Test and Inspection
December 31, 1969 |Estimated reading time: 16 minutes
By John Arena and Roy McKenzie
For the engineer or manager who must examine test and inspection for the first time, figuring out how and when to use the different methodologies may seem daunting at first. A basic understanding of the most commonly employed strategies and equipment, their individual strengths and weaknesses, and their respective roles within the manufacturing process is provided. Potential users quickly can envision the opportunities for process improvement and quality assurance that test and inspection can bring to their factory floor.
Identifying defects in the electronics assembly process is as old as the assembly process itself. Over time, this effort has evolved continuously to encompass a wide range of test and inspection strategies. In their quest to capture defects and in so doing, optimize the assembly process, increase product quality and reduce manufacturing costs today's electronics manufacturers have more equipment options than ever before.
In-circuit Test (ICT)Stationed following the post-reflow inspection process, ICT long has provided manufacturers with a powerful tool to verify the structural integrity of the printed circuit board (PCB) during production. The test philosophy behind ICT is that if a PCB is designed and built correctly, it will perform correctly; a good process will produce good boards. ICT adoption as a primary PCB test method has been instrumental in making high-quality, low-cost electronics available in the past 25 years.
Figure 1. ICT contacts multiple test points simultaneously to test the electrical integrity of PCBs.
A traditional ICT system uses a bed-of-nails test fixture to access multiple test points on the PCB's bottom side (Figure 1). With sufficient access, in-circuit testers can transmit test signals into and out of PCBs at high speed, evaluating components and circuits to quickly identify opens, shorts and defective components for subsequent rework or repair. Additionally, data collected by in-circuit testers can be used for in-line process control, providing real-time feedback to enable fast error correction in the upstream processes.
Recently, however, as components become more complex and packed more densely onto boards, there are fewer test points available on the PCB. While ICT achieved widespread acceptance at a time when PCBs were limited in density, with circuit nodes readily accessible to bed-of-nails fixtures, today's PCBs offer increasingly limited access for test. As test point accessibility declines, in-circuit fault coverage also declines. This loss of access has heightened demand for other test and inspection technologies to supplement or complement ICT in the assembly line.
Figure 2. With flexible, fast-moving test heads, flying probers serve to complement ICT systems in the assembly process.
Flying ProberA form of ICT, the flying prober performs electrical process test without using a bed-of-nails fixture interface between the tester and the board under test. Essentially, the flying prober serves as an ICT with a "universal fixture." The test heads (typically four or eight) move across the board under test at high speed, as electrical probes located on each head make contact and test device vias and leads on the board, providing sequential access to the test points. Mechanical accuracy and repeatability are key issues in designing reliable flying probers especially on dense boards with small lead pitches and trace widths (Figure 2). Advances such as surface linear motors ensure long-term repeatability without accuracy degradation over the prober's life.
The flying prober is a fairly recent addition to the arsenal of test and inspection techniques. Its flexible, fixtureless approach makes it a useful complement to ICT, giving the flying prober electrical access to smaller geometries on dense, small boards.
Flying probers often are used during prototype and production ramp-up to validate line set-up without the cost and cycle time associated with designing and building traditional bed-of-nails fixtures. In this application, flying probers provide fast turn-around and high fault coverage associated with ICT, but without test fixture cost. Some manufacturers also have used flying probers for in-line applications such as sample test and for production test in low-volume, high-mix lines.
Boundary-scan TestBoundary-scan test is designed to compensate for the loss of ICT access. Using the PCB's edge connector or the ICT's bed-of-nails fixture, boundary-scan techniques enable virtual access for testing devices and circuit nodes that physically are inaccessible to ICT and fly-ing probers.
For circuits using chips that incorporate boundary-scan capability, it is a useful complement to ICT, detecting and diagnosing a wide range of structural faults in digital circuits, including 100 percent coverage of pin-level faults. Defects such as stuck-at-pin faults, shorts and opens are isolated and diagnosed quickly. Primarily software-driven, boundary-scan tools generate specialized test patterns suited to the devices being tested, while functional test also can be enhanced by using boundary-scan techniques. One shortcoming of this technique, however, is that devices must be designed to function with boundary scan. Given that not all assemblers may wish to use boundary scan and the additional costs required, chip designers are not always driven to build this specialized function into their devices.
Functional TestAlways employed as a final manufacturing step, functional test provides a pass/fail determination on finished PCBs before they are shipped. Functional testers typically interface to the PCB being tested via its edge connector or a test-probe point and simulate the final electrical environment in which the board will be used. The most common form of functional test, known as "hot mock-up" simply verifies that the PCB is functioning properly. More sophisticated forms of functional test involve cycling the PCB through an exhaustive range of operational tests. However, the time and resources required to generate such tests have limited their usefulness to military/aerospace applications.
As standards such as VXI enter the picture, automated test equipment (ATE) manufacturers move from designing dedicated, proprietary functional testers to more modular test systems that can be configured and reconfigured. Electronics manufacturers who previously might have built their own custom rack-and-stack systems for each individual application or board type find that it makes more economic sense to invest in a flexible, reusable functional test platform. This move from more comprehensive, dedicated functional test to simpler hot mock-ups has pushed manufacturers to improve product quality and good yields prior to functional test, placing increased value on test and inspection within the assembly line.
To perform such a wide array of tests while accommodating a range of products, flexibility is crucial. Open-system architecture, programming ease and the ability to connect to the product in various ways enable functional testers to make the final decision on product quality.
InspectionInspection differs from test primarily in that it uses some form of optics human vision, X-ray, video camera, even laser rather than electrical testing. Generally, inspection is performed earlier in the PCB assembly process and is oriented toward process monitoring and control. Inspection equipment integrates closely with assembly equipment and typically is used in-line.
Human InspectionNaturally, the oldest form of board inspection relies on human inspectors to carefully evaluate the PCB for visible faults. While this method may allow manufacturers to avoid investing in automated inspection equipment, there are numerous drawbacks to human inspection. Foremost is the process unreliability. Numerous research studies have called into question the correlation of human assessments with the actual board condition, as well as the reproducibility of those assessments. Furthermore, new device technologies and features, such as 0201 components and fine-pitch device leads, are barely visible to the naked eye. In addition to these technical challenges, human inspection can become labor-intensive for mass production, and the costs associated with hiring and training a large staff of inspectors often are prohibitive.
X-ray InspectionX-ray systems have proven useful for inspecting solder quality following solder reflow, as well as inspecting components such as bypass capacitors that either are inaccessible to or cannot be tested by ICT. Additionally, the unique nature of X-ray technology enables these systems to inspect components and connections that are obscured visually, such as the solder ball joints beneath a ball grid array (BGA) package.
Some systems use X-ray to produce two-dimensional (2-D) images. These systems provide high defect coverage and sufficient throughput for less complex PCBs, particularly single-sided PCBs, while inspecting a full range of package technologies, including BGAs, chip scale packages (CSP) and flip chip assemblies.
Three-dimensional (3-D) X-ray systems also are available, offering 2- and 3-D imaging for greater throughput and measurement precision for more complex PCBs. While widespread X-ray inspection adoption has been slowed due to high capital costs and throughput limitations, manufacturers whose products include a high percentage of BGAs (or similar devices whose faults are obscured from other vision techniques) will find X-ray useful as costs come down and speeds are increased.
Figure 3. AOI systems rely on powerful imaging technologies to provide thorough coverage of visible assembly defects.
Automated Optical Inspection (AOI)AOI systems rely on camera-based imaging technologies to inspect PCBs for visible faults. The ability of AOI to cover a broad range of defects at production speeds while overcoming the loss of access associated with ICT speed has enabled AOI to serve as a valuable tool for process control in addition to identifying assembly defects (Figure 3).AOI equipment is used in-line following the component placement and solder reflow processes. Each of these processes has a related set of defects and inspection criteria. As a result, some AOI manufacturers have developed unique systems for both post-placement and post-reflow applications.
Post-placement AOI. AOI systems can be positioned immediately following each placement process in the assembly line, from the high-speed chipshooter to the fine-pitch placement machine. Numerous variables in the placement process can affect quality and contribute to defects, including component size, weight and texture, as well as the quality of placement equipment maintenance. By monitoring the placement process for variations, post-placement AOI can help manufacturers minimize their impact on process quality.
Post-placement AOI focuses on the accuracy of the placement operation, inspecting for component presence and absence on the PCB as well as the offset of components rotationally and in the X-Y axes. Additionally, post-placement AOI verifies component orientation, correct component placement in its target area and whether any components show signs of damage.
Some AOI systems employ 2-D imaging, high-speed cameras and specialized lenses to ensure accurate measurement for post-placement AOI. The most successful imaging is performed in full color, with broad-spectrum lighting bringing the PCB and components into sharp relief. Three-tiered image analysis runs concurrently with the inspection process, enabling the AOI system to keep pace with high-speed placement equipment while providing real-time process monitoring
Post-reflow AOI. The requirements for post-reflow AOI differ substantially from those found in post-placement inspection. When stationed following solder reflow, AOI can identify the wide range of defects that the solder reflow process can introduce. Missing, tombstoned, billboarded, misoriented, offset or superfluous components can be caught by post-reflow AOI, along with a complete range of solder defects, including solder bridges, insufficient or excess solder, and solder balls. Other defects include lifted component leads, gold-finger contamination, incorrect jumper position and switch settings, through-hole pins, and improperly inserted compliant and VHDM connectors.
To achieve reliable inspection of such a broad range of defects, post-reflow AOI systems require technology that is both extremely flexible and highly specialized. Instead of a single camera, some AOI systems feature a 3-D imaging system using as many as five cameras, with one looking directly down at the PCB and the others arrayed at the four compass points. The multiple angled views that these cameras provide help ensure complete coverage of all visible defects.
In post-reflow inspection, monochrome lighting and imaging systems typically are used to best highlight the full range of solder joint characteristics. A programmable network of monochrome light-emitting diodes (LED) can target specific components to address shadows or visual obstructions while providing optimum lighting for each area being inspected.
ConclusionElectronics manufacturers have long been motivated to identify and control defects throughout PCB assembly, both for quality control purposes and as a cost-reduction strategy. Several test and inspection methods have been established to meet these demands, each with unique capabilities. Manufacturers who assess the range of test and inspection options and determine which method or combination of methods best meets their specific goals can realize exponential gains in overall yields, product and process quality while increasing the cost-efficiency of their assembly operations.
JOHN ARENA may be contacted at Teradyne Inc., Assembly Test Div., 600 Riverpark Drive, North Reading, MA 01864; (978) 370-2700; E-mail: john.arena@teradyne.com, and ROY MCKENZIE may be contacted at Teradyne Inc., 2625 Shadelands Dr., Walnut Creek, CA 94598; (925) 932-6900; E-mail: roy.mckenzie@teradyne.com.
Completing the Loop with AOI and SPCBy Boris Mathiszik
The current industry trend of manufacturing off-loading by OEMs is not going away. This crucial industry development carries with it many seen and, perhaps, unforeseen consequences.
As an industry, we already are facing consequences such as the rise in technical capabilities of those contract electronics manufacturers (CEM) large and clever enough to compete on a global basis. CEMs quickly are becoming the center of electronics manufacturing technology as they compete among themselves for the lion's share of the assembly business.
To implement new technologies and deal cost-effectively with a changing product mix, important manufacturing aspects will become progressively critical. Among these are line flexibility and complete line optimization. Combined, these aspects hold the key to greater throughput and lower production cost.
Universally, leading edge manufacturers have AOI systems to detect errors, eliminate faulty assemblies and ship better products.
However, there is much more to greater throughput and lower cost than bolting an AOI system to the end of the line. Most important is the complete, real-time use of AOI to do more than eliminate defects after the fact.
Knowledge-driven ManufacturingKnowledge is power particularly in a maturing industry that has, in many ways, reached the apex of automation. Not long ago, when the mandate came from management to improve quality and increase throughput, the solution was more automation. But the industry has hit the wall. Processing improvements are incremental.
As an industry, we have come to realize that the best way to accomplish more with less is to focus on line optimization and to use the equipment already in place to its fullest potential.
As part of this trend, electronics manufacturers are trying to achieve yield improvements with tools like statistical quality control (SQC), statistical process control (SPC) and design of experiment. These approaches focus on three things: defining, measuring and improving a process. In all cases, the single most important key to success is current, valid, real-time data.
Many OEMs and CEMs understand that the industry's standard tunnel vision of looking at the operation of each individual machine in isolation is not the solution for today's manufacturing environment. For OEMs, new markets and greater competition are forcing faster manufacturing speeds and improved line efficiency. For CEMs, customers are concerned with quality and increasingly demand measurable performance data to verify that particular manufacturing standards are being met.
Achieving both goals requires a change from single-machine control to understanding the entire manufacturing process as one integrated picture. With such a "wide angle" perspective, the ultimate goal is to integrate every piece of the process across all machines and vendors.
In short, static SPC, when linked to AOI, can deliver real-time process control using digital technology to seamlessly monitor all machines on an SMT line, resulting in both throughput and quality improvements.
The Four Step ProcessUsing AOI to push throughput and quality requires four steps, beginning with the decision to use AOI and ending with a system of real-time, closed loop control.
Historically, manufacturers turn to AOI to detect errors with greater speed, earlier in the manufacturing process. Generally, vision-based systems have been used as "intelligent conveyors," weeding out defective products.
Advanced AOI use for data collection and real-time SPC has been discussed for years but never used fully. Without advanced, high-speed AOI, there can be no real-time information about defects. Instead, manufacturers still look to QC records defects found after visual inspection or ICT. To achieve and maintain optimum throughput and quality, process variables must be measured in real-time, and problems (processes out of preset parameters) must be identified as they occur. This is the essence and the value of real-time SPC with AOI.
Today's advanced AOI systems incorporate real-time tools that allow manufacturers to better understand what defects are occurring, create real-time charts of the most frequent defects, as well as the most problematic component types and reference designations, and then look at each process to quantify the problem.
Figure 1. An example of a real-time Pareto chart illustrating defects. Using this chart, process engineers can analyze the variable measurement data that AOI collects. Additionally, using this tool, engineers can define process variability to better understand the impact a variable has on defects.
Step One. The first step to improving yield through AOI is to use AOI-collected data to better understand the existing defects.
Once process engineers have a real-time Pareto chart of their defects, they can analyze the variable measurement data that AOI collects (Figure). Beyond a frequency distribution of process defects, they now can define process variability by analyzing the mean, range, standard deviation and other statistics to better understand the impact a process variable, such as placement, may have on defects and nonconformities.
As logical as this first step sounds, many manufacturers do not have access to basic defect knowledge in real-time. And real-time is what matters. Understanding the cause of a particular defect the next morning or week is no longer practical.
Step Two. Set up real-time monitors on those process variables that impact defect rates the most.
In other words, driven by defect information collected by the AOI machine, engineers will set upper and lower control limits for problem areas. In cases where the average is found to be outside these limits, the AOI tool alerts to out-of-spec conditions and operators intervene in the process.
It is important to note the difference between hard and soft defects. The former are simple binary problems, for example, whether a component is on the board or not. Because there is no variability, such defects are simple to track and solve.
Soft defects, on the other hand, are more difficult to characterize because inherently they involve variability. To use AOI for greater productivity, soft defects must be defined clearly and monitored in real-time, allowing immediate intervention when parameters are exceeded. Such real-time measurement cannot be accomplished fully with low-end AOI systems that simply detect binary problems. Rule-based AOI systems are necessary because they provide variable measurement data, not just pass/fail information.
A real-time alarm system, based on reliable detection and measurement, is critical to eliminating defects at first detection. However, to be effective requires discipline. Operators must comprehend problems, correct them and restart the line. Typically, operators do not have the knowledge or training for this, and while they may note the problem and restart the line, the defects continue.
Step Three. To combat the lack of operator training and in response to the need for real-time, closed loop feedback, this step involves using individual data collection monitors. Adding modules to each machine in the process and linking them to the AOI system creates dynamic process control (DPC) or real-time SPC with AOI. With these modules linked to AOI through a line controller, defects automatically are correlated to circuit reference data points and linked to the guilty machine. Operators must intervene because the DPC system will not allow the machine to run until a recommended corrective action is taken and that action is logged in for future reference.
Step Four. Many harbor the mistaken notion that achieving greater yields is a "snapshot" goal a one-time measurement that states, "At a specific point in time, the line achieved 98 percent yield." Excellent results, to be sure, but incomplete at best. Within minutes after that snapshot, a bad batch of components, a clogged screen or a bad feeder could alter line efficiency drastically. Without AOI and DPC, there is no way of knowing in real-time that the line is no longer optimal.
BORIS MATHISZIK, director of marketing, may be contacted at Machine Vision Products Inc., 5940 Darwin Ct., Carlsbad, CA 92008; (760) 438-1138; E-mail: boris@visionpro.com.