x

Trellis Gets $25M For Holographic Switch

Trellis Photonics Inc. is expected to announce a fourth funding round of $25 million and the first customer for its hologram-based optical switch product, the Intelligent Lambda Switch.

Trellis's first product, which will be shipping in the first quarter of 2001, will handle the equivalent of 240 by 240 ports. That may not sound very big compared to the 1000-port switches claimed by other vendors such as Calient Networks Inc. (see Calient Claims Breakthroughs On Optical Switches). But this will be the first of a family of products that will include 480x480, 960x960 and 1920x1920 switch sizes. Trellis plans a second product family for later in the year, which will scale up to 3840x3840 ports.

The technology behind the switches -- electroholography -- is so tightly wrapped up in patents that no other manufacturer could possibly make use of it, according to Tim Cahall, president and CEO of Trellis (see Trellis Touts "Electroholography").

However, Trellis will face stiff competition from companies that are developing other types of optical switching fabric, which can choose from an increasingly long list of weird technologies. Examples are tiny, movable mirrors, also known as MEMS (micro-electrical-mechanical systems) (see Startup Prepares Secret Switch ), acousto-optic filters (see Optical Switches Go Acoustic), and microscopic bubbles (see Switch Startup Raises $10 Million).

Electroholography was developed over a period of 10 years by Aharon Agranat at the Hebrew University of Jerusalem. A laser is used to write a hologram inside a crystal made of potassium lithium tantalate niobate (KLTN). Any mistakes can be overwritten with a new hologram, so production yields should be high. The hologram works like a Bragg grating in that it only reflects a specific wavelength of light. Unlike a Bragg grating, however, it is electrically energized. When voltage is applied, the crystal becomes reflective, directing the beam to an output fiber. With no voltage, all the wavelengths pass straight through.

Inside the switch, the signal from an incoming fiber is split into individual wavelengths. The KLTN crystals, which measure 2-by-2-by-1.5 millimeters each, are assembled into rows corresponding to incoming wavelengths and columns corresponding to the exit fibers. Each input fiber requires a separate matrix of rows and columns.

The result: An 8x8 switch, which has two input and four output fibers, contains 64 crystals, while a bigger 240x240 switch, with six input and six output fibers, contains 1440 crystals.

In addition to scaleability, holograms have two other benefits. First, they have the potential to switch individual wavelengths at single nanosecond speeds. Right now, Trellis's switch offers a speed of about 10ns. "The crystal hasn't been optimized for speed. When you're four to five orders of magnitude faster than the competition you don't spend much time on optimization," says Cahall.

Second, it is possible to monitor every single wavelength in the switch with no additional components. When fully energized, the crystal is about 95 percent reflective, which means that five percent travels unimpeded to the end of the row, where it can either be sent to a local test bed or amplified and brought back to a central network operations center.

Sound too good to be true? Cahall admits that the application for cross-connecting a group of wavelengths is a bit limited. "It's a task that mirrors excel at, but that we would really struggle to get to," he says.

Another potential drawback to Trellis's technology is that it requires a high voltage to switch the hologram -- the company didn't deny that it could be as high as 100 volts. But Agranat, the inventor of the technology, points out that the device doesn't dissipate any power because it is "a pure capacitive load" -- it only draws power when switching. It follows that the power level will be determined by the switching frequency, which shouldn't present a problem for circuit-switched applications. Agranat also says that Trellis is in touch with several vendors who reckon they can supply the electronics that are required.

Trellis wouldn't reveal the name of its customer at this stage, but did say that the customer was one of the investors in its latest funding round.

-- Pauline Rigby, special to Light Reading http://www.lightreading.com

GW Pearson 12/4/2012 | 9:02:04 PM
re: Trellis Gets $25M For Holographic Switch It seems that the switching component has major cross-talk problems (5% or -13 dB). Not really a problem with MEMS, I believe.
Peter Heywood 12/4/2012 | 9:01:42 PM
re: Trellis Gets $25M For Holographic Switch On Trellis's "major cross-talk problems", can you ellaborate?

In particular, could you tell us (or me on [email protected]):

- Whether you think or whether you know this is the case.

- Whether you're basing this on supposition, rumor or facts?

On the face of it, it's hard to see why folk would invest tens of millions of dollars in something that was obviously a non-starter....but I guess it could happen.

Peter Heywood
Light Reading
asherian 12/4/2012 | 8:58:44 PM
re: Trellis Gets $25M For Holographic Switch Based on the description, it seems like the version with 6 input fibers, 6 output fibers, and 40 wavelengths per fiber is actually 40 isolated 6x6 switches rather than a true 240x240.

Is this really the case?
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE