Optical Assembly Must Catch Up

	Alan Rae, Cookson Electronics, Foxborough, Mass.

Optical assembly automation is where electronics assembly was 25 years ago, with no dominant technology or players, an extremely complex patent and IP landscape, and in real need of standardization of package form factor and package stability standards to increase yields and lower costs. In addition, optical assembly has the added complication of fragile fiber "pigtail" handling and connectorization. As a result, most assembly is done by hand, a prohibitive reliability and cost factor that is slowing the rollout of a wider optical network to the consumer.

I recently had the unique opportunity to take an operator's course in fiber optic assembly at Solectron's facility in Charlotte, N.C. I was amazed to discover some of the difficulties and complexities of optical assembly in a marketplace where, in the absence of standards, customers set different specifications even on cleanliness, and the fiber optic assemblies to be placed on the boards are designed for functionality, not manufacturability. Solectron has invested in a very comprehensive training program, which has to make its operators and engineers among the best-trained in the industry.

How did we get here? The last four years saw an unprecedented round of investment in optical networks. According to Merrill Lynch, U.S. companies laid $90B of fiber optic cable, 39 million miles . . . but only 2.6% of this fiber is currently lit. Although our hunger for bandwidth will eventually fill this long-distance fiber, a number of factors are delaying full usage. These include the slowing of e-commerce, DWDM (more channels per fiber), lack of high-speed local access, inability of carriers to gain incremental revenue, slow DSL rollout and slow "wireless Web" rollout.

The huge growth rates meant that there was little incentive for standardization or cost reduction. As a result, there are many standards but most are inappropriate. Why use telecom standards designed for 30 years' use when we know the technology will be obsolete in five?

The recent electronics industry inventory situation, where a $50,000 router could be bought for $4000 on a Web auction, is starting to lift, and we see some evidence of flattening, if not recovery. Once the business climate improves there will be a renewed need for the rack-mounted switches and routers, especially for upgrading metro systems. But here we come to the crux of the problem — service providers have to show a return on investment, and in the local environment the dense equipment use means that the chief cost is in the equipment, not the fiber. We have to lower the equipment cost.

Increasingly, the rack-mount and optoelectronic package assembly is being handled by contract manufacturing — EMS suppliers — working on tight margins to produce complex and costly devices. They also cannot afford to use hand assembly or tolerate yields below 20%, with a high proportion of the scrap being unreworkable.

How can costs be lowered? By developing best practices and standardization where appropriate. The gestation time for a standard is three to six years, far too slow for passive components, which are seeing dramatic consolidation, but appropriate for pump lasers in butterfly packages where standardization of the package (not the contents of the package where IP concerns rule supreme) can improve manufacturability both of the package and in the package.

There are several groups addressing these issues:

• NEMI (National Electronics Manufacturing Initiative) identified the concerns in its roadmap published early this year. Groups have been working and defining the key issues over the last nine months and have six key focus areas—fiber management; hand assembly; accelerated test; in-process and product test; high-frequency, large-circuit design; and package standardization. Teams are already working on areas such as cassette handling of fibers.
• IPC has initiated a set of meetings aimed at nailing down the standards needs.
• The Photonics Manufacturing Equipment Association (PMEA), whose formation was recently facilitated by SPIE, is starting a group closely allied to the IPC and NEMI programs to develop a framework to facilitate cost-effective automation of optoelectronic package assembly.

Many other groups are also looking at optoelectronics and seeing what it means for them, and we all hope that all these groups can put aside their "turf" issues and start to work together to develop meaningful guidelines and standards.