Automation strategies hold the key to manufacturing success
Current assembly procedures There are a number of challenges relating to the automation of fiber-optic component manufacturing. Most first-generation products were developed in laboratories by scientists without manufacturing backgrounds, which resulted in a focus on innovation rather than automation. In the photonics market there is a lack of standards throughout the supply chain, and there has been a limited resource pool of advanced manufacturing engineering in many fiber-optic component companies. Products have short life cycles, so manufacturing processes not only need to be automated but also must include flexible tooling and software for fast product changeover. Lastly, working with optical components and fiber has created new material handling challenges compared with other industries.
Facing these challenges, leading companies have begun using new tools and services to design the newest generation of products (figure 1). Companies that blend these tools and services into an overall strategy for high-volume automated manufacturing are likely to gain success as the fiber-optic component industry continues to mature.
Of course, automated assembly processes are not new in the compound semiconductor industry, and an interesting parallel can be drawn with the manufacture of optical modules for CD players. Like laser sources for fiber-optic applications, CD optical heads contain a laser which is packaged with a photodiode and various other optical components (although there is no fiber coupling). Using automated processes, optical CD heads are assembled in around 2.5 seconds at a cost of about $1, with yields of around 99.7%. Manual assembly of a typical telecom transmitter takes about three hours and has a yield of less than 60%. The message to fiber-optic manufacturers is that someone will find a route to low-cost, high-volume manufacturing.
Financial justification Photonics products require assembly tolerances and process control beyond the capabilities of manual labor. When technicians attempt to assemble these products they are unable to maintain consistent product quality, resulting in poor yields. This is because of the difficulties in repeatedly maximizing alignment, maintaining peak power during epoxy cure and delivering exact adhesive volumes.
For component manufacturers, the annual scrap costs due to poor yield are currently larger than the annual labor costs by an order of magnitude. For example, a photonic component that sells for $1000 might cost $500 and require two hours of assembly and test labor. A yield of 60% is typically achieved when the component is assembled by hand, which means that $200 of scrap (i.e. 40% of $500) is generated for each acceptable product. Two hours of labor might cost $30 in North America or $4 in China but either amount is dwarfed by the scrap cost.
On the other hand, automated assembly processes typically generate yields in the mid-to-high 90% range. Typical first-pass yields for a circuit board with 300-500 components mounted on it are 94-96%. In the example discussed overleaf, an increase in yield from 60% to 90% results in a scrap saving of $150 per unit. At a modest annual volume of 10,000 units, scrap savings will provide a one-year payback of $1.5 million, in both North America and China (figure 2). As yield and cost issues become more fully understood, component makers are quickly concluding that automation will be critical to their success, regardless of manufacturing location.
