Correlating inspection trending with test data to fine-tune accept/reject parameters.

Industry 4.0 automated inspection technology opens the door to enhanced levels of process control. In addition to having to upgrade equipment, however, fully leveraging the power of this technology requires a strong team, an accurate program validation database, and a methodology to track trends in continuous improvement activities. Here is a look at an implementation process for inline 3-D solder paste inspection and 3-D automated optical inspection following reflow on SigmaTron’s SMT lines in Tijuana.

A planned phase two of this implementation includes adding 3-D AOI to secondary assembly operations post-SMT, plus correlating AOI trending with final test data to fine-tune AOI acceptance/rejection parameters. The facility is currently averaging 50ppm defect rates across its SMT lines. The goal of this greater inspection process is to drop that to zero defects; although, given material constraints are driving a defective component rate that represents a third of that 50ppm level on some programs, zero ppms may be unachievable until material availability returns to normal.

The inspection stations communicate with the screen printer and adjust paste deposition based on defects identified, creating a closed-loop system. They also track trends data, enabling the team to pareto top-five defects, and where indicated, implement continuous improvement initiatives beyond process adjustments. The systems communicate to a database accessible via remote computers available to technical personnel. Trending analysis has already driven changes to stencil maintenance frequency and greater focus on monitoring pick-and-place machine placement variation. From a Lean perspective, the machine-to-machine communication aspect reduces process variability and represents a better use of resources by eliminating the need for human intervention within set parameters. Using the data to determine optimum maintenance activity frequency eliminates defects that would otherwise occur as stencils or tooling gradually slips out of acceptable parameters.

Setting up the systems, a team with a Six Sigma Black Belt and two mechatronic technicians performed gage R&R studies for repeatability and reproducibility using the Automotive Industry Action Group’s acceptability definition, where:

• Ideal = a tolerance or study variance percentage of 10% or less, with a contribution percentage of 1% or less
• Acceptable = a tolerance or study variance percentage of 10 to 20% or less, with a contribution percentage of 1-4%
• Marginal = a tolerance or study variance percentage of 20 to 30% or less, with a contribution percentage of 5 to 9%
• Poor = a tolerance or study variance percentage of 30% or greater, with a contribution percentage of 10% or greater

CP and Cpk ratios were also calculated to measure the process capability against the voice of the customer (standards/requirements). In this analysis:

• Very good = Infinity, CP(Cpk) > 1.67
• Good = 1.33 < CP(Cpk) < 1.67
• Need to manage = 1.00 < CP(Cpk) < 1.33
• Need to improve = 0.67 < CP(Cpk) < 1.00
• Poor management state = CP(Cpk) < 0.67

The studies and calculations showed the specification limits for all 3-D AOI were in acceptable GR&R condition and very good CP/Cpk condition. The program validation is used for machine verification every six months. Once control limits were established, programming parameters were adjusted for each printed circuit board assembly to assess acceptability based on IPC-A-610 and customer documentation requirements.

Following implementation, the team is focusing on two improvement tracks. The first deals with improving first-pass yields on product identified through weekly trends tracking. In one example, trends data led to the discovery a fixture had warped and was no longer seating correctly, which caused flex cable terminations to shift off pads. The fixture was replaced, correcting the issue.

The second improvement track is focused on improving the overall inspection process. One area being analyzed is the accuracy of operator decisions to override AOI decisions. Customers are evaluating a more efficient labeling strategy to provide full traceability. Test data have been transformed from raw, delimited data to formatted columns. Now remote computers are running a dashboard that visually conveys real-time SPC/Cpk and first-pass yield data, and allows remote debugging.

This Industry 4.0 approach is creating a system that autocorrects solder-deposition-related issues in real time and provides the trends data to enable the team to monitor tooling, equipment accuracy and customer design issues that may be causing defect opportunities. The comparison of AOI trending with test data is increasing AOI programming accuracy, and better understanding of equipment operational trends is helping fine-tune maintenance and calibration intervals. 

Filemon Sagrero is continuous improvement engineer and a Six Sigma Black Belt at SigmaTron International’s facility in Tijuana, Mexico; filemon.sagrero@sigmatronintl.com.

• Prev
• Next