During the past few years, U.S. electronics manufacturing service (EMS) providers have seen their businesses change from high-volume, low-mix assembly to high-mix, low-volume assembly. There are many reasons for this shift, including mass customization, shrinking inventories and short product life cycles. Another reason is simple dollars and cents: High-volume, low-mix assembly is moving to China, Mexico and other low-cost labor markets. “High-volume assembly is not going to be done anymore in this country,” asserts Steve Pollock, product manager with Manncorp (Huntingdon Valley, PA). “Most of our customers are OEMs assembling military or proprietary products that will never be farmed out to China. They need to produce 10 different boards at volumes of 200 to 500 boards per month." Even with boards that aren’t "top secret," the economics of low-volume assembly often dictate that production stay at home. "You don’t gain a whole lot by moving low-volume assembly to Mexico or China. The cost efficiencies aren’t there," says Bill Coker, director of sales and marketing at Elcoteq (Irving, TX), a contract manufacturer with plants in the United States, Europe, Asia and Mexico. He should know. Before joining Elcoteq, Coker worked for another contract manufacturer with a high-volume assembly plant in Mexico. When assembling large numbers of circuit boards for computers, the plant excelled. When it landed a contract to assemble a small number for boards for medical imaging equipment, the plant struggled. "Low-volume assembly requires a different mentality," Coker says. "You could save a few bucks by moving low-volume assembly to a low-cost labor market, but the plant really needs to be set up to support that." Indeed, both EMS providers and suppliers of electronics assembly equipment are learning to set up their lines differently for high-mix, low-volume assembly. In high-mix production, EMS providers might assemble more than 100 different boards annually. These boards vary in quantity, application, size, assembly processes, components, and quality and reliability requirements. In such an environment, EMS providers face many challenges, including frequent changeovers, engineering change orders, expansive component inventories, low first-pass yields, high defect rates, price sensitivity, and capacity and resource planning. "The big challenge with high-mix assembly is how to control the increased variation in everything, from materials to setups to manufacturing," says Coker. "In high-volume manufacturing, once you’re running consistently, you can tweak the process and monitor your control limits in real time. In high-mix, low-volume manufacturing, you can’t do that. Any variation is going to impact your profit." "High-mix, low-volume manufacturers are much more bottom-line oriented," adds Aaron Saxton, product manager at Universal Instruments Corp. (Binghampton, NY). "They have to build 50 boards, and they have to be profitable on each one. They can’t afford to assemble 200 boards to get their machine or process performing at its best."
For high-volume, low-mix assembly, the emphasis is on high-speed placement. The line might have three or four chip-shooters for every one or two flexible, fine-pitch placement machines. However, with high-mix, low-volume assembly, the emphasis is on flexibility. Because a high-mix line might see seven or more product changes daily, the ratio of chip-shooters to flexible placement machines is 1-to-1. "It’s not how many parts per hour the machine can place; it’s how many jobs can be done in a day," says Pollock. To that end, suppliers have added a variety of features to their equipment to facilitate changeover, says Mark Ogden, product manager with Siemens Dematic Electronics Assembly Systems Inc. (Norcross, GA). Bar code scanners are ubiquitous on today’s placement machines. Before installing a feeder, the operator scans a bar code on the reel. The machine then automatically sets itself to handle that component and tells the operator where to install the feeder. Alternatively, a map showing the required components and their corresponding feeder slots can be printed out and laid over the feeder table. Other suppliers have added intelligence to feeders. Equipped with a microprocessor and an RS-232 serial port, a feeder can tell the placement machine what component it’s carrying and how many are left on the reel. With intelligent feeders, engineers can program the machine to run a series of jobs, and the machine will report if it has enough components to complete the assemblies. Intelligence also enables feeders to be swapped while the machine is running. Another way suppliers are helping to reduce changeover times is by increasing the feeder capacity of their equipment. Some feeders can run two or even three 8-millimeter reels from one slot. Some flexible pick-and-place machines can store 190 8-millimeter feeders. Such a high feeder capacity allows assemblers to store multiple board setups on the machine at the same time. "If you’re running high-mix, it’s to your advantage to have as many feeders as possible," says Pollock. Deciding which components to store on a machine, and in what order, can be tricky. One strategy is to set up feeders to run a family of similar products. The goal is to create a set up that accommodates all the components in the family, but also maximizes throughput and line balancing for each product. Another strategy is to group feeders by customer. This is practical when individual OEMs have their own unique sets of approved component vendors. "Rather than do a lot of changeover, some of our customers have multiple pick-and-place machines, just to get the extra feeder capacity," says Saxton. "They prefer to keep their entire component inventory online." On some machines, feeders are stored on two or more interchangeable tables. By investing in multiple tables, engineers can set up one or more tables off line while other tables are busy assembling product. This can reduce changeover times, but it can get expensive. Assemblers must balance the cost of setups against the capital expense of feeders. "Operations managers may not be comfortable with numerous feeders sitting idle," warns Coker. Placing odd-form components is another issue. In high-mix, low-volume production, EMS providers may be unable to justify automating odd-form assembly. In that case, engineers should take steps to error-proof the manual insertion process. Besides the pick-and-place equipment, the other parts of a high-mix, low-volume assembly line may differ from those in a high-volume, low-mix line. For example, in some cases, it may be more economical to apply solder paste with an automatic dispensing system instead of a printer. This saves the cost of a stencil. A stand-alone convection oven may be more cost-effective than the 7-foot, multizone reflow oven found on high-volume assembly lines. "A lot of small shops can’t even meet the power requirements of these big ovens," Pollock points out.
Compared with high-volume assembly, high-mix production is much more dependent on software for success. Given the need for frequent changeovers, programming software is critical, says Ogden. Engineers should be able to program machines offline and easily import design information, whether it be a CAD file or merely a bill of materials with centroid data. Some software can generate programming data by optically scanning the board or even individual components. Simulation software can help with setup, programming and production scheduling, says Ogden. Pick-and-place programs can be checked and corrected without having to waste a single component or board. Release schedules can be planned according to the time of the order, machine utilization, cycle times, common components and work in process. Engineers can also use simulation to optimize feeder locations and placement sequences for multiple batch assembly. Software can analyze component usage for optimal feeder setup. "Software can look at 10 products and tell you where to put your feeders to get the biggest bang for your buck," says Saxton. "That way, you can put your high runners in the front, and your variable components at the end."