Optimizing component placement: CMM vs. AOI: together, non-contact coordinate measurement machines and automatic optical inspection systems can provide placement process control.

The importance of quality control in printed circuit board (PCB) assembly is intensifying in proportion to the shrinking geometries of boards and components. The challenge for placement equipment is also increasing as these shrinking components are extremely fragile. Inline automatic optical inspection (AOI) systems used for inspection, verification and measurement after component placement can help maintain a level of quality that seems to work well for most standard board configurations. However, AOI systems do not lend themselves to the proactive statistical process control (SPC) that is required for effectively streamlining today's highly complicated assembly lines.

Today, performance baselines must be established based on knowledge of machine characteristics and using in-process and additional monitoring and trend analysis to anticipate and prevent defects. Rework used to be an expected part of production, but now it is avoided as much as possible because of the added cost factor and because many new assembly technologies and materials simply cannot be reworked. SPC can minimize both unscheduled downtimes and defective boards. The ability to adjust the process based on defects when they do occur allows continuous quality improvement.

Two types of placement machines have evolved in the last decade: chip shooters and fine-pitch placement machines. Chip shooters provide high-speed placement of small components such as resistors, capacitors and small packages with low lead counts. Possibly 80 percent of a PCB's population is placed by chip shooters.

Fine- and ultrafine-pitch devices are placed by highly accurate; but slower, machines designed specifically to handle these more fragile components. Because fine-pitch devices with high lead counts present more opportunities for error, the primary concern is the accuracy of the placement machine, assuming that good components are delivered to it.

Where Do Defects Occur?

Most serious defects occur with fine-pitch devices. Inspections after fine-pitch placement still reveal paste and small passive device problems. More importantly, inspections can pinpoint fine-pitch placement defects so that all can be corrected before reflow.

Reflow is the key step. Before reflow, paste deposition can be cleaned and reapplied. Theoretically, misplaced components of any type can also be corrected. Machine calibrations can be adjusted.

After reflow, the story is entirely different, with the possibility of many costly boards turned to scrap. However, to be effective for true process monitoring, any inspection at this point must include precision measurement at the lead level, typically requiring more resolution than is available in AOI systems.

The Placement Standard

The new IPC-9850 Surface Mount Equipment Characterization standard provides a methodology for characterizing the accuracy of pick-and-place machines independent of component and board parameters. It was basically created as a standardized testing tool for manufacturers to quantify their machines before they ship to customers.

IPC-9850 specifies the accuracy required for each of five test components: two quad flat packs (QFP100 and QFP208); micro-ball grid array (microBGA); 1608C capacitor; and small outline integrated circuit (SOIC16). Each machine must place a total of 4,400 of these components onto a PCB and a glass panel.

Accuracy is measured by placing calibrated replicas of these components (glass slugs) onto a calibrated glass plate. The glass plate has a sticky surface, and the glass slugs are the same size and shape as components, but without leads. The plate's sticky surface holds the slugs firmly in place during transfer of the plate from the placement machine and also during any motion of an x-y table on the inspection platform. Following placement of all slugs, the plate is moved to a vision coordinate measurement machine (CMM) where each part location is measured for accuracy (Figure 1).

[FIGURE 1 OMITTED]

Vision Inspection Systems

The two primary vision inspection systems are AOI and CMM. The more known AOI system provides'a fast visual inspection, and it is typically placed in the production line. AOI basically surveys the surface of its programmed area and records visual data about the board. It is exactly what its name implies, an optical inspector.

AOI systems that provide a pass/fail determination on part placement are used to inspect components for integrity and correct position and orientation. These systems offer high-speed automated visual inspection, but, due to the speed demands, they are low resolution and provide qualitative rather than quantitative results.

Using the rapid pass/fail or present/absent capabilities of most AOI machines, a process that is in control produces good products. However, when the process goes out of control, the AOI machine cannot provide quantitative information that will help the process get back within range or determine what caused the discrepancy.

The high-speed requirement of AOI machines results in rapid inspection of many components in a large field of view without measurement. Because fine-pitch placement brings with it higher density and tighter tolerances, most large field-of-view AOI machines lack the resolution needed to provide highly accurate measurements. For example, a typical field of view for a high speed AOI system might be 50 [mm.sup.2]. But, to measure fine-pitch placement at lead level, the field of view should be closer to 5 [mm.sup.2].

A vision CMM also uses optics for inspection, but it is a true metrology tool, not simply an optimized camera. Going beyond surface visual data, this system inspects for missing parts, polarity and orientation (x, y and angle). As a complement to the position and orientation data obtained from computer-aided design (CAD) systems, the use of single-click inspection routines significantly reduces the time required to create a vision CMM inspection program.

The primary value of a vision CMM is in measurement accuracy. The accuracy level is determined by the application's requirements. For example, a cell phone manufacturer may require a stringent process capability (Cpk) of 1.6 for 50 percent on-pad placement of 0201 components and a placement accuracy of 70 to 80 microns. Using 80 microns as 4[[sigma], one standard deviation is then 20 microns. For a high confidence level in the placement accuracy, the measurement system should be at least a factor of four better than the standard deviation. For this example, the measurement accuracy must be in the range of 5 microns.

Measurement accuracy is predicated on proper imaging, including lighting for high contrast and optics for the resolution appropriate to the task. In the research lab, this type of system can be used to fully characterize a process based on individual constituents such as stencils, solder pastes and components.

CMM Inspection and SPC

When considering placement inspection as part of the overall production SPC and quality control, some issues must be addressed. The pick-and-place machine is shipped compliant to the IPC-9850 standard for accuracy. But a machine will experience changes over time, even from a regular maintenance schedule, so recalibration is required. But, any time a machine is recalibrated for any reason, you must assume that the machine is no longer in optimal condition. A vision CMM can reestablish this original precision. To create truly valuable SPC data, the placement machine must always be validated as true to its original calibration.

For enough detail to fully segregate potential defect causes, a high precision sampling formula must be developed to track fine-pitch placement. This formula will quantify the placement tolerances of individual leads on any or all components. With this capability, fine-pitch devices, with their greater opportunity for error, can be sampled at rates in accordance with production line speeds.

A vision CMM will first quantify the placement machine's accuracy and then monitor subsequent component placement on a predefined schedule. The benefit is that the CMM can segregate all possible error sources individually. This capability is very important in maintaining strict process control because, when combined, the individual complexities of components, boards, solder paste and placement machines make pinpointing the causes of errors extremely hard.

Also, a vision CMM does not use a golden board template to compare images. Instead, it uses CAD or user-programmed data to navigate the board and determine actual lead locations (Figure 2). Component locations are compared to nominals that are obtained from the CAD data or the user.

[FIGURE 2 OMITTED]

A typical high-end vision CMM should include camera/laser technology and PC-based control. The lighting options available include coaxial, programmable ring light (PRL) and color light-emitting diodes (LEDs).

A key feature must always be the set-up time between part changeover. Software that uses CAD data of the parts (stencil, PCB, component placement, etc.) can be used to generate an inspection sequence based on an operator's inputs, thus eliminating the programming bottleneck.

The IPC Standard and SPC

IPC-9850 is used,by placement equipment suppliers, systems integrators and equipment buyers. While it was designed as a standard by which placement machine accuracy could be compared, its methodologies can also be used for ongoing process monitoring of the placement equipment in any production line.

Stellar performance in characterization does not necessarily guarantee high quality placement of real components on real boards. IPC-9850 was designed for ease of optical vision inspection using backlighting. However, on a real production line, components differ from the glass slugs in several ways: composition, the degree of precision and dimensional stability of the form, the possibility of lead deformation during placement and the variation in imaging parameters.

Because the monitoring of placement accuracy on the production floor differs from the characterization performed by the equipment supplier, many large original equipment manufacturers (OEMs) use IPC-9850 primarily to qualify new machines and for monitoring during regular preventive maintenance.

For new placement machines, the procedure is to shut the line down for less than 30 minutes once a week to collect data for an SPC baseline. Production is resumed while the data are analyzed off-line. Once the new equipment trend is established, the monitoring schedule is adapted to the new machine.

Besides periodically verifying placement accuracy using slugs, board manufacturers are interested in characterizing the placement accuracy of real components as a function of stencils, boards, solder pastes and components. Accordingly, many manufacturers are investing in high-end vision CMMs because of their flexibility in lighting and image resolution in going from characterization to real assemblies (Figure 3).

[FIGURE 3 OMITTED]

The basic problem in solely relying on glass plate/slugs is that the visual reference needed per the IPC standard for accurate machine placement is not enough for real-world SPC data. Even the best AOI systems cannot deliver this data because of their resolution limitations. However, a capable vision CMM can handle even the most intricate real board configuration because of its much higher resolution, lighting flexibility, enhanced image processing and sub-pixeling.

A typical high-volume OEM will use a vision CMM with a glass mock-up of a board to test and prove their high-speed placement systems both in house during customer evaluations and on-site after a system has been shipped and placed into position on the production floor. Most OEMs also support their customers by providing in-field verification of the placement machine accuracy if the customer moves or modifies it. They will also provide CMM optimization on their machines for particular applications.

While IPC-9850 was designed to use a limited set of precision components, engineers at one company have also devised a method of loading precision-cut silicon die as idealized components in a tape and reel to verify the accuracy of their chip shooters.

Several large electronics manufacturing services (EMS) providers echo the functional tests used at OEMs, with pre-purchase verification, repeating the process at installation, and again if a machine has been moved. In addition, because of the more rugged use seen at EMS facilities, most EMS providers check placement machine accuracy periodically and require stability in the x, y and [theta] axes. These tests allow them to identify drift and placement problems down to the placement head or nozzle level.

Conclusion

AOI inspection after placement in the production line is helpful. However, it does not provide the measurement necessary for trend analysis nor does it provide the data needed to analyze the cause of any problem. The cause has to be determined to prevent recurrence. Error prevention requires knowledge of all aspects of a process, from details of machine performance to details of individual assembly components.

On the production floor, the most accurate first article inspection should be performed with a vision CMM to verify production readiness and set a baseline for subsequent sampling. Where an AOI system provides high-speed pass/fail data, a vision CMM provides in-depth measurement data for trend analysis and process control. AOI checks the process in action. The CMM inspection is more time-consuming, but delivers highly accurate variable data required for gaining control of the process and for statistical trend analysis.

The CMM quantifies the process to ensure placement is working at optimum performance so the AOI can qualify the process; for example, to catch any random fluctuations such as faulty paste application. Both process qualification and quantification are needed for complete process control.

Vijay Patel is a product marketing manager with View Engineering Inc., Simi Valley, CA; e-mail: [e-mail].