If at first you don’t succeed try, try again. -- proverb

This may not be the best moment to follow this advice, but Richard Laughlin, founder of Optical Switch Corp., isn’t letting that deter him.

Laughlin's invention -- an optical switch based on a concept called frustrated total internal reflection -- withered and died after he left the company he founded. Now he claims to have come up with an improved version of the switch, which he hopes to sell or license to someone.

If at first you don't succeed, call it version 1.0 -- Web Witticism

There's just one potential sticking point: Optical Switch never got the switch to work satisfactorily, halting product development completely in late 2001 (see The Telltale Sign). The company survived by switching focus to high-resolution lithography.

Optical Switch Corp. management was a little surprised to hear about Laughlin's plan. "If we thought it [the switch] was fixable, we would have fixed it," says William K. Woodruff III, chairman and CEO of Optical Switch Corp., and provider of the seed funding that Laughlin used to start up the company in 1997.

This raises some pretty important questions. What makes Laughlin think he can fix his switch? He has not yet made a prototype to prove that his new idea will work. And even if he can fix it, will it sell, given that the market for optical switches is stagnant?

Laughlin says he has no direct knowledge of what happened to the product after he left Optical Switch back in April 1999. But he has some ideas about what might have gone wrong. "At the time I left Optical Switch, there were three known problems with the switch, and I had solutions for all of them," he claims.

Sources say that the two big problems with the FiberKey switch were excessive crosstalk and slow switching speed -- and this fits perfectly with Laughlin's recollection.

If you let failure bother you, you'll never succeed. -- J. Peterman

But before going into details, let's set the scene by explaining the principle of frustrated total internal reflection (FTIR). Cast your mind back -- no, not to last week's therapy session, but to high-school physics, which taught us that light impinging on an interface from the inside of a glass block is totally-internally reflected if it strikes the surface at an angle greater than the so-called critical angle.

Press another glass block against the interface, however, and light carries straight on as if the interface weren't there -- total internal reflection is frustrated. With one input and two possible outputs, this forms a basic 1x2 switch (see Optical Switch Comes Into the Light).

FTIR only works well, says Laughlin, if the two surfaces that must come into contact are very smooth, and free from residues of any kind. If these requirements are not met, then losses and crosstalk increase to unacceptable levels.

The first problem he was aware of was that the glass being used in the prototype switch was oxidizing. The fix for that is simple, he says: Use another kind of glass.

"We had some replacement glass sitting on the shelf," he says. "But I wasn't going to build a new switch until we had ironed out some other problems." He doesn't know if the new glass was ever used.

The second difficulty concerned the switch plate, which was a thin wedge of glass that's been polished to nearly atomic smoothness on one side. Optical Switch had to change supplier shortly after Laughlin left the company, and he questions whether the quality of parts from the new supplier was up to scratch. Quality is unlikely to be a problem these days, however, because techniques like "superpolishing" have advanced a great deal since then.

The final problem -- and this could have been the showstopper for Optical Switch Corp. -- was "dimpling" of the switch plate when it was actuated. The actuator was tubular, with the result that the edge of the glass wedge made contact with the switch before the center. To bring the wedge into contact more evenly, the switch had to be slowed down. "This problem was inherent to the design," says Laughlin. "But it has been overcome in the new design."

Laughlin's new design uses a prism instead of a thin switch plate, resulting in a basic 2x2 switch. The simplest way to explain is with the aid of a diagram. The two inputs are A and B'. When the switch is activated, the two outputs A' and B get swapped over. Simple 2x2 switches can be cascaded to make larger switches.

Laughlin has filed a patent on this design. He's also filed a second patent on a variation of it, which uses thin-film filters on the inputs of the device to turn it into an optical add/drop multiplexer (OADM).

The big advantage of this idea over other OADMs, says Laughlin, is that the express channels have an extremely low insertion loss of under 0.5dB. That's because they are reflected once by the filter on the input and do not have to travel into the center of the switch. Losses will increase, of course, if OADM modules are cascaded. But he points out that most traffic passes through, while only a few channels get added or dropped, so the number of modules required is likely to be small.

Laughlin hasn't actually made one of his new devices yet, but is confident it can be done. "The basic technology is proven," he says.

So, will it sell? Lawrence Gasman, director of optical components research at Communications Industry Researchers Inc. (CIR) thinks it might, if the switch can get to market in the next year or so.

"It's basically an optomechanical switch -- which is actually in some ways quite good news," he says. "Optomechanical switches are seen as reliable and cheap -- all of which is good right now." But in the longer term, he believes that optomechanical switches will be superceded.

"There is a short-term window. What we hear when we talk with customers is that they are very content with optomechanical for now, especially for things like protection switching. But they are definitely looking at newer technologies."

The market opportunity for OADMs may also be time-limited he says. Once companies developing OADMs commit to a particular technology, it will be hard to get them to shift, however good the alternative. The favorite technology choice for vendors of OADMs is liquid crystals from the likes of Corning Inc. (NYSE: GLW) rather than optomechanical.

A recent CIR report estimated that the market for OADMs would see substantial growth over the next few years. But the market is starting from almost zero; it is only forecast in the region of millions of dollars per year, not billions, for the foreseeable future (see Optical Add/Drop Muxes).

It is common sense to take a method and try it. If it fails, admit it frankly and try another. But above all, try something. --Franklin D. Roosevelt

— Pauline Rigby, Senior Editor, Light Reading
www.lightreading.com