Compact Disc - The Inside Story

Since my previous article in February 1990, I am strongly compelled to expose some CD myths that have been circulating around. I would like to present some facts, and my own observations.

CD data recording is a purely digital system, with audio samples supplemented by two layers of error correction and other control information. No mechanical system is perfect, so the data read from the disc is not always identical with what was recorded. In fact, of the 7350 data blocks per second being decoded, even a good player will find a few blocks per second with errors on a good CD. I have measured this, and found that CD quality varies enormously; from a block error rate (BER) of less than one per second, up to hundreds per second. The same CD will generally give similar results on different players. The CD standard demands a maximum BER of 1/1000, or about 7 per second.

These errors may seem nasty, but all audio samples in bad blocks are totally and completely repaired by the two error correction systems. Of course the BER may get so high (above about 100 per second) that the errors cannot be corrected. The CD player then pulls a neat trick by doing an "interpolation". This is the process of making a good estimation of what a bad audio sample should be, based on its immediate neighbours. This is over a very short time interval, say 20 microseconds, so the "fudged" sample will go totally unnoticed by even the most discerning listener.

On some CDs, the whole disc may be played without a single interpolation occuring, so the audio samples were a perfect reproduction of what was recorded on the CD master tape. Other discs will have the occassional short burst of interpolations, which can usually be aligned with manufacturing defects or obvious marks such as bad scratches or anything larger than about 1/4 mm in size. Pock-marks in the surface are not as common as they used to be, but I still see pin-holes in new discs. To observe these, hold the disc close to a powerful incandescent bulb and look through the disc metalisation; they are quite obvious. The variation in thickness of the metal layer between different discs can also be noticed, but I have not experienced any correlation with error rates in this respect.

Finger prints and fine dust can affect the BER, but they don't cause interpolations. Another interesting cause of errors is slightly warped discs and imperfect seating of discs on the small platter. These both overstress the laser focussing and fine tracking servo mechanisms, so that the laser cannot quite be kept in perfect position. I have operated player mechanisms out of their cases in bright light with no effect on the BER; shooting down the "green ring" brigade in flames. In fact my current "home brew" player has the mechanism totally exposed.

Misaligned or warped discs can be easily spotted by watching the image of light bounced off the top surface on to a wall or ceiling. This cannot be done for discs with large printed labels. I have improved the seating of discs on the platter (and the BER) by cleaning the surfaces that mate with the inner part of the disc with metho. It was amazing to see the size of particle and quantity of dirt on the cotton bud after cleaning my four year old player. I have also noticed CDs with holes that do not perfectly fit the platter, allowing the minute amount of mis- alignment to make the fine tracking servo work harder, and at times affect the BER. On my player the fine tracking only has a range of about 35 micrometres before the course motor driven lateral tracking has to move the whole head, which gives an idea of the tolerances involved.

It is surpring the amount of physical abuse an operating mechanism will take before interpolations occur. It can be moved all around at any angle with minimal changes in the BER. Impulsive shocks are the worst; sharp finger tapping on the mechanism will produce interpolations every time. I have also operated by player on "lively" unbraced speaker cabinets with little effect on the BER. Even with this relative immunity to external vibration, I still think it is wise to play attention to player damping and suspension of the mechanism.

All along, I have been talking about the effect on the BER of physical things, and unless interpolations occur, all errors are completely corrected and the digital sample reproduction is perfect. Mild doses of interpolation are inaudible, but if it does become chronic, a grainy caste begins to appear over the music. When worse, it sounds like crickets chirping in the background or like excess finger nail clicking on piano keys. Depending on the player, it may give a loud click (like an LP) or give up totally and stop playing. When this happens, you know you have a bad disc. I have only every seen one disc like this. Close monitoring showed it was interpolating almost continuously through the whole disc, and a visual inspection showed hundreds of pin holes. On one player, it was not even possible to skip tracks.

I have no support for stick-on CD rings. They can produce no audible effects, due to the nature of the digital system. I cannot even see any valid reason why interpolation rates on bad discs would be changed. The Armour-All treatment may have some basis, but only as a cleaner of dirt and finger prints. It may "fill" scratches, but should be polished off as much as possible. I prefer to handle my discs very carefully by the edges, and always keep them in closed cases when not being played. Dust specks should preferably be removed with a photographer's type "blower brush", and not any cleaning cloth, no matter how soft. Hold a new disc up to a dim light so that the light bounces off the unlabelled disc surface into your eyes. Notice the lack of scratches. Now "polish" the disc with a cloth and look again. You will probably find the surface is not as pristine as before.

While everyone in pondering the above, I am working on two more articles to discuss the improvement of audio circuits in players, and the two new "single bit" systems. *

* [I still haven't got around to writing these articles, so the original readers have had plenty of time to ponder... GB, Oct '96]