-
- News
- Books
Featured Books
- smt007 Magazine
Latest Issues
Current IssueBox Build
One trend is to add box build and final assembly to your product offering. In this issue, we explore the opportunities and risks of adding system assembly to your service portfolio.
IPC APEX EXPO 2024 Pre-show
This month’s issue devotes its pages to a comprehensive preview of the IPC APEX EXPO 2024 event. Whether your role is technical or business, if you're new-to-the-industry or seasoned veteran, you'll find value throughout this program.
Boost Your Sales
Every part of your business can be evaluated as a process, including your sales funnel. Optimizing your selling process requires a coordinated effort between marketing and sales. In this issue, industry experts in marketing and sales offer their best advice on how to boost your sales efforts.
- Articles
- Columns
Search Console
- Links
- Events
||| MENU - smt007 Magazine
Improving SMT Yield with AOI and AXI
December 31, 1969 |Estimated reading time: 7 minutes
By An Qi Zhao; Xin Yong Yu; Li Ming Gong; Zhen (Jane) Feng, Ph.D.; Mark Evans; and Murad Kurwa, Flextronics International Inc.
As PCB assembly (PCBA) becomes more complex, automatic optical inspection (AOI) and automatic X-ray inspection (AXI) systems are more widely used in electronics manufacturing. AXI has good defect detection capabilities, but its TaKT time becomes a concern when compared to other machines screen printer, pick-and-place, reflow, wave soldering on the SMT line. How can these two testing machines be used effectively to test production?
While some past studies focused on AXI,1-3 this project analyzes AOI and AXI test data to improve assembly test yields. Waste elimination is the most effective means to achieve cycle time reduction. The key elements of six sigma DMAIC define, measure, analyze, improve, and control and statistical tools support the project. One customer's product, which previously had 100% components covered with AOI and >95% covered by AXI, served as a basis for the experiment. AXI test time was reduced by only testing BGAs, fine-pitch ICs, resistor networks (RNs), and some critical-to-function parts. AXI component and pin coverage changed from 98.4% and 98.9%, respectively, to 13.6% and 50.1%. Yields of AOI (top), AOI (bottom), AXI, ICT, and FT increased, despite reduced test.
We also focused on process issues and improvements using daily AOI and AXI test results. Accessing AXI and AOI defect feedback immediately allowed root causes of process issues to be identified and resolved before much waste was created. Test and process engineers used AOI/AXI test results to adjust printer, pick-and-place, and wave solder machine settings.
The ExperimentWith the increase in component count and solder joint count, AOI and AXI systems can add to manufacturing by reducing downstream electrical testing costs.4 Using AXI in combination with conventional test techniques helps to ensure all defects are caught before products are delivered to customers. However, for high-density assemblies, AXI cannot keep up with assembly processing speeds, as its test time is much longer than other SMT machines. A build-up of inventory occurs in front of AXI, extending the manufacturing lead time of the assembly. In this case, balancing the SMT line can be a challenge.
For the experiment, AOI test remained at 100% coverage, and AXI coverage was reduced from 95%. The PCB layer count varies from 6 to 24 for tested assemblies. Test yields (AOI, AXI, ICT, and FT) for five chosen assemblies from the previous 36 months provided a defect summary. Products with stable processes and satisfying test yields (AOI >95%, AXI >80%, ICT >95%, and FT >99%) and with two digits DPMO (<30) were used. We reviewed AOI, AXI, ICT, and FT historical data to summarize process and defect types, then modified the AXI program by reducing components coverage. Based on analysis of this historical data, it was determined that main defects came from fine-pitch gullwing components and RNs.
Reducing AXI Test TimeOver-processing is processing above and beyond the customer's internal and external requirements. AXI with 100% coverage is nice to have in manufacturing because of its proven capabilities in catching solder joint defects. However, long AXI test times prevent manufacturers from improving SMT process with real-time AXI defect feedback. Using 100% AOI and 100% AXI test inspection coverage for all assemblies for a given customer is a substitute to improving upstream processes. To reduce over-processing, we modified the AXI programs (i.e. reduced test coverage) for these assemblies. AOI test coverage remained at 100%. Components that showed not defects for the last six months were removed from AXI coverage; excepting BGAs. AXI test programs were modified to test only BGAs, fine-pitch, gullwing, PTH, and critical-to-function components found defective over the last six months. Therefore, AXI components and pin test coverage was reduced to less than 20% and about 55%, respectively. AXI testing overall was reduced to about 55% of its previous time. This helped balance SMT TaKT time and enabled real-time AXI defect feedback for the SMT line.
SMT ImprovementTo prevent process-related defects, maintain effective AOI and AXI programs, ensure that machines are in good working condition, eliminate defects by identifying root causes, and maintain real-time test result feedback to upstream processes. To ensure all AXI machines are operating under good test conditions, perform tests daily on a baseline board, which has known solder defects/locations. If there is variation in the defects detected, it is likely that the machine needs to be calibrated. Perform machine calibrations per the recommendations of the AXI vendor.
In-house software was developed to analyze the AXI data. AXI yield and defect information was used to eliminate defects. Two feedback loops AXI to AOI; and ICT and FT to AXI allow operators to use test data to improve upstream processes. If AXI found a defect that escaped from AOI, then the AXI team would feed back this information to AOI. ICT and FT teams also provide feedback to AXI if a solder escape was detected at ICT or FT. For example, in Figure 1, Assembly 415-149 location GRN6 (10-pins components with pitch size 25 mils) was found as solder insufficient through visual inspection at ICT. The defect is obvious on this particular part as a heel is not visible. Why didn't AXI detect it? The original AXI setting was focused on the heel only, and it is shown on Figure 2a: the blue bar. The gullwing algorithm was adjusted for different orientation for testing the heel, center, and toe location as shown in Figure 2b. With the new setting, AXI was able to detect the insufficient-solder defect effectively, even for less-obvious insufficient joints. Having stable and effective AOI and AXI programs is the main requirement for defect detection.
Defect Data Analysis and Root CausesA defect data analysis (defect distribution) report focuses on the main defects for each assembly for the next run. This allows operators to solve process defects by identifying the root cause.
For example, in this experiment, many PS2 (RPOTS) locations had insufficient or open defects. The root cause was incoming material with lifted leads. That explained why some boards had defects on PS2 and others did not have defects at this location. The customer allowed for the component vendor to have 9-mils tolerance of lead lift. However, the PCBA had fine-pitch components with a stencil thickness requirement of ≤6 mils. A new, step-up stencil gave PS2 10-mils-height solder paste without impacting other locations. The failure rate at this location was reduced from 1.5% to 0%.
In another case, AOI found many chip capacitors with dewetting. A total of 38 parts with solder joint defects on 1581 boards occurred over a roughly one-month period. These two parts are lead-free components, and the original oven profile didn't meet lead-free soldering requirements, thus creating the dewetting defects. Oven temperature in zones 1, 7, and 8 were modified so that the peak temperature was increased by about 10°C; reflow time above 183°C was reduced by about 5 seconds. The oven's Cpk was maintained at 1.5 or higher. After the profiles were changed, one defective joint was detected on 651 boards tested over a roughly three-week period.
ConclusionAfter reducing AXI coverage for one assembly, AOI (top), AOI (bottom), AXI, and ICT yields changed from 96.6%, 94.6%, 89.6%, and 98.4%, respectively, to 98.2%, 97.2%, 97.1%, and 98.2%, and FT test yields still remained at 100%. Suggested procedure is to use the original AXI program (i.e. high component coverage) if the ICT or FT yields drop several percentage points for any assembly. Yields remained consistent in this experiment, avoiding the need to return to high-component-coverage AXI.
AOI and AXI are more efficient when used to complement each other. Together, they will identify virtually all defects before electronic test. These test tools can also be used as SMT process improvement tools if their test results are properly analyzed and fed back to upstream processes. By balancing the coverage between AOI and AXI, the AXI cycle time can be significantly reduced, eliminating AXI as a bottle-neck. This will also reduce waste, helping achieve lean-concept manufacturing. With AXI coverage reduction, there is a significant cost savings. However, we suggest using 100% AXI coverage for new products, especially when new package types are used. It is necessary to review all test yields and data to confirm a stable SMT process before reducing AXI test coverage. SMT
REFERENCES:1. Glen Leinbach, "Using Production Defect Data to Improve an SMT Assembly Process," SMT, 2001.2. Zhen (Jane) Feng, Jacob Djaja, and Ronald Rocha, "Automated X-ray Inspection: SMT Process Improvement Tool," SMTA, Chicago, September 2002.3. Zhen (Jane) Feng, Eduardo Toledo, Jonathan Jian, and Murad Kurwa, "Reducing BGA Defects with AXI Inspection," Circuits Assembly, July 2005.4. Keith Fairchild, "Evaluation ROI of AXI vs. AOI," Circuits Assembly, October 2006.
Acknowledgements:The authors would like to acknowledge Wei Bing Qian, Jack Shen, Winde Wang, Wendy Zhang, Angery Lee, Allen Wei, Andrew Ho, EB Yeoh, David Pham, Mark Jones, Dan Jackson, Jane Phan, Jayapaul Basani, Gary Liu, Sam-TS Wong, Barbara Koczera, Heidi Hanson, and Flextronics Shanghai's engineering and production teams.
An Qi Zhao, Xin Yong Yu, and Li Ming Gong can be contacted at Flextronics Shanghai, No.77 Yong Sheng Road, Malu, Jiading, Shanghai, 201801, China. Zhen (Jane) Feng, Ph.D.; Mark Evans; and Murad Kurwa can be contacted at Flextronics International, 1710 Fortune Drive, San Jose, Calif., 95131.