Blame it on Moore's Law. The consistent doubling of processor speed every 18 months now seems normal, almost natural to us. It is easy to be impressed with our demiurgical abilities and to believe we can mould digital technology however we desire.
The story of the liquid crystal display suggests otherwise. Despite massive worldwide investments in research and development, LCD technology remains expensive and plagued by poor image quality. Yet the dream of a thin flat-panel display and the new portable devices such a display would enable is powerful enough that Japanese conglomerates continue to invest billions of dollars in LCD development. Even if they eventually succeed - and they probably will - it will say more about persistence than about our mastery of the technology.
The story begins in the 1960s, when scientists first realized that liquid crystals could be useful for television displays. Liquid crystals ex-hibit the molecular symmetry characteristic of a crystal, but not along every axis. This results in their unique optical properties. Depending on how its molecules are align-ed, a liquid crystal will either scatter light or let light pass through. By using an electrical field to control molecular alignment, a thin liquid crystal layer can be coerced to display any image.
The stencil for the image is created by applying electric current to specified regions of the liquid crystal. Then, when the LCD is illuminated by reflected ambient light or a special backlight, the charged regions appear dark. As soon as the current is shut off, the regions fade back to translucence.
For small LCDs, such as those used in digital watches, each pixel is controlled with a separate wire. But this is impractical for displays that have thousands of pixels. Instead, the liquid crystal is sandwiched between verti-cal and horizontal wire grids. The intersections of those wires define individual pixels. This matrix allows m x n pixels to be addressed with only m + n wires.
In a passive-matrix display, any given pixel - at position (x, y), for example - is activated by applying current to the wires in the corresponding column and row. This works because while the voltage of a single wire is too small to affect liquid crystal alignment, the combination is sufficient. However, matrix addressing does not allow two pixels in different rows and columns to be activated simultaneously. If we tried to turn on pixels at both (x, y) and (a, b), we would end up actually activating four pixels: (x, y), (x, b), (a, y), and (a, b). Consequently, the screen needs to be filled in one row at a time.
The usual method is to start by setting the voltage level for all of the column wires and sending an electrical pulse down a specific row wire. Then do the same thing for the next row. Because the activated pixels immediately begin to fade to trans-lucence after they've been triggered, the screen must be continually redrawn.
This is where passive-matrix displays run into problems. The more rows in a display, the longer it takes to redraw. If it takes too long, users will notice pixels fading. One solution is to use liquid crystals with slower response times. But rapidly changing images, such as video or computer animation, appear jerky and discombobulated. It's a problem familiar to anyone who has seen the mouse cursor on a laptop momentarily disappear while moving.
Active-matrix displays solve these issues by using switches to regulate the flow of electricity. Each pixel in an active-matrix LCD is controlled by a thin-film transistor (TFT), a very simple switch connected to the pixel's row and column wires. When the switch is triggered, it uses current stored in a ca-pacitor to maintain a steady charge on the pixel during an entire screen-refresh cycle.
With active-matrix dis-plays, we no longer have to worry about pixels fading between refreshes, and that allows the use of liquid crystals with faster response times. The result is a higher-contrast display capable of handling full-motion video.
Unfortunately, the performance strides of active-matrix displays come at a high price. The added complexity of attaching TFTs to every pixel makes active-matrix displays difficult to manufacture and production yields discouragingly low. This is especially true for colour displays, where each pixel comprises three sub-pixels (red, green, and blue). Even one defective TFT among the 921,600 used in a typical colour display results in a visible flaw.
This sort of problem is routinely overcome by semiconductor manufacturers. But it will take a few years for makers of LCDs. And even then, prices will remain high as manufacturers struggle to recoup billion-dollar capital investments.
This provides a window for competing technologies to emerge before active-matrix LCDs hit their exponential growth curve. One possibil-ity, active addressing, was introduced in 1992 by In Focus Systems. Instead of selecting a single row at a time, active addressing en-codes signals so they can be multiplexed and sent to many rows simultaneously. This allows for faster refresh times without the extra cost and complexity of TFT arrays. However, active ad-dressing still doesn't match the image quality of active-matrix. Another possible contender is field-emission display, a sort of flat-panel cathode-ray tube that prom-ises reduced manufactur-ing complexity and lower power needs.
Yet active-matrix LCDs have a feeling of inevitability to them - too much has been invested in their success for them to lose. Although the technology is being pulled along by a strong growth curve, it is still in the long, flat region that comes before liftoff. Meanwhile, we are all being dragged willy nilly over the rocks. n
Steve G. Steinberg (steve@wired.com) is a section editor at Wired US.