Guide to Reliability of Electrical/Electronic Equipment and Products--Manufacturing/Production Practices (part 1)

Home | Articles | Forum | Glossary | Books


Figure 1 of Section 1 presented a block diagram of the major steps involved in delivering a product with end-to-end reliability (i.e., from customer requirements and expectations to customer satisfaction). This section addresses the details of the block labeled manufacturing/production. Figure 1 of this section expands this block to present some of the key elements that constitute the manufacturing/ production of electronic products prior to delivery to a customer.

1. PRINTED WIRING ASSEMBLY MANUFACTURING

1.1 Introduction

Manufacturing/production of electronic products by its nature involves the assembly of components on a printed circuit board (PCB). The entire assembly is called a printed wiring assembly (PWA). In the electronics industry printed wiring assembly, Printed Card Assembly (PCA), and circuit card assembly (CCA) are used interchangeably to describe a fully populated and soldered PCB.

In the fast-paced and cost-sensitive world of electronic manufacturing major changes continually take place. Original equipment manufacturers (OEMs) no longer view manufacturing as a core competency or a competitive advantage.

They want to focus on technology, product innovation, brand marketing, and sales. So they divested themselves of their manufacturing facilities, selling them to contract manufacturers (CMs) and established long-term relationships with them. [Contract manufacturers are also referred to as electronic manufacturing service ( EMS) providers and contract electronic manufacturers (CEMs). I use these terms interchangeably.] As a result, OEMs have become increasingly de pendent on contract manufacturers for some or all of their manufacturing requirements, especially PWA services.

Computer makers were the first to adopt an outsourcing manufacturing model, followed by both telecommunication equipment and wireless (cell phone) manufacturers, nimbly reacting to instantaneous changes in the marketplace. It is no longer unusual for a half dozen different brand-name computers to come off a single assembly line. (However, such products are not identical. Each OEM designs products with unique characteristics and specifications, while the EMS provider keeps customer jobs strictly separate to avoid conflicts of interest and other difficulties.)


FIGURE 1 Electronic equipment manufacturing tasks.

1.2 Outsource Manufacturing

When considering outsourcing as a manufacturing strategy, it is important for a company to understand what is driving its need for outsourcing. Is it simply cost? Is it that management feels distracted by having to deal with many functions and activities it feels are ancillary to its true mission? Is it a strategic decision to not maintain a rapidly changing technology that, when closely examined, is associated with a process that can be deemed as a support or management process?

Outsourcing originally began with the OEMs' need to manage the manufacturing peaks and valleys resulting from volatile, often unpredictable, sales volumes. In order to perform their own manufacturing, OEMs had to face three difficult choices:

1. Maintain sufficient staff to deliver product at sales peaks, knowing that workload would later drop along with volume.

2. Staff at some compromise level, carefully scheduling and distributing tasks to accommodate peak loads.

3. Hire staff at peak times and lay them off when sales drop.

Obviously, each of these options presented drawbacks. Contract manufacturers provided a solution: they offered to handle these peaks with their production capacity, and as OEMs gained more experience in dealing with them, routine manufacturing also shifted to EMS companies.

Since manufacturing operations are notoriously expensive to build and maintain, OEMs initially got out of manufacturing to reduce costs (reduce fixed assets and people). The inherent unpredictable market demand, sales volumes, product profitability, and resulting financial returns made an OEM's manufacturing operations exceedingly difficult to manage and maintain and thus led many OEMs to outsource that function. They needed to stop a hemorrhaging bottom line because they couldn't control their own processes; they refused or were un able to address deficiencies; they wanted to get rid of a headache; or they just jumped to outsourcing because everyone else was doing so. What initially weren't well thought out were the ramifications of and support required to manage their outsourcing decision/strategy. They just did it. Many companies eliminated assembly line workers and purchasing personnel in a carte blanche manner when they implemented outsource manufacturing, thinking they no longer needed them.

The ramifications included lost critical skills, negative impact on morale of both affected and remaining employees, learning curve loss, inability to both innovate and provide flexibility and fast response to customers, and loss of control of manufacturing quality to a third party. However, the OEMs painfully learned that more (not less) skilled and technically competent workers were required to sup port the outsource strategy.

The CMs' focus on manufacturing allows them to achieve the lowest product costs and high product quality. The OEMs take advantage of the CMs' strengths to stay competitive in the electronics industry without most of the manufacturing overhead costs. The use of EMS providers adds enormous flexibility to an OEM's arsenal of available tools. Because they make a broad range of products for an equally varied group of customers, EMS providers have accumulated a wider array of knowledge, experience, and expertise than their OEM customers. As a result, EMS providers often suggest other manufacturing (including test) strategies and tactics to improve product performance, manufacturability, quality, reliability, and cost.

Moreover, variations in throughput requirements, product changes, and product mix generally present less of a challenge to an EMS provider than to an OEM. Many products go through regular manufacturing cycles. In the automobile industry, for example, test development and product requirements peak at the beginning of each model year. Others, including computers and cell phones, experience a volume bump during the holiday season. The EMS providers smooth out their own production schedules and optimize their equipment utilization by carefully managing the needs of individual customers and market segments so that their peaks seldom coincide, leveraging the advantage of year-round full production. This is because the demands of both the aggregate customer base and different market segment needs tend to provide an averaging function over a large number of manufacturing opportunities.

==========

Table (coming soon) 1 Benefits of Outsourcing Manufacturing

Increased focus on core competencies Improved manufacturability and operating efficiencies resulting in reduced manufacturing costs Improved quality Faster time to market and time to technology Increased flexibility and ability to respond more quickly to market changes Improved competitiveness Reduced capital investment Increased revenue per employee Latest technology availability/implementation (manufacturing and test tools) More choices of manufacturing and test strategies Design for manufacture strategies based on large volume of many different designs from different OEMs Reduced and/or spread of manufacturing risk Better assurance of supply in dynamic and changing market conditions

==========

Table (coming soon) 2 Challenges of Outsourcing

Greater exposure and risk when problems occur.

Lack of control of manufacturing process and activities.

Loss of manufacturing expertise and technology knowledge.

Little or no headcount reduction. One of the proposed early advantages of outsourcing PWAs was to reduce personnel headcount since it was thought that the manufacturing support functions were no longer required with an outsourcing strategy. This was found to be false. Various OEM manufacturing, components, and business experts are needed to support, backfill, and validate CM issues. Just as much, if not more, support is required to conduct and manage an effective PWA outsourcing strategy.

Greater coordination required with CM.

Inaccurate and changing build forecasts require much handholding.

A dedicated champion and central interface (point of contact) at the OEM is required to manage the outsourcing activity to prevent mixed messages to the CM.

Long distances for travel and communication.

Additional time required for problem solving.

Ability of CM to perform failure analysis and drive corrective action.

Dealing with proprietary information.

Need for appropriate support (normally not considered when outsourcing):

Manufacturing engineering PWA test engineering Component engineering Failure analysis and problem resolution

==========

Table (coming soon) 1 summarizes the benefits of OEMs' outsourcing their manufacturing requirements. It is important to understand that once established, OEMs hardly ever dismantle these relationships.

Despite its growing popularity, outsourcing presents its own challenges, as Table 2 illustrates. Working with EMS providers requires tighter management practices over the entire manufacturing operation as well as better process documentation. Both are imperative to ensure that what the OEM wants and what the EMS provider builds coincide. Although many engineers regard these requirements with disdain, tight management can minimize the likelihood of design errors and ensure that production processes remain in control, thereby maximizing product quality and functionality and increasing chances for market success.

Outsourcing creates a new and complex relationship system in the supply management chain. In the case of printed wiring assemblies, the quadrangle shown in Figure 2 best describes the relationship between OEMs and CMs. The insertion of the CM and the component supplier's authorized distributor between the OEM and component supplier serves to both separate the supplier and OEM and complicate the supply chain. This figure lists some of the key functions of each link in the supply chain and indicates the changing nature of these functions.


FIGURE 2 Electronic equipment manufacturing relational structure and dependencies.

The Changing Role of the Contract Manufacturer

Early on (late 1980s and early 1990s) the sole service provided by the CM was consigned assembly services; the OEM provided the components and the bare PCB, while the EMS provider performed the assembly operation. As time passed, more and more value-added services have been added to the EMS provider's portfolio. These include component engineering [component and supplier selection and qualification, managing the approved vendors list (AVL), alternative sourcing issue resolution, etc.]; test development and fixturing; material procurement; and most recently full product design, integration, and assembly services as well as shipment and distribution of the product to the end customer. Component costs are reduced because OEMs benefit from the purchasing power, advanced processes, and manufacturing technology of the CM.

Today, many OEMs perform little or no manufacturing of their own. Organizations are using outsourcing to fundamentally change every part of their business because no company can hope to out-innovate all the competitors, potential competitors, suppliers, and external knowledge sources in its marketplace world wide. Companies are focusing on their core competencies-what they are best in the world at-and then acquiring everything else through a strategic relation ship with a CM, in which the "everything else" is the outsource service provider's core competencies. Some have referred to the CM as "infrastructure for rent." With this shift in functions, the OEM looks like a virtual corporation focusing on overall system design, marketing and sales, and managing the outsource service providers to ensure delivery of the product to the customer when requested. The EMS provider becomes a virtual extension of the OEM's business and looks more and more like the vertically integrated OEM of old. The trend toward the "virtual corporation"-in which different parts of the process are handled by different legal entities-will continue to accelerate.

The role of the CM is becoming more critical in the global electronics manufacturing sector. Currently (2002), the majority of outsourcing is given to EMS companies that provide more value-added services rather than single-function assembly houses. High-tech product development requires many different technical disciplines of expertise (industrial design, mechanical and thermal de sign, electronic circuit design, specification and documentation, test development, and turnkey manufacturing). For electronic equipment manufacturers, the many new technologies becoming available have caused them to consider outsourcing new product development and new product introduction (NPI), requiring a major adjustment in the strategic thinking and structure of modern companies.

One of the most important ways for large, complex-system OEMs to control product cost is to outsource systems build. The trend for OEMs to use CMs to build complex, multi-technology systems is accelerating. At the system level, integrating these subsystems into a complete functional system or larger subsystem presents a significant challenge. Until recently, complex systems have required engineering skills beyond the scope of most CMs, which is why these systems have remained integral to an OEM's engineering and manufacturing staffs. But this has changed as CMs have acquired these skills. A representative from a major aerospace corporation stated that it is now more cost effective to purchase completed systems than it is to purchase components and assemble them in-house.

The contract manufacturing services industry is a high-growth dynamic one. The services provided by contract manufacturers can be broadly segmented into PCB assembly, box or system build, and virtual corporation. The traditional role of the CM (PCB assembly) is decreasing as the scope of responsibilities and services offered increases, with more emphasis being placed on box build and system design. A more detailed description of each of these contract manufacturing services is now presented:

PCB assembly. The CM assembles the components on the PCB to create a PWA. The PWA is tested for opens/shorts and to verify proper component placement. Functional testing of the PWA may be included. The OEM retains PWA design responsibility.

Box or system build. This includes the PCB assembly as well as system integration to create the final product. The CM will functionally test the final product or "box" and ship it to the OEM or the OEM's customer.

The OEM retains product design responsibility.

Virtual corporation. The CM designs the product, assembles the PCB and builds the box. This is a turnkey relationship in which the CM assumes design responsibility for the product.

The trend depicting the changing nature of CM services provided is shown in Figure 3.


FIGURE 3 Evolution of contract manufacturing services.

Practical Aspects of the OEM-EMS Relationship

With ever-increasing competitive pressure on time to market, functionality, cost, and quality, it is essential that the OEM relationship with its CM is built on a solid foundation that includes organization, strategy, systems, and supplier management as its cornerstones. For an OEM-EMS provider relationship to be effective, it should be a win-win relationship that supports key corporate objectives of both partners. At a minimum, these include the improvement of customer loyalty; optimization of return on assets; and continuing growth of revenues, profits, and corporate value. Let's look at some of the practical issues involved in an OEM-EMS provider relationship that expands on the summaries of Tables 1 and 2.

Loss of Control. For the OEM, the primary drawback to outsourcing is loss of control over the manufacturing process. Geographical distances affect delivery schedules for both inventory and final product, resulting in higher transportation costs than with in-house operations. Calling in a product designer or manufacturing engineer to solve a production problem generally is slower and more expensive when the process involves an EMS provider. Situations in which the EMS provider's manufacturing facilities are far from the OEM's design team underscores the need for local control by the provider. Manufacturing and test strategy engineering and modification also must occur locally because manufacturing departments cannot afford to wait 2 days or longer for answers to engineering problems. Time zone distances aggravate product shipment delays.

OEM Proprietary Information. Out of necessity, an EMS provider will learn intimate details of product design and construction. Revealing such proprietary information makes many OEMs uncomfortable, especially when the same provider is typically working for competitors as well (no OEM puts all of its eggs in one CM basket). This situation has become increasingly common and offers no easy solution. The OEMs must select EMS providers that have a reputation for honesty and integrity. This requires a leap of faith that the provider's reputation, potential loss of business, or (as a last resort) threat of legal action will sufficiently protect the OEM's interests.

Quality. Probably the largest concern with outsourcing is that of quality.

The continuing exodus to outsource manufacturing can be a source of quality issues. When making a transition from in-house manufacturing to outsourcing, some things may be omitted. In an environment of rapid technological change, the complex OEM-CM relationship raises a number of valid questions that must be addressed: Will the quality level of the OEM's product prior to the outsource decision be maintained or improved by the outsource provider? Are corners being cut on quality? How much support will the OEM need to provide? What happens and who is responsible for resolving a component issue identified by the OEM or end customer in a timely manner? Who is responsible for conducting a risk assessment to determine the exposure in the field and if a given product should be shipped, reworked, retrofitted, or redesigned? When a low price is obtained for a computer motherboard from an offshore supplier, what is the quality level? When the customer initially turns a product on, does it fail or will it really run for 5 years defect-free as intended? If a CM does a system build, who is responsible for resolving all customer issues/problems? Obviously, it is imperative for OEMs to work closely with contract manufacturers to ensure that relevant items are not omitted and that corners are not cut concerning quality.

The key to managing supplier quality when an OEM outsources manufacturing is to integrate the outsource provider into the OEM's design team. Even before a product is designed, the supplier has to be involved in the design and specifications phase. Quality can best be enhanced by connecting the EMS provider to the OEM's design centers and to its customers' design centers before the supplier starts producing those parts. One can't decree that a product should have a certain number of hours of reliability. Quality and reliability can't be screened in; they have to be designed in.

Today's OEMs are emphasizing the need for robust quality systems at their CMs. They must maintain a constant high level of quality focus, whether the work is done internally or externally, and be actively involved with the quality of all CM processes. The reason for this is that the OEM's quality reputation is often in the hands of the CM since the product may never come to the OEM's factory floor but be shipped directly from the CM to the customer. This requires that a given CM has a process in place to manage, among other things, change in a multitude of production locations. Outsourcing requires extra care to make sure the quality process runs well. Technical requirements, specifications, performance levels, and service levels must be carefully defined, and continuous communication with the CM must be the order of the day. Outside suppliers must understand what is acceptable and what is not. Quality problems are often experienced because an outsource supplier works with one set of rules while the OEM works with another. Both have to be on the same page. The big mistake some OEMs make is to assume that by giving a contract manufacturer a specification they can wash their hands of the responsibility for quality. To the contrary, the OEM must continuously monitor the quality performance of the CM.

Communication. An OEM-CM outsource manufacturing and service structure requires a delicate balancing act with focus on open, accurate, and continuous communication and clearly defined lines of responsibility between par ties. (See preceding section regarding quality.) OEM Support. Effective outsourcing of any task or function typically re quires much more support than originally envisioned by the product manufacturer. In many companies the required level of support is unknown before the fact. Many companies jumped into outsourcing PWA manufacturing with both feet by selling off their internal PWA manufacturing facilities, including employees, CMs without fully understanding the internal level of support that is required for a cost-effective and efficient partnership. The product manufacturers felt that they could simply reduce their purchasing, manufacturing engineering, and manufacturing production staffs and turn over all those activities and employees associated with printed wire assembly to the outsource provider. Wrong! In reality it has been found that outsourcing does not eliminate all manufacturing costs. In fact the OEM has to maintain a competent and extensive technical and business staff (equal to or greater than that prior to the outsourcing decision) to support the outsource contract manufacturer and deal with any issues that arise.

More resources with various specialized and necessary skill sets (commodity purchasing specialists, component engineering, failure analysis, manufacturing engineering, etc.) are required at the OEM site than were often considered. Originally, one reason for outsourcing was to eliminate the internal infrastructure. Now a more focused internal support infrastructure is required. Unfortunately, without a significant manufacturing operation of their own, OEMs may find a dwindling supply of people with the necessary skills to perform this function, and they often compete for them with the EMS providers.

Most large OEMs place no more than 20% of their outsource manufacturing needs with a single EMS provider. This means that OEMs need to have the personnel and support structure in place to deal with at least five EMS providers, all with different infrastructures and processes.

Selecting a CM. Selecting a CM requires as much diligence as any other large investment because all CMs are not alike. Just as there are tiers or categories of competent suppliers, so are there similar distinct tiers of CMs, each with different base characteristics and core competencies. A given CM's manufacturing processes, technologies, and infrastructure must be synergistic with the needs of the OEM's products and its customers. The right EMS partner can provide a smooth, efficient, and problem-free manufacturing process that produces a high quality, profitable product. The wrong choice can hurt even a good product's quality and financial performance.

Choosing the right EMS partner depends a great deal on the nature of the OEM's product. Some EMS providers concentrate on reducing costs. They specialize in high-volume, low-mix, low-margin products such as cell phones and personal computers. Their strength lies in reducing cost and product-to-product variation. These CMs expend their resources developing efficient manufacturing and test strategies (creating, debugging, and optimizing test programs), preferring to concentrate their efforts on programs and strategies rather than on the products themselves. The automobile industry, for example, prefers this approach. Most automakers would rather spend extra time and money optimizing fixtures, pro grams, and developing self-tests than incur an increase in component cost.

Other EMS providers emphasize achieving high quality and reliability. For example, some complex, low-volume products offer higher margins and require higher quality, making them less sensitive to manufacturing costs. Typically, product failure consequences demand that defective products never reach custom ers. Manufacturing pacemakers, for example, is not inherently more difficult than making cell phones or PCs. But test program development costs, as well as product upgrades and other changes, must be amortized over a much smaller product volume. While PC and cell phone failure may cause user dissatisfaction and annoyance, even incur warranty costs, pacemaker failure can cause user death. The EMS providers address these customer requirements as an extension of the OEM's own resources, often serving as partners during the design stages and early stages of manufacture. The link between the OEM and EMS provider is tighter here than in high-volume/low-mix cases.

So there are several perspectives in selecting a CM: high-volume, low mix production versus low-volume high-mix and large- versus small-sized EMS providers. At one extreme, an EMS provider may populate boards from OEM supplied bare PCBs and parts. The provider may perform some kind of manufacturing defects test (such as in-circuit test or x-ray inspection). The OEM receives finished boards, subjects them to environmental stress screening and functional testing as appropriate, then assembles, tests, and packages the systems for shipment. Sometimes several EMS providers manufacture different boards for large systems. Here, the OEM retains control over the manufacturing and test process, including the methods employed at each step.

At the other extreme, an OEM may provide the CM with only design and functional specifications, leaving every aspect of the production process, deciding strategies, and tactics to the EMS provider. Many EMS providers offer extensive engineering services extending far beyond buying bare boards, parts, chassis, and performing assembly operations. These providers can essentially reengineer boards and systems to enhance their manufacturability, testability, quality, and reliability. For example, a board measuring approximately 24 in. on a side that boasts 26% node access for an in-circuit test (ICT) may defy conventional at tempts to achieve acceptable yields. Sometimes the solution requires changing strategy, e.g., supplementing ICT with x-ray inspection. Then too EMS provider's engineers may be able to redesign the board to provide additional access, add self-test coverage to supplement conventional test, and thus make the board more "manufacturable" and testable. Where reengineering is appropriate, an EMS provider that specializes in such activities generally will produce higher-quality, higher-reliability products at lower cost. The EMS provider may even box systems up and ship them directly to distributors or end users. This arrangement requires that the provider share engineering responsibility with the OEM, pooling expertise to best benefit both parties. The OEM gets a better-engineered product for less, thereby increasing profit margins.

A large EMS company generally encounters a wider variety of manufacturing and test strategies and philosophies than a smaller company. Bringing an unusual situation to a small EMS provider may not lead to the best solution simply because of a lack of sufficient experience with similar problems. Yet the wider experience cuts both ways. Rather than evaluate each situation on its own merits, the large provider may attempt to fit every problem presented into one of the alternatives in a menu of suggested strategies. The OEM must always ensure that the EMS provider's strategic decisions address the constraints and parameters in the best possible way.

Conversely, a small OEM generally does not want to engage a CM so large that the amount of business generated gets lost in a sea of much larger projects.

A small company often can get much more personal attention from a small EMS provider (subject to the same size caveat as before). Rather than impose a "best" solution, the small contractor may be more willing to design a manufacturing and test strategy in cooperation with the OEM. Smaller CMs are generally more flexible than their larger counterparts. Although a small CM may have fewer resources and less experience, if the CM's skills match the OEM's needs, the result should be a strong and mutually beneficial relationship.

Between these points lies a continuum of alternatives. Deciding where along that continuum an OEM should "drop anchor" represents the crux of the CM selection and planning process.

EMS Resources Required. It is important that EMS providers have the right and sufficient resources to meet the OEM's design, manufacturing, and support needs. A problem that the EMS provider can resolve reduces the engineering load on the OEM, and the time saved reduces the product's time to market. Striking the right balance is needed since EMS services vary widely.

Design-Related Issues

Three issues pertaining to design have come to the forefront. First, as networking and telecommunications equipment manufacturers pare their workforces and sell their manufacturing operations to EMS providers, they are becoming more dependent on IC suppliers for help in designing their systems, the complexities of which are becoming difficult for contract manufacturers and distributors to handle effectively. Thus, IC designers will need to have a systems background.

Second, three levels of EMS design involvement that can affect design for manufacturability have been identified, going from almost none to complete involvement. These are (1) OEMs either give the company a fully developed product; (2) look for EMS engineering support in the middle of product development at or about the prototype phase; or (3) engage with the OEM at the conceptual design phase. Contract manufacturers are moving toward or already providing new product introduction (NPI) services that help get high-volume projects underway more quickly. The sooner the customer and CM can get together on a new project, the sooner such issues as design for manufacturability and design for test can be resolved. The closer the CM can get to product inception, the quicker the ramp to volume or time to market and the lower the overall product cost.

The third issue is that significant delays are being experienced by OEMs in their product development efforts. According to a report published by AMR Research Inc. ( Boston, MA) in May 2001, product design times for OEMs in PC, peripherals, and telecommunications markets have increased by an average of 20% as a result of the growing outsourcing trend. The report identifies two troubling issues: the OEM design process is being frustrated by an increase in the number of participants collaborating in product development, and OEMs are losing their ability to design products that can be easily transferred to manufacturing.

In their quest for perfection, OEM design engineers create a design bill of materials that has to be translated into a manufacturing bill of materials at the EMS level because design engineers aren't familiar with how the EMS providers produce the product. They've lost the connection with manufacturing they had when manufacturing was internal. OEMs are designing their products one way, and EMS providers preparing the product for manufacturing have to rewrite (the design) to be compatible with the manufacturing process. The disconnect is in the translation time.

Because OEMs in many cases have relinquished production of printed wiring assemblies and other system-level components, their design teams have begun to move further away from the manufacturing process. When the OEMs did the manufacturing, their design engineers would talk to the manufacturing personnel and look at the process. Since the OEMs don't do manufacturing anymore, they can no longer do that. The result is increased effort between OEM design and EMS manufacturing. Time is being wasted both because the design has to be redone and then changes have to be made to make the product more manufacturable by the EMS provider.

The EMS Provider's Viewpoint

An OEM looking for the right EMS partner must recognize that the provider's manufacturing challenges differ from those the OEM would experience. First, the provider generally enjoys little or no control over the design itself (although, as mentioned, this is changing). Layout changes, adding or enhancing self-tests, etc. will not compensate for a design that is inherently difficult to manufacture in the required quantities. An experienced EMS provider will work with a customer during design to encourage manufacturability and testability and discourage the reverse, but ultimate design decisions rest with the company whose name goes on the product.

One primary concern of the CM in accepting a contract is to keep costs down and avoid unpleasant surprises that can make thin profit margins vanish completely. The OEMs need to be sensitive to this situation, remembering that the object is for everyone in the relationship to come out ahead. Keeping costs low requires flexibility. For the EMS provider, that might mean balancing manufacturing lines to accommodate volume requirements that may change without warning. A board assembled and tested on line A on one day may be assigned to a different line the next day because line A is running another product. Similarly, volume ramp-ups may demand an increase in the number of lines manufacturing a particular product. Achieving this level of flexibility with the least amount of pain requires that manufacturing plans, test fixtures, and test programs are identical and perform identically from one line to the next. Also, one make and model of a piece of manufacturing or test equipment on one manufacturing line must behave identically to the same make and model on another manufacturing line (repeatability and reproducibility), which is not always the case.

The CMs depend on OEMs for products to maintain high volumes on their manufacturing lines to maximize capacity and lower the overhead associated with maintaining state-of-the-art equipment that may not always be fully loaded (utilized). A contract manufacturer may send production from one geographical location to another for many reasons. Tax benefits and import restrictions such as local-content requirements may encourage relocating part or all of a manufacturing operation elsewhere in the world. The EMS provider may relocate manufacturing just to reduce costs. Shipping distances and other logistics may make spreading production over several sites in remote locations more attractive as well. Again, seamless strategy transfer from one place to another will reduce each location's startup time and costs. To offset the geographical time differences, crisp and open communications between the EMS provider and the OEM are required when problems and/or questions arise.

The first 6 to 12 months in a relationship with an EMS provider are critical.

That period establishes procedures and defines work habits and communication paths. Planners/schedulers should allow for longer turnaround times to implement product and process changes than with in-house projects.

The Bottom Line

Connecting outsourcing to an OEM's business strategy, selecting the right opportunities and partners, and then supporting those relationships with a system de signed to manage the risk and opportunities are the essential success factors for implementing an outsourcing business model. For an OEM, specific needs and technical expertise requirements must be evaluated to select an appropriate EMS provider. The selected CM's manufacturing and test strategy must match the OEM's. Identifying the right partner requires that a satisfactory tradeoff among quality, cost, and delivery factors be found. If the product reach is global, the EMS provider should have a worldwide presence as well. Because the product could ship from manufacturing lines anywhere in the world, the EMS provider should strive for consistency and uniformity in the choice of manufacturing and test equipment, thus minimizing the variation of produced product and the effort and cost of that wide implementation and distribution. In the final analysis, there are no canned solutions. One size cannot fit all.

Reference 1 discusses the process for selecting contract manufacturers, CM selection team responsibilities, qualifying CMs, integrating CMs into the OEM process flow, material and component supplier responsibilities and issues, and selecting a manufacturing strategy.

2. PRINTED WIRING ASSEMBLY TESTING

2.1 Introduction

Defects are inherent to the processes used in creating the products we make.

Virtually no step in these processes has 100% yield-it just isn't possible. So effective manufacturing becomes an engineering problem in choosing how to minimize the conflicts and thus the potential defects. As a result, performing effective testing to identify and correct defects is as important to the manufacturing process as buying quality materials and installing the right equipment.

In fact, the manufacturing process has become part of what is considered as traditional test. Test is thus comprised of the cumulative results of a process that includes bare board (PCB) test; automated optical inspection (AOI); x-ray, flying probe, manufacturing defect analyzer (MDA), ICT, and functional test solutions. Electrical testing is an important part of manufacturing because visual inspection is not sufficient to ensure a god PWA.

============

Table (coming soon) 3 Typical Defect Spectrum

Process defects Missing components Wrong component orientation Wrong component value (resistors, capacitors) Shorts (process or PCB) Opens (process or PCB) Missing screws Solder issues (missing solder, insufficient solder, solder balls, insufficient heal)

Wrong label Misalignment Wrong firmware Component defects Defective components Part interaction defects (combined tolerance issues, timing faults) Design and tolerance issues

============

The electrical design, the physical design, and the documentation of all impact testability. The types of defects identified at test vary from product, de pending on the manufacturing line configuration and the types of component packages used. The difficult part of testing is the accumulation and processing of and action taken on defect data identified through the manufacturing test processes. The manufacturing defect spectrum for a given manufacturing process is a result of the specific limits of that manufacturing process. As such, what is an actionable item is ultimately process determined and varies by line design. A typical PWA manufacturing defect spectrum is shown in Table 3.

Opens and shorts are the most predominant failure mechanisms for PWAs.

There is a difference in the failure mechanisms encountered between through hole technology (THT) and surface-mount technology (SMT). Practically, it is difficult to create an open connection in THT unless the process flow is not reaching part of the assembly, there are contaminants in the assembly, or a lead on the part is bent during the insertion process. Otherwise the connection to the part is good. The bigger problem is that solder will bridge from one component lead to the next to create a short. When soldering using SMT, the most significant problem is likely to be an open connection. This typically results from insufficient reflow. Since it is more difficult for manufacturers to catch (detect) opens with ATE, some tune their assembly processes toward using additional solder to make the manufacturing process lean more toward shorts than it would if left in the neutral state.

Testing SMT or THT will find all of the shorts unless SMT doesn't allow full nodal coverage. Either technology allows electrical testing to find most opens.

For example, when testing a resistor or capacitor, if there is an open, it will be detected when the part is measured. The issue is finding open connections in ICs.

For the typical test, a mapping is made of the diode junctions present between IC pins and the power/ground rails of the IC under test. Normally this works fine, but in many cases an IC is connected to both ends of the trace. This looks like two parallel diodes to the test system. If one is missing due to an open connection in the IC, the tester will miss it since it will measure the other diode that is in parallel, not knowing the difference.

Every test failure at any deterministic process step must have data analyzed to determine the level of existing process control. Data can be compared to the manufacturing line's maximum capability or line calibration factor to achieve a relative measure of line performance quality. The result is a closed loop data system capable of reporting the manufacturing quality monitored by the various test systems. In general, it has been observed that 80% of the defects encountered at the various test stages are process related. Looking at the defect spectrum of Table 3 it is seen that the nature of the defects has not changed with technology.

Rather it is the size of the defect spectrum that has changed.

2.2 Visual Inspection

After a printed circuit board has been tested and assembled (i.e., all components soldered) the assembly is first subjected to a visual inspection to look for gross manufacturing defects. These include defects in solder joints, assembly defects involving wrong or missing components, and solder defects under devices such as ball grid arrays (BGAs) via x-ray. Visual examination is efficient and cost effective in detecting manufacturing defects and alleviates the burden of a more expensive electrical test and time-consuming diagnostics. The objective is to find the defects at the lowest cost point.

Visual inspection employs both manual and automated techniques such as optical magnifiers, optical comparators, closed-circuit television (CCTV) magnified display, automated optical inspection, and x-ray. There is an increased trend toward the use of x-ray inspection using x-ray-based machine vision systems, which provide pin-level diagnostics. The reasons for this are

1. The shrinking sizes of passive components and finer linewidths of PCBs result in much denser PWAs than in the past. Thus, solder joints are much more critical as they become smaller and closer together, making human inspection more difficult and inaccurate.

2. The increased use of chip scale and ball grid array packages in which soldered connections around the periphery of an IC package are re placed by an underside array of solder balls that are used for electrical connections. The result is that the solder joint connections are hidden from view underneath the package and cannot be inspected by humans or conventional optical inspection or machine vision systems.

3. Densely populated PWAs are so complex that there is no way to access enough test nodes. While board density makes physical access difficult, electrical design considerations (radiofrequency shielding, for example) may make probing extremely difficult if not impossible. X-ray inspection provides an excellent solution for inspecting solder joints where limited access or board topography makes it impossible to probe.

4. A functional failure (via functional test) only identifies a segment of problem circuitry and the diagnostics are not robust.


FIGURE 4 Typical electrical tests conducted during PWA manufacturing.

2.3 Electrical Testing

Historically the first microview of the quality and predictability of the manufacturing process has been at PWA electrical test (either ICT or functional test).

Even though a PWA may undergo a rigorous visual inspection, the only way to ensure that all components are functional, contain no electrical shorts or opens, and that the circuitry performs as designed is by conducting an electrical test.

Electrical test at the PWA, module, and system levels exists for the following reasons:

Verify that the PWA, module, and system operates as designed

Continuously improve the manufacturing process and thus the quality

Reduce costs

Ensure the level of quality and reliability demanded by the customer is met Board test can be divided into two phases: manufacturing process test and functional test. Manufacturing process test verifies that the PWA has been assembled correctly by checking for correct component interconnection and component orientation. The PWA is tested for opens and shorts as well as resistance, capacitance, inductance, diode junctions, and active part operation. Manufacturing defect analyzers and in-circuit testers are used to perform this step. The MDA tester, often called an analog ICT, assumes that the components are good and endeavors to find how they may be improperly installed on the PCB. In-circuit testing takes testing one step further by checking ICs in the circuit for operation. Going from MDA to full ICT improves the test coverage by a few percent, but at multiple times tester and test fixture development cost.

Functional test verifies that the PWA operates as designed. Does it perform the tasks (function) that it was designed for? Figure 4 is a block diagram of the electrical tests performed during manufacturing and the sequence in which they are performed, from simple and least expensive (visual inspection, not shown) to complex and most expensive (functional test).

In the 1960s and 1970s PWAs were simple in nature and edge connectors provided easy test access using functional test vectors. In the 1980s, increased PWA density led to the use of ICT or bed-of-nails testing to mechanically access hard-to-get-at nodes. In the 1990s, due to continually increasing PWA density and complexity, boundary scan, built-in self-test (BIST), vectorless tests, and unpowered opens testing gained increased importance with the decreasing use of ICT.

Table (coming soon) 4 summarizes and compares some of the key features of the various electrical test methods being used. These are further discussed in the following sections.

=============

Table (coming soon) 4 Comparison of Test Methods

Self test

Limited defect coverage Limited diagnosis Inexpensive Can help ICT (boundary scan)

No test fixture required

Functional test

Catches wider spectrum of defects

Better diagnostics than self-test

Most expensive test

Computer run

Test fixture required

Long test development time and costs

Easy data logging/analysis of results

In-circuit test

Highest defect coverage

Precise diagnostics

Very fast

Least expensive test

Requires test fixture and programming

Facilitates device programming

Easy data logging/analysis of results

Vectorless open test

No models required

Overclamp and probes required

Fast test development

High defect coverage

Reversed capacitors detectable

=============

In-Circuit Testing

The most basic form of electrical test is by use of a manufacturing defects analyzer, which is a simple form of an in-circuit tester. The MDA identifies manufacturing defects such as solder bridges (shorts), missing components, wrong components, and components with the wrong polarity as well as verifies the correct value of resistors. Passive components and groups or clusters of components can be tested by the MDA. The PWA is connected to the MDA through a bed-of-nails fixture. The impedance between two points is measured and compared against the expected value. Some MDAs require programming for each component, while others can learn the required information by testing a known good PWA. The software and programming to conduct the tests is not overly complicated.

Conventional in-circuit testers are the workhorses of the industry. They have more detection capability than MDAs and extended analog measurement capability as well as the added feature of digital in-circuit test for driving integrated circuits. The programming and software required for conventional in circuit testing is more complicated than for an MDA, but simpler than for functional testing.

An in-circuit test is performed to detect any defects related to the PWA assembly process and to pinpoint any defective components. The ICT searches for defects similar to those found by both visual inspection and by the MDA but with added capability; ICT defect detection includes shorts, opens, missing components, wrong components, wrong polarity orientation, faulty active devices such as nonfunctioning ICs, and wrong-value resistors, capacitors, and inductors.

To be effective, in-circuit test requires a high degree of nodal access, there fore the tester employs a bed-of-nails fixture for the underside of the PWA. The bed-of-nails fixture is an array of spring-loaded probes; one end contacts the PWA, and the other end is wired to the test system. The PWA-to-fixture contact pressure is vacuum activated and maintained throughout the test. The fixture contacts as many PWA pads, special test pads, or nodes as possible such that each net can be monitored. (A node is one circuit on the assembly such as GROUND or ADDR0.) With contact to each net, the tester can access every component on the PWA under test and find all shorts. The industry standard spring probe for many years was the 100-mil size, typical of through-hole technology probes for the vast majority of devices that had leads on 0.100-in. centers. 75-mil probing was developed in response to the needs of SMT, then 50-mil, then 38-mil; and now even smaller spacing probes are available.

A major concern in ICT is test fixture complexity. Fixtures for testing THT PWAs are generally inexpensive and reliable compared with the fixtures required for testing SMT PWAs if the latter are not designed with test in mind. This leads to high fixture costs, but also makes it extremely difficult to make highly reliable, long-lasting fixtures. Dual-sided probing allows access to both sides of a PWA (whether for ICT or functional text). Complex PWAs may require the use of clam-shell bed-of-nails fixturing to contact the component side as well as the back or underside of the PWA, but it adds significantly to the cost of the test fixture.

Also, sometimes not all nodes are accessible, compromising test coverage.

In-circuit testing is constantly being refined to provide improved test coverage while minimizing the number of required test (probe) points. This is accomplished by including testability hooks in the design and layout of the PWA such as using boundary scan devices, clustering groups of components, being judicious in test point selection, adding circuitry to provide easier electrical access, and placing test points at locations that simplify test fixture design.


FIGURE 5 Boundary scan circuit showing latches added at input/output pins to make IC testable.


FIGURE 6 Integrated circuit with boundary scan in normal operating mode (left) and with boundary scan enabled for testing (right).

By way of summary from Section 3, the use of boundary scan has been effective in reducing the number of test points. In boundary scan the individual ICs have extra on-chip circuitry called boundary scan cells or latches at each input/output (see Fig. 5). These latches are activated externally to isolate the I/O (wire bond, IC lead, PWA pad, external net) from the internal chip circuitry (Fig. 6), thereby allowing ICT to verify physical connections (i.e., solder joints).

Boundary scan reduces the need for probe access to each I/O because an input signal to the latches serially connects the latches. This allows an electrical check of numerous solder joints and nets extending from device to device. The reality of implementing boundary scan is that most PWAs are a mix of ICs (both analog and digital functions) with and without boundary scan using a mix of ad hoc testing strategies and software tools. This creates a problem in having a readily testable PWA.

Conventional in-circuit testing can be divided into analog and digital testing. The analog test is similar to that performed by an MDA. Power is applied through the appropriate fixture probes to make low-level DC measurements for detecting shorts and the value of resistors. Any shorts detected must first be re paired before further testing. The flying probe tester, a recent innovation in ICT, performs electrical process test without using a bed-of-nails fixture interface be tween the tester and the board under test. Originally developed for bare board testing, flying probe testing-together with associated complex software and programming-can effectively perform analog in-circuit tests. These systems use multiple, motor-operated, fast-moving electrical probes that contact device leads and vias and make measurements on the fly. The test heads (typically four or eight) move across the PWA under test at high speed as electrical probes located on each head make contact and test component vias and leads on the board, providing sequential access to the test points. Mechanical accuracy and repeat ability are key issues in designing reliable flying probers, especially on dense PWAs with small lead pitches and trace widths.

Flying probers are often used during prototype and production ramp-up to validate PWA assembly line setup without the cost and cycle time associated with designing and building traditional bed-of-nails fixtures. In this application, flying probers provide fast turnaround and high fault coverage associated with ICT, but without test fixture cost. Flying probers have also been used for in-line applications such as sample test and for production test in low-volume, high-mix PWA assembly lines.

The second phase of analog test-an inherent capability of the conventional in-circuit tester-consists of applying low-stimulus AC voltages to measure phase-shifted currents. This allows the system to determine the values of reactive components such as capacitors and inductors. In taking measurements, each component is electrically isolated from others by a guarding process whereby selective probes are grounded.

In digital ICT, power is applied through selected probes to activate the ICs and the digital switching logic. Each IC's input and output is probed to verify proper switching. This type of test is made possible by the use of a technique called back driving in which an overriding voltage level is applied to the IC input to overcome the interfering voltages produced by upstream ICs. The back driving technique must be applied carefully and measurements taken quickly to avoid overheating the sensitive IC junctions and wire bonds (Fig. 7). An example of how this happens is as follows. Some companies routinely test each IC I/O pin for electrostatic discharge (ESD) susceptibility and curve trace them as well be ginning at the PWA perimeter and working inward. They then move farther into the PWA back driving the just-tested ICs. These devices are weaker than the devices being tested (by virtue of the test) and fail.

In-circuit testing is not without its problems. Some of these include test fixture complexity, damage to PWAs due to mechanical force, inefficiency and impracticality of testing double-sided PWAs, and the possibility of overdriving ICs causing thermal damage. Because of these issues many companies have eliminated ICT. Those companies that have done so have experienced an increase in board yield and a disappearance of the previously mentioned problems.


FIGURE 7 Back- (or over-) driving digital ICs can result in adjacent ICs being overdriven (a and c), resulting in overheating and temperature rise (b), leading to permanent damage.

Vectorless Opens Testing

Vectorless opens testing uses a special top probe over the IC under test in con junction with the other standard probes already present. By providing a stimulus and measuring the coupling through the IC, open connections can be accurately detected. Vectorless test can be used effectively even when IC suppliers are changed. It doesn't require the expensive programming of full ICT techniques.

Functional Testing

Although an in-circuit test is effective in finding assembly defects and faulty components, it cannot evaluate the PWA's ability to perform at clock speeds. A functional test is employed to ensure that the PWA performs according to its intended design function (correct output responses with proper inputs applied).

For functional testing, the PWA is usually attached to the tester via an edge connector and powered up for operation similar to its end application. The PWA inputs are stimulated and the outputs monitored as required for amplitude, timing, frequency, and waveform.

Functional testers are fitted with a guided probe that is manually positioned by the operator to gain access to circuit nodes on the PWA. The probe is a trouble shooting tool for taking measurements at specific areas of the circuitry should a fault occur at the edge connector outputs. The probe is supported by the appropriate system software to assist in defect detection.

First-pass PWA yield at functional test is considerably higher when pre ceded by an in-circuit test. In addition, debugging and isolating defects at functional test requires more highly skilled personnel than that for ICT.

Cluster Testing

Cluster testing, in which several components are tested as a group, improves PWA testability and reduces the concerns with ICT. One begins with the components that are creating the testing problem and works outward adding neighboring components until a cluster block is defined. The cluster is accessed at nodes that are immune to overdrive. Address and data buses form natural boundaries for clusters. Cluster testing combines the features of both ICT and functional test.

Testing Microprocessor Based PWAs

The ubiquitous microprocessor-based PWAs present some unique testing challenges since they require some additional diagnostic software that actually runs on the PWA under test, in contrast to other board types. This software is used to exercise the PWA's circuits, such as performing read/write tests to memory, initializing I/O devices, verifying stimuli from external peripherals or instruments, generating stimuli for external peripherals or instruments, servicing interrupts, etc. There are a number of ways of loading the required test code onto the PWA under test.

1. With a built-in self-test, the tests are built into the board's boot code and are run every time the board is powered up, or they can be initiated by some simple circuit modification such as link removal.

2. With a test ROM, a special test ROM is loaded onto the PWA during test. On power up, this provides the necessary tests to fully exercise the board.

3. The required tests can be loaded from disk (in disk-based systems) after power up in a hot mock-up situation.

4. The tests can be loaded via an emulator. The emulator takes control of the board's main processor or boot ROM and loads the necessary test code by means of this connection.

When a microprocessor-based board is powered up, it begins running the code contained in its boot ROM. In a functional test environment this generally consists of various test programs to exercise all areas of the PWA under test. An emulator provides an alternative approach by taking control of the board after PWA power up and providing the boot code. Two types of emulation are used: processor emulation and ROM emulation.

Processor Emulation. Many microprocessor manufacturers incorporate special test circuitry within their microprocessor designs which is generally accessible via a simple three- to five-wire serial interface. Instructions can be sent through this interface to control the operation of the microprocessor. The typical functions available include the following:

Stop the microprocessor.

Read/write to memory.

Read/write to I/O.

Set breakpoints.

Single step the microprocessor.

Using these low-level features, higher level functions can be constructed that will assist in the development of functional test programs. These include Download test program to microprocessor under test's memory.

Run and control operation of downloaded program.

Implement test program at a scripting language level by using the read/ write memory or I/O features.

Recover detailed test results, such as what data line caused the memory to fail.

Control and optimize the sequence in which test programs run to improve test times.

Emulators are commercially available that already have these functions pre programmed.

ROM Emulation. If microprocessor emulation is not available, ROM emulation can be a viable alternative. A ROM emulator replaces the boot ROM of the (DUT) device under test. This means that there must be some way of disabling or removing this. Once connected, the ROM emulator can be used in two different ways. In the first, the DUT is run from the ROM emulator code (after the test code has been downloaded to the ROM emulator), rather than from the PWA's own boot code. This removes the need to program the ROM with boot code and test code. Alternatively, the PWA's own boot ROM can be used to perform the initial testing and then switch to the ROM emulator to perform additional testing.

In the second method, the emulator is preloaded with some preprogrammed generic tests (such as read/write to memory and RAM test, for example). These are controlled by means of a scripting language to quickly implement comprehensive test programs.

Most commercially available emulators come with preprogrammed diagnostic tests that include ROM CRC test; RAM test; I/O bus test; PCI/compact PCI/PMC bus test; and ISA bus test, to name several. In addition, tests can be written using simple scripting languages rather than C or assembly language.

Since PWA throughput is driven by test time, Emulators provide a reduction in test time, thus increasing PWA throughput. They also possess the advantage of having an easy-to-modify test code. This means that redundant test code can be removed, test sequencing can be optimized, and the test code compressed, leading to dramatic improvements in test time with minimal programmer input.

The major disadvantage of using an emulator is cost. However, due to the benefit of improved test development times and PWA test throughput, this investment can be rapidly recouped.

Testing Challenges

Limited electrical access is the biggest technical challenge facing in-circuit test.

A steady decline in accessible test points is occurring due to shrinking board geometries. The decline in electrical accessibility is due to both the use of smaller test targets and a reduction in the number of test targets. Designers driven to create smaller, denser PWAs view test targets as taking away board real estate that does not add value to the end product. The problem is compounded by the increased use of smaller parts such as 0402 capacitors and by denser digital de signs employing micro-BGAs, chip scale packages, and chip-on-board (or direct chip attach) methods.

Smaller test targets mean more expensive and less reliable fixturing. Reduced test target numbers mean reduced test coverage. This typically forces test engineers to write more complex cluster tests at the expense of component level diagnostics.

Very-high-frequency PWA designs, such as for cellular phones, are also a challenge to in-circuit test. When PWAs operate at 1 GHz and higher, the component values become very small and difficult to measure, assuming that test points are available. At such frequencies, test points become small radiating antennas, causing designers to avoid their use. This is why cellular phone manufacturers traditionally have not embraced in-circuit technology as readily as in other markets, having chosen instead to use x-ray and optical inspection systems.

NEXT>>

Top of Page PREV.   NEXT Article Index HOME