& cplSiteName &

Polatis Reveals Switch Secrets

Light Reading
News Analysis
Light Reading
9/28/2001

For the better part of a year, Polatis Ltd. has been bragging about the low-loss and other characteristics of its all-optical switch while keeping tight-lipped about the technology it’s based on. Next Monday, however, the UK startup is promising to reveal all at the ECOC exhibition in Amsterdam.

At the show, Polatis will demonstrate elements of its 64x64 port fiber switching subsystem, one that acts like an automated patch panel, linking together input and output fibers.

The key point is that the light passing through the switch loses very little power -- less than a single decibel, according to Andrew Dames, Polatis’s founder and CEO. For an equivalent subsystem, this compares with several decibels made with arrays of tiny tilting mirrors using 3D MEMS (microelectromechanical system) technology. That’s a huge difference, bearing in mind that decibels use a logarithmic scale (see Optical Units Reference).

It just so happens, the comparison with 3D MEMS subsystems is particularly appropriate. In effect, Polatis’s switch is like a 3D MEMS switch without the mirrors, which are the major source of light loss.

Here’s the score. If you trace light through a 3D MEMS switch, it comes out of the input fiber, goes through a collimator (a lens that focuses it) and onto a mirror. The mirror is tilted so the light is steered onto a second tilting mirror which then bounces the light through a second collimator into the output fiber. Polatis has removed the tilting mirrors. Instead, it sways the collimators so they shoot light directly at each other. "It’s so simple that it feels like cheating," says Dames. It appears to work, though.

Dames acknowledges that the idea isn’t new. Bell Labs made such a switch a long time ago but the collimators were moved with chunky bits of mechanical equipment like step motors and worm screws. It measured several feet across and switching speeds were very slow.

Polatis has managed to shrink all of this into a much smaller space and reduce switching speeds to less than 10 milliseconds -- fast enough for protection applications.

An 8x8 switching subsystem would be about the size of half a video cassette, according to Dames. That’s quite a bit bigger than the equivalent 2D MEMS subsystem sold by OMM Inc., but not impossibly big. Polatis’s 64x64 switch measures 81x400x198 millimeters (about 3x15x8 inches). Other details together with pictures and performance specs are on its Datasheet.

The key to much of this is the way in which Polatis moves its collimators. According to Dames, it uses "high quality piezo actuator elements". These are multilayer ceramic devices that expand and contract when electric currents are applied to them. "They’re in a gray area that people consider solid state but it’s actually a bit of material waggling about," says Dames. They’re already used for other applications, such as in heart pacemakers and fuel injection systems in automobiles.

One of the nice things about piezo actuators is that they can be designed to meet specific reliability standards, according to Dames. In Polatis’s case, the actuators have been designed for 1,000 million cycles; the company has already run tests to prove this in practice.

Polatis designed its actuators and is getting them manufactured by specialists. All of its hardware is "very amenable to subcontracting," says Dames. "It’s allowed the company to really concentrate on the control systems from Day 1." Polatis has "a lot of people with massive experience in sensing," he adds.

As well, the use of established technologies makes it easier for system vendors to incorporate Polatis subsystems into their equipment. "They’re dealing with components they’re already familiar with, like collimators," Dames points out.

Dames says the technology could be used to make switch modules with up to 256x256 ports. Beyond that, it would be better to cascade modules, something that would be easier than normal to do because of the subsystem's low losses. On the other end of the scale, the same technology could be used in tiny devices "all the way down to single on-off devices," he says.

A couple of companies are working on similar developments to Polatis's, except that they’re waggling the fibers rather than the collimators. One of them, Creo Products Inc. (Nasdaq: CREO; Toronto: CRE) is aiming to develop a switch that could scale to 4000x4000 ports (see From Graphics to Optics). The other, Germany’s AIFOtec Fiberoptics AG, is developing very small devices that aren't really in the same market (see Startup Moves Fibers, not Mountains).

Both Creo and AIFOtec use magnetism to move their fibers, which attracted some quizzical comments on the message board following the Creo article (see Not Another Switch!!!, for example). Doug Richardson, Creo’s director of technology, points out that magnetism, in the form of electrical coils, already has a long track record of reliability in devices such as loudspeakers and CD players. He also points out that moving a whole collimator is much more challenging than moving a wisp of fiber because it's so much heavier.

Dames says the big problem with waggling fibers rather than collimators is that the end of the fiber needs to be prepared very carefully. Maintaining adequate quality control is a challenge. In contrast, Polatis uses collimators "straight out of the packet". Their quality is already well-established.

All the same, Creo appears to have a more compact switch than Polatis. Richardson says Creo's 32x32 switch would occupy a single 1U high shelf and its 1000x1000 switch would occupy 7x23x30 inches. Creo also appears to have the edge on switching speed -- less than 5 milliseconds, half that of Polatis -- although the difference might not mean that much in practice.

The biggest challenge facing Polatis is almost certainly raising more cash. "We’re in the middle of trying to complete the second round financing," says Dames, who hopes to close the round "before Christmas". Polatis announced in late May that it was trying to raise $15 million (see Polatis Kicks Off Fund Drive).

It seems likely that funds are running low. Polatis has only raised a total of £2.5 million (about $3.7 million) so far, some of it from 3i Group PLC and some of it from Alta Berkely Venture Partners (no web site). Originally, Polatis was spun out of Sentec Ltd., a technology development company also founded by Dames (see Polatis Ltd.). Polatis currently has 30 staff and has been operating since June 2000. It's also spending quite a bit on a second manufacturing center (see Polatis Opens Optical Switch Center). Dames appears confident that Polatis will find new funding, noting that the company is already making a simpler device -- a frequency switch -- for which it has a customer. An unnamed equipment vendor is designing Polatis’s frequency switch into a future product, according to Dames.

On the other hand, the market for large scale all-optical switches, like the one being developed by Polatis, appears to be further away than once thought. At least two developers of MEMS switches -- OMM and Onix Microsytems Inc. -- have already mothballed their plans to develop larger 3D products because of this (see OMM-inous News and Onix Follows in OMM's Footsteps). Not everybody agrees with this analysis, however. Some startups, like Integrated Micromachines Inc. (IMMI), are still charging ahead with 3D MEMS developments, and a new startup in this field, MEMX Inc., has only just emerged (see MEMX Starts Anew on 3D MEMS).

On the face of it, Polatis has some technology that’s less risky than 3D MEMS, so if MEMX managed to find some investors to back its ideas, maybe Polatis can too. Mind you, that’s a big maybe in the current environment.

— Peter Heywood, Founding Editor, Light Reading
http://www.lightreading.com

(4)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
mvissers
mvissers
12/5/2012 | 1:23:33 AM
re: Polatis Reveals Switch Secrets
In a transport network the loss of a complete network element is important. For an optical switching equipment, the optical fabric should typically be (equipment) protected; i.e. working fabric and protection fabric. Around these W and P fabrics you need a splitter and selector adding up to 6 dB of additional loss. The optical signals will be ingressing/egressing via connectors that worst case have 0.5 dB loss (up to 2 dB total in system).

The loss in the actual fabric device is not longer determining the loss of the network element...
iamnoone
iamnoone
12/4/2012 | 7:47:48 PM
re: Polatis Reveals Switch Secrets
I am wondering what the demand is for switches bigger than 8x8. From what I've heard from metro systems people, the sweet spot is 32x32 which I believe the 2D MEMS will be very hard pressed to achieve with decent yield. Thus I believe that even if the demand for really big switches (256x256 and above) is still a bit further out, the demand for 32x32 or 64x64 is at least as immediate as for the 8x8's. I'm guessing that OMM and Onix are mistaken in lumping any switch bigger than a 16x16 in the "big switch" category that might not be in demand for a while. The "intermediate switch" category might be the near-term winner.
hawkman
hawkman
12/4/2012 | 7:47:45 PM
re: Polatis Reveals Switch Secrets
The Metro areas need larger than 8x8, in other words they are not a long term solution. 32x32 is a good place to start for metro, but 2d mems will give a 8+ db loss for that number of ports. The people that are using 8x8s are only using them because there is nothing bigger available. 32 and 64 are good numbers, however your assumption about 256 not being in demand is incorrect. That was generated by the marketing departments of companies that failed at a 3d solution. Agere says they have one, and I am sure someone from there will respond about their 64x64, but it has yet to materialize in a usable form. I can have a foundry make a MEMS mirror array, I want to buy a system whose controls I can integrate, I dont want to have to design the whole control system.
manoflalambda
manoflalambda
12/4/2012 | 7:47:35 PM
re: Polatis Reveals Switch Secrets
Hi all,

Does the Polatis module contain any control elements within it? I had heard that the Agere module would contain its own mirror control/movement devices and some sort of micro-controller.

Also, have they quoted a max or just an ave loss of 1 dB? I had heard 1 dB alone for loss at the output collimator lens in other free space switches. I think the module has MU connectors on the outside for another 1/2 dB or so. Does loss increase between the fiber and collimator as the collimator moves?

Salute,
Manoflalambda
Featured Video
Upcoming Live Events
October 22, 2019, Los Angeles, CA
November 5, 2019, London, England
November 7, 2019, London, UK
November 14, 2019, Maritim Hotel, Berlin
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events