The humble CD-ROM has gone from being the plaything of well endowed HiFi enthusiasts less than two decades ago, to becoming the most widely used means of bulk information distribution on the planet. Now we have the DVD which offers multiple Gigabytes of storage in the same format, with Read/Write variants beginning to emerge. Are you finding this Babel of storage formats and technologies confusing ? In this feature we will attempt to dispel some of this chaos by delving into the basic ideas which are the basis of rotating optical storage, and discuss the evolving standards.

Where to start ? Probably the early sixties, with the invention of the laser, which is essentially nothing more than a very sharply tuned optical oscillator. As such the laser has a very important quality, in that it can produce light which is coherent and essentially of a single wavelength - purists will no doubt lambast me for this gross simplification, but delving into the resonant behaviour of Fabry-Perot cavities is a little outside the scope of storage devices !

Purely monochromatic light allows a user to focus it into an extremely small dot, which is indeed the central issue in high density storage.

Alas a bench-top laser the size of two shoe-boxes isn't the most practical way to build a compact low cost storage device. These did not become feasible until the advent of the solid state laser diode. The laser diode is a miniature laser, produced by essentially blending the technology of the humble Light Emitting Diode (LED) with that of the optical resonator.

The audio Compact Disk (CD) was the first optical bulk storage device to hit the market in appreciable quantity. It was well and truly a quantum leap in technology over the century old groove and stylus technology it replaced, and other than the unfortunate choice of cumbersome 14-bit encoding, intended to reduce the cost of consumer digital to analogue converters, the Compact Disk has proved to be a great success. Philips and Sony did well in this instance, while not managing to do too badly for themselves in the process.

It did not take very long for the voracious computer storage industry to recognise the potential of the Compact Disk, so by the late eighties moves were afoot to standardise a data storage format for use with the medium. Today it is almost unthinkable for software to be distributed through a satchel of magtapes. We pop a 650 MB CD-ROM into the drive, and mount an ISO 9660 filesystem.

The latest step in this technological evolution is the Digital Video Disk (DVD), which like its forebearer, has become the target of the storage industry in the form of the Digital Versatile Disk (again DVD). With many Gigabytes of capacity, it outstrips the now ubiquitous CD-ROM in capacity and transfer rates. With moves afoot to standardise technology for Write Once Read Many (WORM) and Read/Write (R/W) variants of the DVD, we are now poised on the threshold of multi-Gigabyte, dirt cheap, universally readable optical archival storage.

What are the technical issues in this technology, what are the risks and benefits, and how portable will this new technology be ?

The best way to gain insight is to dive into the inner workings of these machines to see how they work.

The Compact Disk

The Compact Disk bears much mechanical similarity to the traditional magnetic disk, be it floppy or Winchester. A platter rotates at a set angular velocity, and a read head is stepped to a track, upon which a stream of data is read from the surface of the platter. Alas this is where the similarity ends.

The platter of any optical, or magneto-optical disk is manufactured from a good quality optical material, mostly a polycarbonate plastic, and in some instances, where durability is critical, a drawn sheet glass material.

In a mass production CD, the polycarbonate is injection molded into a die, one side of which is a Nickel master. Ones and zeroes are encoded into a spiral of tiny pits on the master, which appear as bumps on the polycarbonate substrate.

The side of the platter with the impression of the metal master in it is then coated with a highly reflective material, usually aluminium which is vacuum sputtered on to the substrate. The metalised layer on the CD is then coated with acrylate lacquer which is hardened by a UV lamp. Finally a label is either offset or screen printed over the acrylate.

This produces the CD which we all know and love. Reading it is where things get interesting.

The CD is read by the optical head through the polycarbonate and thus the pattern of pits on the surface of the master is seen through the 1.2 mm thick layer of plastic, replicated on the reflective surface.

In terms of function, the optical head focusses the beam from the 0.78 micron infrared laser into a tiny dot on the reflective surface of the CD. As the CD rotates, this dot passes over the pits in the reflective surface. These pits are seen by the laser as bumps, which raised above the surrounding surface or "land", scatter the laser beam. An optical detector which senses the reflected light, will see a drop in the intensity of the reflection as the beam passes over the pit, since much of the energy of the beam is being reflected in directions other than that where the beam arrives from. Thus a 1 or a 0 can be resolved.

The optical head is itself a very clever piece of miniaturised optical engineering. The head assembly is very large and heavy compared to a Winchester disk, and is usually mounted on a set of rails along which it rolls. Physically the arrangement compares best to the trusty voice coil arrangement we love and know. in this manner, the head can be positioned coarsely over the platter.

The optics are designed to both transmit the laser beam and read the reflection from the CD surface through a single optical path and focus lens. The head itself resembles a tubular T-piece, with the laser shining down the barrel. At one end sits the focus lens, just above the platter, at the other the laser and its collimating lens.

A 45 degree angled prism is used as a beam splitter to reflect out the T junction, into the detector assembly. A quarter wave thickness birefringent optical plate sits between the prism and the focus lens. This arrangement converts the horizontally polarised light from the laser into circularly polarised light, which reverses its sense of polarisation when it reflects off the CD platter. The prism then bounces the reflected beam out at 90 degrees to the optical path between the laser and the surface, into the detector. In this fashion none of the light from the laser can leak into the detector to degrade the signal. The detector sees only the reflection from the disk platter.

Focussing the beam precisely into a spot with a 1/2 micron diameter is not a trivial task. The focus lens is therefore embedded in a miniature voice coil, and electrical current through the coil is used to drive the lens up or down to focus the spot on the surface.

Because the optical head is much too large and heavy to move quickly enough to track the stream of pits travelling beneath it, a fine tracking mechanism is required. The spot must be maintained in the centre of track it is reading data from.

Therefore a second coil is used to tilt the focus lense, so the beam can be rapidly steered at right angles to the track. Messy ? Inevitably so, but the standard design used is close to identical across millions of drives and is quite elegant.

How do the control circuits which drive the focus and beam steering coils know how far to drive them ? This question could be rephrased as "what does your CD-ROM drive have in common with a laser guided bomb ?" The answer is that both employ what is termed a "four quadrant detector". The photo-diode which senses the reflection off the platter is really a cluster of four detectors, arranged like a Maltese cross or cloverleaf. When the beam is properly focussed and centred on the track, all detectors get equal illumination. Via clever trickery with a cylindrical lense in front of the detectors, the pattern of light which falls on the four detectors produces a particular imbalance between the four detectors, depending on whether the beam is focussed above or below the surface, or left or right of the track.

In many respect the optical head in your CD drive is a marvel of modern mass production miniature optical engineering. The ability to select and follow one track at a density of 16,000 tracks per inch is not a trivial chore for a piece of consumer electronics.

What comes out of the optical head and its amplifiers is an electrical signal which changes in strength depending on whether the beam is hitting a pit or the land between pits. Thus 1s and 0s are read off the surface.

The raw data stream in this format is not the whole story. Needless to say the CD surface will have imperfections, and the polycarbonate surface 1.2 mm above the data is likely to have scratches, fingerprints and perhaps residue of yesterday's lunch or morning doughnut on it. Therefore the raw data stream will contain every once in a while erroneous 1s or 0s, or even runs of erroneous 1s and 0s.

Therefore an error detection and correction code is required. Since the data stream is serial in nature, a forward error control technique is required.

The Sony-Philips CD standard uses a Cross Interleaved Reed Solomon code (CIRC) which puts on average about 8 bits on the platter for every 7 bits of real data. The code is claimed to correct an error burst of up to 4000 bits, and interpolate 12300 bits. Despite this, as we well know, people manage to mistreat their CDs badly enough to defeat even this scheme.

The ISO 9660 standard is based upon the ECMA-130 standard, which divides the surface into sectors. Each sector contains 12 bytes for synchronisation, a 4 byte header, 2,048 bytes of user data, a 4 byte CIRC error detection checksum, an 8 byte space, and finally, 276 bytes of CIRC error correction data. Two modes are used, one with ECC and one only with error detection.

The hardware in the drive must synchronise to the sector, and on the fly perform the decoding and if required correction of the data. The block of clean data is then buffered, and sent off to the host over the SCSI interface.

The path between a stream of pits on the surface of the CD to the host is thus a tortuous one !

Recordable and ReWritable CD-ROMs

The latest evolution of the Compact Disk in its standard format is the Compact Disk - Recordable (CD-R), and Compact Disk - ReWritable (CD-RW). The former is essentially WORM technology, where the CD is can only be written once, the latter allows multiple writes and reads of the medium.

The CD-R differs from a standard CD in several respects. While the polycarbonate disk is injection molded the same way, instead of a pattern of pits in the master, a wobbled continuous groove is employed. The next step in the fabrication process is the application of a cyanine, phthalocyanine or azo dye layer. A layer of Gold or Silver is then sputtered over the dye, to act as reflector. The acrylate lacquer and screen print are then applied. various manufacturers will manipulate the composition of the dye, the thickness of the dye, the type of reflector material and the thickness of the groove to achieve best performance.

A drive to write such media must of course contain the appropriate encoder hardware and firmware, and be supported by a software tool which generates the appropriate ISO 9660 format bit pattern for recording. The laser must be capable of operating at a high and a low power output setting. At the low setting, it reads the medium. At a high setting, it burns a spot in the dye. Where the dye is burned, its optical properties change - either it is deformed or a bubble is created. As a result when the medium is read by a laser, the burned and unburned areas along the track appear as ones or zeroes.

The durability of a CD-R medium depends critically upon the quality of dye, the quality of the lacquer, and the quality of application. If any of these are compromised, water and oxygen may penetrate and corrode the dye, producing an increasing number of bit errors as the medium ages.

The CD-RW is conceptually similar to the CD-R, but the dye is replaced with a "phase change" alloy material. Such a material has the property of changing its phase between crystalline and amorphous, depending on how it is hit with the laser write pulse. Because each of the two phases has different optical properties, this produces a readable contrast.

Durability much like with the CD-R depends on the quality of materials and application. Compatibility may be an issue with phase change technology, in that a standard CD drive may not be able to read a particular manufacturer's CD-WR medium, once it is written. The general trend is that the CD-WR drive will read arbitrary "standard" CD or CD-R media.

Digital Versatile Disk

The evolution of the DVD has been much less haphazard than that of the Compact Disk, which was devised with no regard for computer applications.

The Compact Disk was a great leap forward for the music industry, but left the movie industry out in the cold. The basic technology and recording density permitted only analogue video disks, which were never very popular. By the 1990s th movie industry decided that it wanted a digital optical disk which could fit a whole 2 hour feature movie. Needless to say the computer industry wanted as many Gigabytes as it could get.

Toshiba and Time Warner developed the SD (Super Disk), while Sony and Philips devised the MMCD (MultiMedia CD). These proposals were merged in 1995 to define the DVD standard. The intent was to avoid a rerun of the VHS vs BetaMax fracas.

Another important design feature was the provision of a common recording format for data and backward compatibility with the existing CD/CD-R media. The latter means that consumers preserve their investment in media, the downside of which is that some of the warts of the original Philips design live on.

The technology which enabled the DVD to record with a much higher density is the use of a red, 640 nanometre wavelength laser, instead of the 780 nanometre laser in the CD. The shorter wavelength means a smaller spot, and a reduction in track pitch from 1.6 microns down to 0.74 microns, and a reduction in minimum pit length from 0.8 to 0.4 microns. Needless to say the mechanism used for focus and tracking had to be much more accurate.

Increasing the density eightfold while using the standard CD medium size did not yield the capacity required, so another change was incorporated. The DVD employs a disk which is a laminate of two 0.6 mm thick disks. The inside surfaces of the two halves of the disk contain the recorded data. In this fashion a DVD can be double sided. However, this was still deemed to provide too little capacity, so a dual data recording layer was adopted. Depending on how deep the laser is focussed, it reads pits either in the first to second layer.

A DVD therefore can exist in four different formats:

The use of a symmetrically bonded laminate provided another benefit - the medium would not warp as a result of humidity or temperature affecting the polycarbonate. Any stresses between the two halves would be balanced.

Backward compatibility with the CD was a blessing in disguise, while a second lens and laser were required for the optical read, the second lense also provided for a different depth of focus for dual layer reading.

A typical first generation DVD-ROM drive will provide rotational speeds of 3500-1200 RPM, and due to the higher density achieve a much faster read rate in comparison with a CD at the same RPM. A sustained rate of 2.7 MB/s is quoted for the Toshiba W1101 reading a DVD-ROM. Typical seek times are about 200 msec. A 20 year lifetime is claimed for the media.

Writable variants of the DVD are a contentious issue, as vendors jockey for the adoption of their pet standards.

The DVD-R recordable medium using established CD-R dye technology and the standard has been agreed, defined by the "Book D" component of the "DVD Book" standard (Book A defines the DVD-ROM). The DVD-R provides 3.8 GB single sided, or 7.6 GB double sided.

Where we see divergence is in the ReWritable variants. Toshiba/Hitachi/Matsushita/JVC/Pioneer defined their "Book E" standard, while Sony/Philips/HP/Mitsubishi/Ricoh/Yamaha have submitted their proposal, designated DVD-RW to the European ECMA standards body for adoption. To add fat to the fire, a third proposal, DVD-NEC/MMVF by NEC is also being touted.

For a single sided medium, the standards yield different capacities:

The DVD-RAM is the most mature, but smallest in capacity. The DVD-RW can read the CD-RW medium, and has more capacity. The DVD-NEC uses Magneto-Optic technology and is incompatible with the other media. The DVD-ARM and DVD-RW are compatible with the CD-ROM, DVD-ROM and DVD-R media.

The OSTA standards body is doing its best to persuade the vendors to tweaks their respective standards to provide at least a "Multi-Read" level of compatibility so that they can read all earlier media, ie CD-ROM, CD-R, CD-RW, DVD-ROM and DVD-R, even if they are mutually incompatible. VHS vs Beta again ? Looks like it.

How this pans out remains to be seen, since the two largest consortia are both very large and a fight in the marketplace may take some time to resolve.

Where does this leave the hapless sysadmin, when the IT manager storms in saying "when do we get this fantastic ReWritable DVD I just saw in trade journal XXX for backups ?" This is indeed an interesting question.

In principle, the best strategy, whatever standard you opt for, is to verify that the drive can read samples of your existing CD-R or CD-RW media. If the drive can't read it reliably, then it is obviously not the best choice.

Regardless of the squabbling over standardisation, the DVD-ROM, DVD-R and DVD-(whichever-turns-out-to-be-rewritable) will provide a massive leap in storage capacity for the industry consumer, and home user intent on doing backups or archives. Home users will benefit especially, given the costs of high capacity tape media. If the rewritable DVD goes the route of the VCR, the longer term cost of the technology will plummet.

What is the next step beyond the DVD ? A blue laser resulting in even higher recording density is the current target, and already prototypes have been reported. Another possibility is a holographic optical head and recording materials. The current generation of DVDs is well under the technology limits of the basic model.

The evolution of CD/DVD technology is an instance of where the commodity product model has mostly worked very well and to the benefit of the consumer. Pity it happens so seldom.