The Problem With Rotating Disks

As mentioned in previous articles, what determines the minimum spot size on an optical disk is the signal to noise ratio:

• The better the ratio the smaller each spot can be;
• The smaller each spot the higher the information density;
• The higher the information density the more information you can hold and the faster you can read it.

Through transitive logic then, the higher the signal to noise ratio, the more information a disk can hold and the faster you can read it. Today I am going to look at how the rotation of the disk plays into this.

Information can be thought to be stored on a spiral on the disc surface. The read head reads this as a linear string of information. Think of it this way – you could just as easily print the track from a disk onto a Mylar ribbon a few meters long. Were you to pull the tape across the read head at the same speed as the linear equivalent of the rotation of the disc, the electronics of the disc drive would read it in exactly the same way. Therefore, the faster you can move the information in front of the read head, the faster it can read it.

There are two things to remember here:

1. Under normal operation an optical disc rotates, and
2. Any substrate material has a finite tensile strength.

When anything rotates, centripetal force acts to pull it apart. The higher the angular velocity, the greater the force. If the particular type identity of the ‘anything’ is pizza, the force serves to stretch the dough. Properly applied, this will fit the dough to the size of the pan. Improperly applied, the dough will tear and fly apart into non pizza-type objects. The critical division between proper and improper centripetal force application within this scenario is an improper level of force exceeds the tensile strength of the dough.

The same phenomenon occurs with optical disks. As the rotational speed increases, so too does the centripetal force acting to tear the disc apart. Once the tensile strength is exceeded, the disk will fail catastrophically. This places a practical limit on the speed at which the storage medium can be moved passed the read head.

This means there is a finite limit to the area a read head can cover in a given time frame. Multiply this area by the number of points you can store on it and you get a finite limit to the speed at which you can transfer information off the medium. This also means transfer speed is directly related to information density. Once again then, information transfer speed is effectively limited by the signal to noise ratio.

There is a second factor the spinning nature of a properly operating disc injects into the optical storage equation: motion blur. Consider shutter speed on a camera – it takes a certain amount of time, however short, for the camera to gather enough light to expose the film. An electronic sensor, while much faster, still requires a certain number of photons to register an image. The photo sensor in an optical drive is no different.

If the object you are trying to photograph is moving, what you will get is essentially an infinite number of photographs during that time overlaid on the image resulting in a blur. In an optical drive you may only be concerned with a single point at a time, but that point is moving. During the exposure, that point is moving through the sensor region.

This forces an engineer to make tradeoffs in the design. If the sensor region is as large as a single point on the disc, then either the preceding or succeeding point will be detected during part of the exposure. Because light gathered from the second point is not light from the point you want, it must be counted as noise. I have already been over why this is a bad thing.

One solution is to shrink the sensor region or shorten the exposure time. The problem with this is it reduces the amount of light gathered during the exposure. This means a reduction in signal also reducing the all important S/N ratio.

To counteract this loss in light gathering power, an engineer could boost the power of the laser. This too can only be done to a certain point. Eventually the power of the read laser will equal the power of the write laser, destroying the information on the disc in the process. This may not always be an undesirable phenomenon. I could understand why an online video rental company would want to send one time use discs to its customers rather than persistent media.

Sony’s solution to this problem may be the most workable yet. This article describes an electro-mechanical tracking system. It is a tilting mirror similar to what you would find in a DLP projector. Instead of reflecting light to produce images, it tilts the optical path to track the point on the rotating disc. This technique bears more similarity to the operation of a siderostat than image stabilization (contrary to what the article claims).

The downside of this approach is it is a mechanical system. This means it will eventually break. It may take a long time to break, possibly even longer than the bearings on the rotor or the read head. But, at the speed they claim to be running it at it will fail within a human life span.

Ian Anderson