26

Early CD-ROM drives were considered both expensive and slow (in comparison to HDDs). This limited their adoption until around 1992, when both the prices fell and the speeds increased dramatically.

In the mid 90s (roughly 1993–1997), I recall CD-ROM speeds increasing very rapidly; probably doubling in speed every 6–12 months. I got my first PC, which included a 2x CD-ROM drive, in 1993. And it seems that 32x drives were the normal option by 1997–98, even for "budget" PC's.

For perspective, the original 1x CD-ROM's of the late 80s transferred data at 150 KiB/s, while a 40x drive from the late 90s would do 6000 KiB/s.

I know that, fundamentally, the speed up is down to higher rotational speeds. However, the pattern of doubling of performance, and at a high rate, suggests the innovation was probably in the digital circuity. Like better error correction or better peripheral interfaces to the CPU. Though, maybe it was just about tweaking the analog aspects (motors, optics) instead? It seems weird to me that small spinning motors would be an innovation area in the 1990s, especially after so much of that had already been perfected for hard drives; additionally, the apparent exponential rise in the speeds.

So where were the innovations happening during this interesting period: motors, laser/optics, digital circuitry? How is this rapid speed increase best understood?

user3840170
  • 23,072
  • 4
  • 91
  • 150
Brian H
  • 60,767
  • 20
  • 200
  • 362
  • 6
    Don't forget the increase in speed of the receiving end. A late 1980s computer would probably have trouble to do anything useful of 6 MBytes/sec. – UncleBod Oct 04 '22 at 14:52
  • 1
    @UncleBod Fast SCSI-2 was around by then. Though true that was for high-end computers, not "budget" PC's. – Brian H Oct 04 '22 at 15:06
  • 5
    I suspect the apparent exponential rise is a red herring, and there were actually a series of step changes, as different problems were overcome. Going from "as fast as necessary for audio" to 2x or 4x may have been trivial; then different approaches to precision and balance were needed. There were many rumours of 12x being the limit above which damaged disks would explode under vibration. – IMSoP Oct 04 '22 at 16:08
  • 4
    Same reason why cd drives never really went above ~48x or so. Too much of a chance of the disk exploding... – TLW Oct 04 '22 at 23:43
  • Are you asking about average speeds or peak speeds? If I remember correctly the rotational speed could change as the head went in or out on the faster drives. – Mark Ransom Oct 05 '22 at 01:56
  • @MarkRansom You're probably thinking of Constant Linear Velocity. – IMSoP Oct 05 '22 at 08:41
  • 1
    It's kind of strange that we allowed 5.25 plastic discs to go beyond 10000rpm while 3.5 hard drives never exceed 7200rpm. – user3528438 Oct 06 '22 at 04:12
  • 1
    @user3528438 3.5" HDDs absolutely hit 10000 RPM. They just cost A Lot, so Seagate, WD, etc only market them to Enterprises. – RonJohn Oct 09 '22 at 08:54
  • @user3528438 - a lot of hard drive platters are made from aluminosilicate glass, which has a worse specific tensile strength than plastic. Glass has comparable tensile strength to polycarbonate from what I've seen (~35MPa)... at twice the density (2.5 versus 1.2 g/cm^3). Aluminum platters can do better of course. – TLW Oct 15 '22 at 19:13
  • @RonJohn I thought those are all 2.5" platters. – user3528438 Oct 16 '22 at 03:42
  • @user3528438 that’s traditionally laptop size, no? – RonJohn Oct 16 '22 at 16:51
  • 1
    @RonJohn The drive is the same 3.5" form factor but inside it the platters are smaller. – user3528438 Oct 17 '22 at 01:50

2 Answers2

22

While reading a CD, the main head assembly will simply move slowly from the inside of the disk outward, but unless the disk is absolutely perfectly centered--something that usually doesn't happen--the laser head itself will need to move in and out with each rotation in order to follow the track. Early CD mechanisms were designed to accommodate about five in/out wobbles per second, but spinning the disk faster will make it necessary for the head to move faster while following the tracks. Beyond the higher mechanical requirements, this poses another problem: at slower speeds, the head could move in response to indications that the beam was on one side or the other from where it "should" be, but as speeds increased it became necessary for the head positioning electronics to "predict" how the track would move, so as to need only small adjustments beyond that.

Another factor is that I don't think drive manufacturers really knew how fast drives could go without creating a risk that poorly manufactured CDs might explode catastrophically in such a manner as to destroy not just the disk, but also the drive as well. In many cases drives from more established brands wouldn't achieve speeds quite as high as those from lesser-known brands, which would compete on the basis of their higher speeds. If a lesser-known brand whose top spin rate was e.g. 4000RPM (compared with a normal "1x" top speed of about 500RPM) was on the market for a few months, and there were no apparent complaints despite the fact that consumers of budget drives might be more likely to use lower-quality media, then name brands could figure they could safely use comparable speeds.

From a practical standpoint, most drives ended up being limited to about 8000 RPM, which is 16 times the speed of a 1x drive when reading data near the center of a disk. Because one rotation's worth of the spiral track near the edge of the disk could hold about 2.5x as much data as one rotation near the center, and drives could maintain the 8000RPM speed even when reading data that was further out, such drives could read data near the edge of the disk 40 times as fast as a single-speed drive could, and were thus marketed as "40x drives".

supercat
  • 35,993
  • 3
  • 63
  • 159
  • 6
    I think the general story here is probably about right, but the "watch and see if people complain" methodology seems slightly implausible. More likely, it was a case of "stress test and release to market with a known safety tolerance". Cheaper brands might be more likely to cut corners on that testing and push the boundaries for a quick buck, but ultimately the mechanical problems of balance and precision were what needed to be innovated. – IMSoP Oct 04 '22 at 15:52
  • 2
    Wouldn't spinning the disk faster tend to reduce the amount of "wobble" because inertia? – Brian H Oct 04 '22 at 16:05
  • @IMSoP: I'm sure drive companies did their own testing, but there's no way that a drive manufacturer could be certain they'd identified the worst quality disk that anyone might want to use in thier drive, and I recall some manufacturers who were asked about whether their speeds would catch up to those of their competitors indicating they had no intention of pushing speeds that fast because of the dangers caused thereby. – supercat Oct 04 '22 at 17:53
  • 4
    @BrianH: If the center of the spiral coincided with the center of mass of the disk, the spinning mechanism was on a lightweight free-floating suspension, and there were no issues with resonance, then spinning the disk faster could reduce the distance the head would have to move to follow the track of an off-center disk. That's a lot of "ifs", however. Increasing rotational speed will increase centripetal acceleration and make it harder for bearings to accommodate it smoothly. Worse, any imbalance will like cause the axis of rotation to tilt slightly, setting up nasty resonance effects. – supercat Oct 04 '22 at 18:02
  • The OP asked about reading but IIRC there was quite a learning curve for increased speeds for writing as well. I seem to remember that reading at high speed was quite reliable reasonably quickly but for writing drives were marketed that could write at higher multiples but if you actually tried it you'd find you'd written coasters at the higher speeds. It also seems to me that the speed ramp for DVDs (and writable DVDs) was steeper and more reliable earlier than that for CD-ROMs. Is that correct? – davidbak Oct 04 '22 at 19:31
  • 5
    @davidbak: Writing a disk at 8x speed requires a more powerful laser than writing at 4x speed, but not exactly twice as powerful. Using too much laser power will create spots that are too big to be read reliably, while using too little will produce spots with insufficient contrast. From what I understand, drives that could write media with just the right characteristics at 16x speed, and media that could be written at 16x speed by drives with just the right characteristics, were created before there was any standardized way of characterizing drives and media to ensure compatibility. – supercat Oct 04 '22 at 20:01
  • 6
    At least the bigger brands didn't market their drives as "40x", but marked it " 40x max" instead because it's just the maximal rate the drive can achieve, not the rate it will get everywhere (as it was on CLV drives like 4x drives). Customers didn't notice the distinction between 6x (CLV) and 16x max (CAV), though, so the 16x drive sounded nearly thrice as fast. – Michael Karcher Oct 04 '22 at 20:26
  • 1
    From experience I can say they at some points definetly went over the "poor manufactured CD will explode" ... some later drives would detect excessive vibration and slow down... and there had been some tools around to "file down" CDs to perfectly round to reduce vibration... – PlasmaHH Oct 05 '22 at 06:46
  • @PlasmaHH - did they work better than the green pens that were sold to make the music sound better? – davidbak Oct 05 '22 at 17:51
  • @davidbak: If the CD was uniformly balanced except at the edge, I would think such a device might possibly help, but I would think a device which would spin the disk on a somewhat flexible bearing, and that included a strobe light that was triggered by a rotation sensor, would be more useful, among other things to distinguish the 99% of disks that are balanced just fine from those that might cause trouble. – supercat Oct 05 '22 at 18:01
  • 1
    @davidbak: they worked very well to make CDs where the center hole was not in the center make wobble less and thus achieve higher speeds. Not so much for CDs with variable thickness or stupidly unbalanced thick printing. – PlasmaHH Oct 05 '22 at 18:29
  • 2
    I believe the failure point was somewhat higher than the 40x you gave. Checking my memory with google I found multiple brands of old 56x drives for sale and 52x ones from Sony (as brand name as it got at the time). Also a forum thread about a 60x drive having launched but the poster being unable to get the (overseas) maker to send one, with speculation that it finally hit the bad disk breaking limit. I believe I had a 52 or 56x drive; and despite buying the cheapest cdrs I could find never had one fail mechanically. – Dan Is Fiddling By Firelight Oct 05 '22 at 21:04
  • @DanIsFiddlingByFirelight, checking my stock of old hardware, the fastest one I've got is a 52x. There were faster drives out there, but they were accompanied by rumors of exploding disks, where 52x drives weren't. – Mark Oct 06 '22 at 00:28
  • Comments are not for extended discussion; this conversation has been moved to chat. – wizzwizz4 Oct 06 '22 at 08:07
7

'for perspective, the original 1x CD-ROM's of the late 80s transferred data at 150 KiB/s, while a 40x drive from the late 90s would do 6000 KiB/s.'

no. The original drives read at 150Kb/s, varying the speed of the disc to keep the data rate constant. 8x drives could read at 1200kb/s and some 12x drives at 1800 Kb/s. constant linear velocity, so same data rate anywhere on the disc. All drives rated faster than 12x used constant angular velocity with the stated speed being the rate readable at the outermost edge of the disc only, with the readable speeds at the inside of the disk being approximately 3x slower, and even increasing the stated speed because later CD-Rs are at the limit of the spec - 700Mb instead of 650, The increase in speeds was all marketing nonsense. In practice a 40x drive was no faster than a 12x drive from a few years earlier -the hardware was almost the same anyway.

camelccc
  • 645
  • 4
  • 9
  • 2
    It's too bad that software publishers didn't master disks which didn't need all available space with one copy of the data written from the inside out (which could be accessed more quickly by slower CLV drives) and one copy written at maximum density from the outside in (which could be read more quickly with CAV drives). That would have allowed the latter to actually achieve something close to their advertised data rates. – supercat Mar 08 '24 at 19:02