Repository logo

An outlier detection approach for PCB testing based on Principal Component Analysis

dc.contributor.authorHe, Xin, author
dc.contributor.authorJayasumana, Anura P., advisor
dc.contributor.authorMalaiya, Yashwant K., committee member
dc.contributor.authorReising, Steven C., committee member
dc.description.abstractCapacitive Lead Frame Testing, a widely used approach for printed circuit board testing, is very effective for open solder detection. The approach, however, is affected by mechanical variations during testing and by tolerances of electrical parameters of components, making it difficult to use threshold based techniques for defect detection. A novel approach is presented in this thesis for identifying boardruns that are likely to be outliers. Based on Principal Components Analysis (PCA), this approach treats the set of capacitance measurements of individual connectors or sockets in a holistic manner to overcome the measurement and component parameter variations inherent in test data. Effectiveness of the method is evaluated using measurements on different types of boards. Based on multiple analyses of different measurement datasets, the most suitable statistics for outlier detection and relative parameter values are also identified. Enhancements to the PCA-based technique using the concept of test-pin windows are presented to increase the resolution of the analysis. When applied to one test window at a time, PCA is able to detect the physical position of potential defects. Combining the basic and enhanced techniques, the effectiveness of outlier detection is improved. The PCA based approach is extended to detect and compensate for systematic variation of measurement data caused by tilt or shift of the sense plate. This scheme promises to enhance the accuracy of outlier detection when measurements are from different fixtures. Compensation approaches are introduced to correct the 'abnormal' measurements due to sense-plate variations to a 'normal' and consistent baseline. The effectiveness of this approach in the presence of the two common forms of mechanical variations is illustrated. Potential to use PCA based analysis to estimate the relative amount of tilt and shift in sense plate is demonstrated.
dc.format.mediumborn digital
dc.format.mediummasters theses
dc.publisherColorado State University. Libraries
dc.rightsCopyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see
dc.subjectprincipal component analysis
dc.subjectopen defect
dc.titleAn outlier detection approach for PCB testing based on Principal Component Analysis
dcterms.rights.dplaThis Item is protected by copyright and/or related rights ( You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s). and Computer Engineering State University of Science (M.S.)


Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
13.58 MB
Adobe Portable Document Format