The Analog Compact Disc Page 2
Fig.2 The relationship between the original digital audio data, the EFM-coded data, the CD pit structure, and the recovered HF signal. (Reproduced with permission from Principles of Digital Audio by Ken C. Pohlmann, Howard W. Sams & Company, Second Edition, 1989.)
How CDs are made
The entire disc-manufacturing process is shown in fig.3. In the first step, mastering, the binary data on a master tape are converted to pit-and-land structures. A large glass disc, called the glass master, is coated with photoresist, a material that changes its properties when exposed to light. The coated disc is put on the mastering machine's turntable, spun, and exposed to a laser beam turned on and off by the binary data we wish to record on the disc. (footnote 3). The exposed master is then "developed" by applying a liquid chemical that dissolves the photoresist only where it has been exposed to the recording laser beam. Dissolved areas in the photoresist are the pits.
Fig.3 CD manufacturing process. (Reproduced with permission from Principles of Digital Audio by Ken C. Pohlmann, Howard W. Sams & Company, Second Edition, 1989.)
The developed glass master is coated with silver, then put in an electroplating tank which applies a thin layer of nickel to the silver. When separated, the nickel coating can become a stamper. Alternately, the nickel coating becomes a "metal master," from which "metal mothers" are formed, which in turn can create hundreds of stampers. Fig.4 shows a metal master being separated from a silvered glass master. Interestingly, the technique for making CD metal masters, mothers, and stampers is identical to that used in LP manufacturing. Even the chemical solutions in the plating baths are the same.
Fig.4 Metal CD master being separated from a silvered glass master.
The stamper is trimmed and put in an injection-molding press. Polycarbonate beads are heated to their melting point, then injected under high heat and pressure into the mold cavity containing the stamper. The pits on the stamper's surface are thus transferred to the polycarbonate disc. This process takes about eight seconds. Fig.5 shows a disc's worth of polycarbonate (about 18gm) in bead form; fig.6 is the clear disc as it comes from an injection-molding machine.
Fig.5 One CD's worth of polycarbonate beads.
Fig.6 Clear CD from the injection-molding machine before metallization.
In the next step in CD manufacturing, metallization, a thin, reflective layer of aluminum is either sputtered or vapor-deposited on the clear polycarbonate disc. A lacquer coating is then applied to the reflective layer to protect the aluminum and prevent it from oxidizing, and the label is silk-screened onto the disc. The finished discs are put into jewelboxes and packaged for shipment.
Bits is bits?
Although CD manufacturing appears to be a straightforward process of stamping inviolate ones and zeros into a plastic disc, these manufacturing techniques introduce analog-like variations in the quality of the HF signal read from the disc.
Because the HF signal recovered from the finished CD is created by the tiny pit-and-land structures, it follows that any changes in pit shape will affect the HF signal. Well-formed pits produce a good-looking HF signal; poor pit geometry creates a poor-quality HF signal.
A clean HF signal is essential not only to low error rates and good tracking ability, but also to sound quality. Although the HF signal undergoes significant processing before the raw audio data and timing clock are recovered, many digital designers agree that the HF signal's shape and quality affect how the disc sounds. Some high-end transports even have circuits to clean up the HF signal before it's sent to the decoding electronics.
That the HF signal's quality affects the sound is suggested by many examples of audible differences where there should be none. In 1986, Doug Sax (footnote 4) first alerted me that CDs made from the same CD master tape, but pressed in different factories, sound different. Doug routinely buys CDs of records he's mastered and compares them to the original CD master tape from which the CD was made. He has found a huge variability in sound quality between different pressing plants—some plants produce discs that sound very similar to the original; others make discs that sound dreadful. The only difference is in the manufacturing process. Indeed, engineer Bob Katz's experience, described in the companion piece to this article, further suggests that, although the binary 1s and 0s on two CDs may be the same, it doesn't necessarily follow that the discs will sound the same.
Chesky Records supplied me with three sampler discs cut from the same master tape but manufactured using different techniques. The first disc was made conventionally—polycarbonate with aluminum metallization; the second disc was made from polycarbonate, but with gold metallization; the third disc was molded from a material called "Zeonex," and metallized with gold. Since all three discs were made from the same master tape, any sonic differences between them should be solely the result of the manufacturing process, disc material, and metallization.
Footnote 3: The laser beam stays on all the time, but is passed through a modulator that either deflects the beam or allows it to pass straight through. The modulator is a crystal with a particular lattice orientation, across which an electrical signal representing the data is applied. The voltage across the crystal causes the lattice to change orientation, deflecting the beam off its path toward the CD master disc. You can put a photodetector on this "rejected" beam, decode it, and listen to the music while mastering.
Footnote 4: Doug Sax is a mastering engineer, co-founder of Sheffield Lab, and the father of modern direct-to-disc recording. See my interview with him in October 1989 (Vol.12 No.10).