If Integrated Micromachines is right, then most other vendors are using the wrong type of tilting mirrors

June 12, 2000

5 Min Read
Switch Startup Raises MEMS Questions

Some important issues concerning the use of MEMS (micro-electro-mechanical systems) in all-optical switches have been brought to light by the latest unveiling of a startup in this area.

The startup in question -- Integrated Micromachines Inc. (IMMI) http://www.micromachines.com -- went public last week with an announcement of a $15 million investment to develop switches.

However, IMMI is far from being a new player in the MEMS manufacturing business. It's been making them for pressure sensors, used in car air bags, for the past five years.

IMMI set out to develop MEMS-based switches and then got side-tracked into making pressure sensors when the telecom market didn't materialize, according to Denny K. Miu, the company's president and CEO. Now it's back doing what it originally intended, with a lot of experience of making and selling MEMS under its belt.

The company plans to offer a complete range of switches to original equipment manufacturers, from arrays of tiny 2x1 port units to 1000x1000 port monsters. The bigger switches will be assembled from 64x64 or 128x128 port subsystems, linked together

IMMI is aiming to differentiate itself from competing vendors in two particular respects -- the material used to make the MEMS, and the technology used to turn the tiny tilting mirrors.

Material Matters

IMMI uses single crystal silicon while many of its competitors use "polysilicon" -- a deposit of silicon on another material, typically an insulator.
Polysilicon also tends to make very thin mirrors, which can warp, according to Miu. In addition, it's hard to attach reflective coatings to polysilicon, which means light losses are high. In contrast, IMMI's switches will have remarkably low losses; Miu is targeting 1.5 decibels for a 64x64 port switch.

"As a mechanical material, polysilicon is a poor choice. As an optical material, it's even worse," says Miu. Other vendors, including Lucent Technologies Inc http://www.lucent.com, have reached the same conclusions and are moving to single crystal material for their mirrors, according to Miu.

Optical Micro Machines Inc (OMM) http://www.omminc.com begs to differ. It uses polysilicon to make similar range of switches to IMMI. And it says polysilicon lends itself to the automated processes it uses when manufacturing its tilting mirrors. It's also more robust.

"OMM is shipping products and generating revenues," notes Conrad Burke, OMM's senior vice president of marketing. "We've gone through rigorous testing." In contrast, IMMI is no where close to testing or shipping any of its developments. Miu expects to have samples of 2x1 switches in two months' time.

Tilting Technologies

On the technology for tilting the mirrors, IMMI uses electro-magnetic developments, similar to those used in electric motors. It delivers "1,000 to 1,500 times" more turning force than the electrostatic technologies used by most vendors, according to Miu

This has two big benefits, according to Miu. First, IMMI's mirrors can turn through a very wide angle -- 50 degrees compared to 22 degrees in developments by Calient Networks, Inc. http://www.calient.net and 10 degrees in developments by Lucent, both of which use electrostatic technology to turn their mirrors. Having a big turning angle delivers significant benefits in bigger switches, which have two two arrays of mirrors facing each other. It means the arrays can be closer together, which reduces the size of the switch and also reduces losses.

The second benefit of using electro-magnetic technology for turning the mirrors is that much more robust hinges can be used, which reduces the likelihood of damage from shocks, according to Miu. "Our structure will be 1,000 times stronger because we have 1,000 times more force," he says.

This may turn out to be vitally important. Calient says it has tested a number of MEMS designs for compliance with U.S. regulations covering resistance to earthquakes. Its mirrors stayed in place during these tests, but that could be because they are an order of magnitude smaller than the ones developed by other vendors (see Calient Claims Breakthroughs On Optical Switches).

Larger mirrors - similar to the ones used by Xros Inc. http://www.xros.com - broke and "flew off" during Calient's tests, according to John Bowers, the startup's CTO. This is because the hinges have to be very delicate so they can be turned with a weak electrostatic force.

The suggestion that Xros's MEMS may not resist shocks is rejected by Greg Reznick, its CEO. He says that its switch developments have undergone rigorous testing without any failures and its mirror hinges "are over-engineered by an order of magnitude." Nortel Networks http://www.nortelnetworks.com examined all of these issues in great depth before agreeing to acquire Xros for $3.25 billion, he adds.

The downside of using electro-magnetics for turning mirrors is overall switch size, according to Reznick. It could explain why IMMI is planning to make a 1,000 port switch with smaller size building blocks. Miu says that's not the case. IMMI is designing building blocks because carriers want scalability, he says.

Size has certainly been a big issue for switches made with MEMS from Texas Instruments http://www.ti.com, which also uses electro-magnetic technology to turn mirrors. Its MEMS were used in a 72x72 port switch developed by Astarté Fiber Networks, Inc. http://www.starswitch.com. It's said to be huge - a 2-meter sided cube.

By Peter Heywood, International Editor, Light Reading, http://www.lightreading.com

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like