Silicon dream TV refuses to switch off

It reads like a VIP list of failures - Hewlett-Packard, Toshiba, Intel and Philips.

Each of these technology powerhouses tried to conquer a promising technology for making thin, big-screen televisions - called LCoS, or liquid crystal on silicon - only to back out in defeat.

"The roadside is littered with those who have tried and failed," said Sandeep Gupta, the chief executive of MicroDisplay, a privately held designer of LCoS chips.

As the TV market moves to bigger and better screens, LCoS is one of a few technologies that, in theory, fit the bill to replace bulky cathode-ray tube TVs and pricey plasma displays.

In an LCoS TV set, light reflects off one or more small microchips made up of a layer of liquid crystal and a layer of transistors, projecting an image on to the front of the screen.

The pictures from LCoS sets can be rich and bright. But, as more than one technology giant has discovered, LCoS is also a black hole for investment cash.

Meanwhile, Texas Instruments using another rear-projection technology - digital light projection, has sold five million DLP engines, used in cinemas, projectors and TVs.

"TI has done a fantastic job marketing DLP," Bob O'Donnell, director of personal technology at IDC, a market research firm, says.

A year ago, Intel announced at North America's biggest consumer electronics show it would reshape TV with LCoS products.

"It's real," proclaimed Intel president Paul Otellini in Las Vegas, adding that TVs built with its high-definition displays would be on the market by the end of 2004.

That forecast deeply embarrassed the world's largest chip maker, which delayed the project and then cancelled it in October. Intel says it overestimated the economic pay-off, although experts say it had an unrealistically complicated design.

It was deja vu for Chris Chinnock, a senior analyst with display market researcher Insight Media. He has watched project after project on LCoS - developed in the 1990s by IBM and Japan's JVC - being cancelled or quietly shelved.

"It has cast a serious pall and doubt about the technology," said Chinnock.

Among the first large company to try to commercialise LCoS was Hewlett-Packard, which originally expected high volumes of components in 1999. Chinnock, who followed the HP deal closely, said the project was shelved soon after.

"They couldn't get the price and performance," he said.

In 2002, France's Thomson pulled the plug on an US$8000 TV set built with LCoS panels from Three-Five Systems. Three-Five later spun off the LCoS business into a company called Brillian, which earlier this year lost a lucrative deal with retailer Sears, Roebuck amid component shortages.

Just before Intel put aside its LCoS adventure, the giant Dutch electronics company Philips backed out of its LCoS project, saying it realised it wasn't "big enough" to bring mature products to market quickly. Japan's Toshiba also halted its LCoS plans after a supply snafu with Hitachi, Chinnock said.

What is it about LCoS that seems destined for failure and what keeps bringing companies back?

For one, the technology promises a seemingly straightforward technical solution to a problem facing the entire TV industry - how to make big, gorgeous TV displays on the cheap. It's an especially attractive idea for chip makers, since LCoS displays get better and better as the silicon components get more advanced.

And it can be done. JVC is making a big push on a mainstream LCoS set this year and Sony is using the technology in its high-end projectors.

"I think what we've found and concluded is that these were really failed approaches to the LCoS solution, which does not necessarily mean that LCoS is dead," Chinnock said.

Among those trying to turn the technology into a profitable business is MicroDisplay, based in San Pablo, California.

Gupta said LCoS could be a maddening technology to develop, with engineers fixing one problem only to uncover an even deeper flaw. There were eight technological disciplines required to make a good LCoS product, from optical expertise to software to analog chip design.

MicroDisplay says it now has the advantage. It has signed a TV customer in South Korea, Uneed Systems, with more to announce next year.

"After 22 designs, 320 man-years, a 50 per cent staff of PhDs, and US$50 million, you end up with a design that works," Gupta said.

- REUTERS