A strategic approach to automating photonics manufacturing: full automation in manufacturing and assembly is the key to success in photonics industry.

Consider this opportunity: Despite the recent telecom slowdown, the long-term market for photonic devices is forecast to rebound to a 30 percent growth rate in 2002 and reach $20 billion in the next five years. As telecom companies guard their capital budgets carefully after the crash of 2001, reducing product costs is the key for any manufacturer to capture a share of this ripening market.

The challenges in cycle times, costs and yields combine to make photonic components one of the most demanding manufacturing processes in any industry. Manufacturing tolerances are excruciatingly tight, and manufacturing processes are changing quickly at best and are unproven at worst. Assembling photonic components manually can no longer support the industry's projected growth, and long-term success now demands a successful automation strategy.

State of Photonics Manufacturing

Manufacturers of photonic components face several major challenges and market demands:

* Products are still evolving from the laboratory. Most of the product design effort has been focused on innovations in performance, and very little effort has been spent on design for automation (DFA) features. In fact, many devices have been completely designed by scientists who are just taking their first products into production. Many companies also have a very limited pool of manufacturing professionals.

* Rapid advances in the core technology and performance mean photonic products have short lifecycles and are subject to a steady stream of innovations and iterative design changes. Today's reduced volumes also mean that any automation investment must handle multiple products on a single line. Manufacturing processes must be automated with flexible tooling and software for fast product updates and changeovers.

* A lack of standards exists in packaging and handling throughout the supply chain in the photonics market, which complicates the already demanding task of precisely handling flexible fiber.

* All the processes demand precision in handling, placing and inserting a nasty combination of components including flexible fiber. Bonding processes require active control of components during the cure cycle. Also, while most discrete components are manufactured in a pass/fail quality environment, higher precision in photonic component assembly means higher efficiencies and increased revenue.

So, how to begin? The following steps are necessary for successfully automating the assembly of miniature photonic and electronic components:

* Make sure the process is defined and stable.

* Focus on yield.

* Combine DFA and process simulation.

* Understand cycle time and throughput,

* Buy industrial equipment.

A Defined, Stable Process

Two manufacturing processes can be automated: The manufacturing process that mates the major components including the fiber pigtail itself; and the material handling process that delivers components and materials to the actual assembly process.

Most financial gains including high yields and reduced manufacturing costs can be achieved by automating the core process of epoxy bonding the glass fibers. During its curing, the epoxy's changing mechanical properties can draw optical components out of alignment. To prevent any misalignments, the epoxy bonding process can be fully automated with specialized tools. These tools will maintain alignment during the curings, and they also measure the bonding to prevent any stresses.

To automate the core component assembly process, begin by defining, documenting and validating the process. Consider automation issues from the outset, but develop a stable process before doing any direct automation development. Attempting to apply automation to an unstable or weak process will guarantee project delays, cost overruns and low yields. In the second stage, automation of the material handling, parts feeding and packaging will ensure a path to high-volume manufacturing and lowest absolute unit cost.

Yield Improvements

Automating a stable core assembly process delivers the highest returns, With the savings coming from yield improvements, not reduced labor costs. Consider a nominal photonic component with a direct manufacturing cost of $500, which includes two hours of assembly and test labor. Typical manual assembly yields for such a product are 60 percent, which means scrap costs will add $200 to the total cost of each good unit. However, improving the process yield to 90 percent, which is very attainable with automation, reduces total costs by $150, regardless of labor costs.

DFA and Process Simulation

Photonics represents one of the best opportunities to leverage DFA disciplines. Carefully designed assembly tools have attributes that facilitate automation by providing an analytical DFA scoring approach, which allows designers to quickly assess part and process features and assign a score that fundamentally represents a degree of difficulty. The composite automation "grade" identifies risk areas that should be addressed to ease the automation process. After the initial DFA analysis and identification of problem areas, simulation can utilize three-dimensional (3-D) computer-aided design (CAD) data for both DFA and process simulation purposes.

Simulation tools allow process and product designers to analyze the cumulative tolerance stack-up of both part and process tolerances (Figure 1). With assembly tolerances measured in nanometers, all the contributors to position error must be understood, including the device packaging, components, stages, bonding delivery system, tooling, fixturing and sensors. Collision avoidance analysis is also critical, given the costs of precision tooling, grippers and the components themselves.

Cycle Time and Throughput

Barriers to throughput are not always the obvious issues. Leveraging simulation software and services enables a thorough understanding of the drivers and dynamics of system cycle time and throughput. Individual process steps must be identified and understood in terms of how they interrelate.

For example, the search for first light is the initial step in the photonic alignment and assembly process. The process seems simple; clamp the fiber or ferrule in a chuck and move it in a predetermined search pattern until light from a stabilized laser source is sensed. While time to first light is a function of the search algorithms, the stage design can also have a major impact. Because most alignment motions are within the stroke of the actuators, operator involvement and repositioning are usually not required. Also, the long travel and low mass of the design enable high speed dithering, scanning, first light and final alignment in less than 15 seconds.

Industrial Equipment

Because so much of the technology in photonic components is coming directly from laboratories, a common mistake is to buy lab grade equipment for manufacturing. However, after all the planning, analysis, process development and design, always buy industrial manufacturing and assembly equipment.

Micropositioning stages provide a good example. Robust design means the stage should be able to take the moderate bumps and shocks that are common in the manufacturing environment and still deliver 5 nm resolution. An intrinsically compact design reduces sensitivity to vibration and improves thermal stability, so look for a compact design and small footprint. An industrial design also means easy access for recalibration, repair or replacement by field personnel and a clean interface with an industrial control.

Finally, full automation should only be designed and implemented after first implementing the semi-automated process, gathering data on the results and securing tight process controls. Use simulation and other engineering tools developed in traditional robotic material handling applications to evaluate cell layout, throughput and cycle time considerations.

Summary

The key factors for automating the assembly of miniature photonic and electronic components include:

* Design, characterize and verify the process manually.

* Use 3D simulation tools.

* Use a DFA process in product design.

* Implement automation in phases.

* Partner with equipment suppliers with standardized, flexible automation components to minimize risk and schedule and maximize redeployment.

Analysts agree that manufacturers of photonic components who successfully develop and execute full automation strategies will be the first-to-volume in their respective markets and surely become the market share leaders.

Joe Campbell is vice president of marketing for Adept Technology Inc., Livermore, CA; [e-mail].