STEP 9: Root Cause Analysis in Test-and-Inspection

Estimated reading time: 6 minutes

By Tim Anderson, Omron electronics

Integrating automated root cause analysis during PCB assembly, based on inspection data, provides an effective way to boost profit margins. It minimizes labor costs due to rework and time spent reviewing false calls. Software that supports root cause analysis helps optimize inspection settings, identify imperfections in incoming components, and highlight areas for improvement in operator training.

Over the last decade, domestic PCB assemblers have experienced a steady squeeze on profitability through a combination of component and board miniaturization, smaller and more varied lot sizes, conversion to lead-free solder, increased offshore outsourcing, and zero-defect customer directives. The survivors of these dramatic changes strive to increase profit margins through improved inspection and process improvement to reduce waste and rework. Software integrating the results of inspections after solder paste deposition, component placement, and reflow stages provides a way to identify, document, and correct root causes of failure.

To achieve zero-defect production, most small- and medium-size PCB assemblers in the U.S. implemented post-solder inspection to prevent bad product from getting out to customers. The systems of choice have been AOI and X-ray, depending on the components and solder used. While this basic prevention step solves the supply chain quality problem, it does not address the process faults. To increase the bottom line, assemblers must do root cause analysis, improve the process, and lower costs.

Many assemblers are adding post-paste inspection to determine the quality and geometry of the solder paste brick. The shape can directly affect the solder results during reflow. Modern screen printers include 2D inspection of the solder built in. However, 3D inspections are the industry standard and deliver more accurate results.

Data Mining Inspection Results

By placing post-paste inspection at the front and post-solder inspection at the back of the line, assemblers gather the raw data necessary to determine where faults occur in the process. Right now, assemblers address only about 50?60% of the problems that occur after soldering, because it is so complicated to determine the root cause. With data mining, specific faults can be traced and corrected. Analysis software tools can perform the analysis and point out problem areas. Dedicated software reduces waste by correcting problems before more value is added to the board, reducing rework and process faults, and maintaining the line improvement more effectively than with AOI alone. Companies involved in board assembly for automotive subsystems, factory automation, consumer products, and similar sectors benefit significantly from applying data-tracking software tools.

Analysis software enables an experienced engineer to document case-by-case the history of faults and reveal patterns or unsuspected root causes. The default fields for case reporting include what faulty type, the occurrence date and time, the component type, process that initiated the defect, primary and secondary factors, displacement variation, mounting precision, and multiple other elements. A consistent set of data should be reported each time a defect is spotted, for uniform analysis.

Figure 1. Sample equations for determining board-level statistical data on defects.

null

Setting Quality Targets

Before using software systems, it is important to define the acceptable limits of defects. Two formulae were used establish acceptable levels in a 10-week quality improvement project at one company. Pass rate is a percentage determined by dividing the number of OK boards by the total number of boards produced, then multiplying the result by 100. Target for this equation was 80%. Similarly, the defect rate, in ppm, is determined by dividing the number of real failed components by the total number, multiplying the result by 106. Target defect rate was 50 ppm. To more clearly define the sources of potential faults, additional formulas were employed to analyze data gathered by in-line AOI systems.

At the component level, the escape rate was determined by the number of OK components that failed inspection criteria divided by the total component count, and the result multiplied by 106. Target was 0%. Good percentage is defined as the number of actual OK components divided by the total number of components, then the results multiplied by 106. This should equal 1 ? (defect rate + escape rate). Finally, the false call rate is a percentage determined by dividing the number of real no-go components that passed inspection by the total component count, and multiplying the result by 106. Target for false calls is 100 ppm. At the board level, defect, escape, good, and false call rates are determined using similar formulae (Figure 1).

Documenting Error Cases

AOI systems capture images of faults, enabling a skilled engineer to identify the problem and recommend corrective action.

Bridge Caused by Excessive Solder. The AOI system post-print detects excessive solder forming a bridge before component placement. The operator did not handle the fault call completely and allowed the board to proceed through placement and soldering. The error was picked up again post-solder. The case file mined by the software system notes that this was a bridge caused by excessive solder paste, discovered by post-paste and post-solder AOI. It also documents the screen printer as the problem initiator and that too much solder paste in the location was the primary defect factor. Analysis determined that, near the solder pads, a solder resist material lifted up the stencil. Solder paste was forced under the stencil, making a bridge. An improvement plan included a complaint to the manufacturer, and operator instruction on how to handle this type of board pollution.

Figure 2. Tracking the defect rate in one study from no controls, to AOI only, to AOI combined with root cause software.

Fillet Error, Component Wettability. Post-solder AOI identifies a component wettability problem that prevents solder joints from forming. Some of the components’ copper leads were not completely plated and had oxidized during stock time in production. Components are stored in dry-pack conditions in the warehouse, but not on the floor. This storage/handling and incoming quality problem was confirmed by completing a production run using components straight from dry pack that had no reflow defects.

Fillet Error, Material on Solder Pad. An unknown material on a solder pad was identified in post-paste inspection. It probably reached the board before screen printing. The paste deposits were OK at this point. Between paste deposition and chip mounting, the solder paste was completely wiped away from one pad and mostly removed from another. This led to an investigation of whether the problem was caused by the manufacturer or in production. It resulted in additional training to prevent wiping the solder off by accident.

Figure 3. Wettability issues could be traced to plating on the component leads, not solder print, as the board was approved by post-print and post-mounting AOI.

The necessary pass rates were achieved during the last five production runs, exceeding the 80% target. Pass rates before root cause analysis had been below 40%. However, target defect rates were achieved only in two of the nine weeks studied.

Steady Reduction in Faults. Another manufacturer performs AOI at all three locations in the assembly process: post-paste, post-placement, and post-solder. On each line, the inspection systems feed data to the specialized analysis software, which can identify the fault quickly before a board moves to the next stage. The reduction in faults, counted as solder points in ppm, was significant.

Conclusion

In one example, when AOI alone was applied, the defect rate remained high, but fell by 9 ppm over the course of a month. The next month, data collected from the various inspection systems was integrated and mined by root cause analysis software. Faults identified by the software in various aspects of the solder application and placement processes were corrected. By the end of that month, quality improved from 79 down to 15 ppm. Over the following months, the plant was able to maintain the improved level of quality (Figure 2). Each of the AOI systems contributed valuable data that led to the finding faults’ root causes.

Integrating the data collected by inspection systems at post-paste, post-placement, and post-solder causes a dramatic reduction in faults, ultimately leading to less waste and rework, and higher overall productivity. SMT

Tim Anderson, sales manager, inspection systems, Omron Electronics LLC, may be contacted at tim.anderson@omron.com.