How CDs Work

How CD Players Work

You press open, put in a CD, hit the play button and relax back listening to wonderful music. But how is music taken off that little 12cm disc? All you can see is a silver (and sometimes gold) coloured surface. Unlike records where you can actually see the track and in some cases the “wavy” patterns that make up the sound wave, CD’s have a much smaller track that contains the music encoded in a digital format.

Most CD’s are made from polycarbonate and aluminium (a few special ones are actually gold inside, it supposedly is prone to less errors hence they sound better). The aluminium is the shiny layer inside that you can see, interestingly on some discs you can actually see through them, like a bright light! On the surface of that layer are a series of millions and millions of little bumps (or pits depending which way you look at it) that have been stamped at the factory into the aluminium. The stampers can be use many times and cheaper CD’s may have been stamped with worn out dies and have higher error rates than others. The bumps form a track that starts in the centre of the CD and spirals outwards. Their spacing is the same, so the disc must slow down the further out you move from the centre. They start spinning at about 500 RPM and end up around 200 RPM. The CD player automatically controls this change in speed as it needs a constant rate of data from the disc. (CD-ROM drives in computers and some car CD players can speed this up to read more).

Okay, so we have a disc spinning at high speed but how does the data get read? There is a small laser diode that shoots a mostly infrared beam through some mirrors and prisms that reflect the beam through a lens. Just before this is a diffraction grating that splits the laser into three beams, the extra two are for keeping it on track. This lens is the bit that often becomes dirty and can result in skipping or no playback. The lens is suspended with tiny motors that can move it up or down and left or right. This movement, again controlled by the data feedback from the disc, allows the lens to focus and to track bumps and jumps. So if you bump the CD player, the laser will try and move in the opposite direction to keep its spot. For minor jolts, it works. It also moves to follow the outward spiral track of data, whilst also be moved as a whole along a motorised sled.

The laser beam shines through the surface of the disc then reaches the metal surface. It will either hit a bump or no bump. The height of the bumps is related to the wavelength of the laser beam. If you remember your high school physics, you remember about waves cancelling or adding. This is basically what happens here, so the reflection is either a bright reflection or a dim reflection. This reflection goes back down, through the lens and via a special prism that sends it away from the laser diode towards a detector. The detector has about 5 or more sections, the centre section usually just reads the main data stream from the disc, the others are used for tracking; keeping the laser pointed where it should be. The signal is then carried via wires to a so-called RF chip. It is called RF (radio frequency) because the data rate is very high and doesn’t yet resemble anything like audio. If you view the signal on an oscilloscope, it looks a bit like overlapping waves and is called the eye pattern. The digital data is then extracted from this signal which, believe it or not, is actually analog. Not audio analog, think of it more like the old dial-up modems with the digital signal carried by analog noise but at a much high frequency.

Now we have a true digital signal, but still not a straight digital audio signal. Processing has to be done now. There is a lot of extra data, such as track positions, error correction data and packing data. The track positions are obvious; the player needs to know where it is and where it has to go. When you first load up a CD, the player will read the TOC or table of contents. This is like an index of the disc. Error corrections takes two forms, firstly it can actually work out the missing data using mathematical calculation, like a checksum. Secondly, it can best guess the missing data and fill in the gaps; even though it may not be 100% correct it should be pretty close. These types of error correction you generally wont hear. Lastly it will simply skip over the damaged part and you will certainly hear this!

All the data is store in blocks and eventually all the extra bits are used or removed and we finally end up with the well know 44.1kHz (44,100 samples per second), 16-bit data stream that actually carries the audio. Some units will still carry out extra processing at this stage, one example is upsampling, where the digital data is chopped and converted into an even finer data stream, often 24 bit / 96kHz. Finally it goes to the digital to analog converter that we have looked at previously!

Written by Leon Gross, originally published in Audio & Video Lifestyle magazine.