5230E (ATA-33) - 7.85|
Toshiba XM-6702B (ATA-33) - 6.80|
Mitsumi CRMC-FX4820T (ATA-33) - 4.80|
DAE Transfer Rate - End
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 49.99|
Toshiba XM-6702B (ATA-33) - 14.52|
Mitsumi CRMC-FX4820T (ATA-33) - 10.25|
Delta OIP-CD4800A (ATA-33) - 8.03|
Creative CD5230E (ATA-33) - 7.97|

Along with the advent of CD burners comes a need to extract the audio from our music CD's in order to get it over to the CD-R media. As write speeds increase in CD-R/W drives, so does the need for faster audio extraction. Although many recommend extracting to hard drive, then burning from HD to CD-R (arguments for this range from more accurate audio duplication to fewer failed burns, or "coasters"), lots of folks love the quickness with which a CD-DA can be duplicated when going directly from CD-ROM to burner. However, in order for this method to succeed, the audio tracks must be extracted at a speed consistently equal to or greater than the burner's write speed. For most of today's burners that means 8X (CD-RW's can currently be written at only 4X, but it is unlikely that one would duplicate a CD-DA to CD-RW). Therefore, in our opinion, in order for a drive to have acceptable DAE capabilities, it must extract audio at a minimum of 8X, if not higher (assuming it will be used for this purpose). Anything less would be likely to contribute to failed burns unless you purposely lowered your burner's write speeds or imaged your CD-DA to hard disk before writing it to CD-R. For this reason we will focus on DAE speeds at the inner tracks, where it will be slowest. We have included links to DAE transfer rate graphs so you can see how each drive extracts as the process moves toward the outer tracks.

Looking at the results, Kenwood's 72X again stands out above the other drives. With a minimum extraction speed of about 30X from our test CD-DA, Kenwood's unit nearly breaks 50X by the time it gets to the outer tracks. Delta, the next closest competitor, extracts digital audio at a CLV of 8X, barely acceptable. Squeezing in behind the Delta is the Creative drive, also at a CLV of about 8X. Next is the Toshiba, which extracts in a CAV fashion; starting at a less-than-impressive 6.8X, but finishing at a more respectable 14.5X. This means the first few songs on a CD-DA will not extract at acceptable speeds, but that the rest of the tracks likely will. Finally, Mitsumi's offering brings up the rear with a minimum extraction rate of only 4.8X. At least the extraction process improves, as you can see by the CAV transfer rate graph. By the outer tracks, the Mitsumi has broken the 10X barrier, still only barely acceptable.

When examining the quality of DAE, we found all five drives to be acceptable. The Kenwood was the only drive not to receive a perfect 10 in CDSpeed99's quality test, which compares extracted sectors to the original for any differences. CDSpeed99 gave the Kenwood 72X a rating of 9 out of 10, which means a few differences were found, but not enough to be noticeable audibly. Indeed, in listening to extracted .wav files from the Kenwood (as well as the rest of the drives), I could not discern any difference between the original and the extracted audio. However, some audiophiles may find even the smallest difference between copy and original to be unacceptable. The Kenwood may not be the drive to conduct DAE with for such folks.

CD-R Media Compatibility and Performance

 Optical Testbed I CD-R Media PerformanceDetails... 
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 7300|
Creative CD5230E (ATA-33) - 3833|
Delta OIP-CD4800A (ATA-33) - 3670|
Toshiba XM-6702B (ATA-33) - 3560|
Mitsumi CRMC-FX4820T (ATA-33) - 3493|
Windows 98 SE using FAT 32
Kenwood UCR-421 (ATA-33) - 9973|
Delta OIP-CD4800A (ATA-33) - 7520|
Toshiba XM-6702B (ATA-33) - 7210|
Mitsumi CRMC-FX4820T (ATA-33) - 7037|

When it comes to CD-R compatibility, all five drives were more than capable of reading our test media. The only real issue was with the Creative drive, which had problems reading the outside edge of our disc. Note in the transfer rate graph, all is well until the last few megabytes, where things come to a grinding halt and the drive drops severely in performance. Since this is limited only to the extreme outside edge of our CD-R, a region many discs don't even use, it should not affect most users (assuming the issue is not specific to our particular test unit). The other drives, with the exception of the Kenwood, read our CD-R at speeds equal to or greater than their CD-ROM performance. The Kenwood was a hair slower in reading our CD-R relative to its pressed CD performance, but was still ahead of the competition despite this.

CD-RW Media Compatibility and Performance

 Optical Testbed I CD-RW Media PerformanceDetails... 
Windows 98 SE using FAT 32
Creative CD5230E (ATA-33) - 2480|
Delta OIP-CD4800A (ATA-33) - 1230|
Toshiba XM-6702B (ATA-33) - 1150|
Mitsumi CRMC-FX4820T (ATA-33) - 1140|
Windows 98 SE using FAT 32
Creative CD5230E (ATA-33) - 5030|
Toshiba XM-6702B (ATA-33) - 2330|
Mitsumi CRMC-FX4820T (ATA-33) - 2310|
Delta OIP-CD4800A (ATA-33) - 1230|

Things were not so rosy for our five drives when it came to CD-RW performance and compatibility. Although only the Kenwood utterly failed this test, all drives took a significant hit in read speeds with our test CD-RW disc. The drive that held up the best here was the Creative 52X. It performed at roughly 16X-32X CAV, far faster than the other four drives. The Toshiba and Mitsumi are 8X-16X CAV-capable with our CD-RW and the Delta read our test disc at a CLV of 8X. The Kenwood, as mentioned earlier, refused to recognize our test CD-RW (Verbatim brand). This could be a very sore point for some folks, as Kenwood advertises CD-RW compatibility, and their last two firmware updates supposedly addressed compatibility issues. Our experience with two firmware revisions did not alleviate the problem. For some folks, this could be a non-issue, but if you plan on ever reading CD-RW media in your drive, be aware that Kenwood's drives are known to have difficulties with CD-RW media.

Editor's Choice: Kenwood UCR-21's Editors ChoiceAlthough Kenwood's highly touted 72X unit failed our CD-RW compatibility test, that was about the only weakness we could find. On the other hand, it trounced the competition in the disc and file copy tests as well as the DAE tests, and kept up admirably in the heavily access time-weighted CD-ROM Winmark despite clocking in over 33ms slower than the 48X Toshiba, our current access time leader. For those who may shy away from the Kenwood's relatively high sticker price (or the CD-RW issues), Creative's 52X unit does deserve an honorable mention. Very speedy in its own right, the CD5230E gives up quite a bit to the Kenwood in DAE speed and noise levels, but nearly makes up for it in CD-RW compatibility and read performance. While the Kenwood has the better "traditional" CD-ROM performance, Creative gets the nod as the more well-rounded performer. A final mention should go to the Delta unit for its bang-for-buck prominence. Barely-acceptable DAE and CD-RW performance kept it out of contention, but for those whose needs do not lie there, the Delta is a steal at only forty bucks.

See also: SCSI CD-ROM Roundup


Copyright © 1998-2005, Inc. All rights reserved.
Write: Webmaster