Corona Claims 40-Gig First
Parallel optics like this, which use multiple channels delivered over fiber ribbons, are designed to replace the electrical connections that make up the backplanes of telecom equipment. At high data rates, electrical links cannot support the distance needed to send signals from board to board or rack to rack in high-performance telecom installations.
What's unexpected about Corona's product is that it's here so soon. The working group of the Optical Internetworking Forum (OIF), the body that sets standards for short-range optical links like these, has barely begun the process of agreeing on specifications for OC768 modules. In fact, it only finalized the standards for OC192 (10 Gbit/s) six months ago (see OIF Sets Short-Range Sonet Standard).
In fact, it may be too soon. "Even if you designed an [OC768] link, what are you going to connect it to? There's no silicon to support OC768 yet," points out Tracy Earles, director of product management for Picolight Inc., a company that also makes parallel links.
Corona denies it's too soon, and support for this view lies in the fact that it's shipping to customers. According to its president, Bryan Gregory, customers come from two camps: those that demand the OC768 capability, and those that don't care precisely what the aggregate capacity of the device is -- as long as it's a tiny transmitter that can move huge amounts of data.
So far, Corona doesn't appear to have any competitors in the OC768 market. But in applications that simply require raw bandwidth for the backplane, there are plenty of contenders. These include companies like Agilent Technologies Inc. (NYSE: A) and Picolight, both of which offer 30-Gbit/s interconnects. And then there are vendors like Blaze Network Products Inc. and TeraConnect Inc. that are developing monster interconnect products using multiple fiber ribbons to deliver bandwidths in excess of 100 Gbit/s (see Parallel Optics Boosts Bandwidth).
Corona's product stacks up well against the competition, says Gregory. It has two big advantages. For starters, its module is extremely small -- a single transmitter measures just 13x13 millimeters. Corona's competitors agree that this is quite an achievement. "It's very impressive," says Picolight's Earles.
A second advantage of the device is that it's "highly manufacturable," says Gregory. It's designed to resemble an electronic chip as much as possible, so that equipment manufacturers or their outsourcers can use automatic pick-and-place machines to plonk it on a printed circuit board, just as they do with electronic chips. Corona's modules can be supplied on a "tape and reel" -- resembling a giant reel of movie tape -- which is fed into the pick-and-place equipment.
"Typically, manufacturers place all the normal chips on the board, and then at the end of the process they take the optical module and hand-place and solder it on the board," says Gregory. "Having a human being touching, soldering the part affects your yields. What's more, if [equipment makers] are going to be doing the densities that they say they need then you truly need to automate this part of the process."
But not everyone endorses this approach. Picolight, for one, says that the best way to increase the board yield is to make devices pluggable, as it does with its 30-gig product (see Picolight Ships 32-Gig Interconnects). That way, the electronic chips can be tested before the optics are placed on the board. It also allows customers to repair or upgrade their optical backplane modules without throwing out the entire board.
Corona is based in Lombard, Ill. It raised $12.4 million in first-round funding led by Intel Capital. — Pauline Rigby, Senior Editor, Light Reading
http://www.lightreading.com Want to know more? The big cheeses of the optical networking industry will be discussing this very topic in a session at Opticon 2001, Light Reading’s annual conference, being held in San Jose, California, August 13-16. Check it out at Opticon2001.