Applying multivariate analysis to analyze and improve component rejection by pick and place machines

ORIGINAL ARTICLE
  • 4 Downloads

Abstract

Pick and place (P&P) machines are often adopted to precisely place electronic components onto a printed circuit board (PCB). If the component is not picked up successfully, then the pickup information is recorded as a “pickup miss.” If the component is picked up successfully, then the charge-coupled device camera at the bottom of the P&P is used in conjunction with a light source to examine the component from the bottom up, thereby determining whether the component’s appearance meets the specification requirements. If the component’s appearance does not meet the specification requirements, a “recognition miss” is recorded in the pickup information sheet and the component is sent into the rejection box; otherwise, the component is placed on the corresponding position of the PCB. Because passive components are tiny, component rejection due to pickup misses and recognition misses frequently occurs, thereby increasing material costs and production time. In this study, passive components of sizes 0402 and 0603 were explored. Specifically, rules were first compiled for component rejection that are attributable to recognition misses (i.e., the component size and angle exceeded the allowable errors for a P&P machine), followed by examining whether actual component rejection records met these rules. Next, components discarded into the rejection box were analyzed to determine abnormal appearance as the main factor causing component rejection. Pickup information was also collected concerning production line mass production and other types of production-related big data. Multiple regression analysis, the chi-square automatic interaction detection (CHAID), and JRip method were used to establish rules for predicting component rejection. The results revealed tape roll over, splice tape joints, excessively long nozzle maintenance intervals, and unstable feeders as key factors influencing the rejection rate. Accordingly, related suggestions were proposed for reducing the rejection rate. Finally, the prediction accuracy of the various methods was assessed, showing that the CHAID method attained prediction errors within 1.0% at various nodes.

Keywords

Pick and place component rejection Pickup miss Recognition miss Big data Data mining Multiple regression analysis CHAID JRip method 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Huang CY, Huang HH (2014) Process optimization of SnCuNi soldering material using artificial parametric design. J Intell Manuf 25(4):813–823.  https://doi.org/10.1007/s10845-012-0720-z CrossRefGoogle Scholar
  2. 2.
    Qin X, Kumbhat N, Raj PM, Sundaram V, Tummala R (2014) Finite element analysis and experiment validation of highly reliable silicon and glass interposers-to-printed wiring board sMT interconnections. IEEE Trans Compon Packag Manuf Technol 4(5):796–806.  https://doi.org/10.1109/TCPMT.2013.2296780 CrossRefGoogle Scholar
  3. 3.
    Huang CY (2015) Innovative parametric design for environmentally conscious adhesive dispensing process. J Intell Manuf 26(1):1–12.  https://doi.org/10.1007/s10845-013-0755-9 CrossRefGoogle Scholar
  4. 4.
    Yu W, Olorunyomi M, Dahlberg M, Djurovic Z, Anderson J, Johan L (2007) Process and pad design optimization for 01005 passive component surface mount assembly. Soldering Surface Mount Technol 19(1):34–44CrossRefGoogle Scholar
  5. 5.
    Synkiewicz BK, Skwarek A, Witek K (2014) Voids investigation in solder joints performed with vapour phase soldering. Soldering Surface Mount Technol 26(1):8–11CrossRefGoogle Scholar
  6. 6.
    Sony (2010) Root cause analysis for component rejection, SMT manufacturing factory. (in Chinese)Google Scholar
  7. 7.
    Panasonic Corporation of North America (2005). Electronic Assembly Mounting System. CM402-L Instruction manualGoogle Scholar
  8. 8.
    EIA Standards and Technology Department (2008) EIA-481-D, Global Engineering DocumentGoogle Scholar
  9. 9.
    Demetgul M (2013) Fault diagnosis on production systems with support vector machine and decision trees algorithms. Int J Adv Manuf Technol 67(9–12):2183–2194.  https://doi.org/10.1007/s00170-012-4639-5 CrossRefGoogle Scholar
  10. 10.
    Huang CY, Lin YH (2013) Applying CHAID algorithm to investigate critical attributes of void formation in QFN assembly. Soldering Surface Mount Technol 25(2):17–127CrossRefGoogle Scholar
  11. 11.
    Wilk-Kolodziejczyk D, Regulski K, Grzegorz G (2016) Comparative analysis of the properties of the nodular cast iron with carbides and the austempered ductile iron with use of the machine learning and the support vector machine. Int J Adv Manuf Technol 87(1–4):1077–1093.  https://doi.org/10.1007/s00170-016-8510-y CrossRefGoogle Scholar
  12. 12.
    Ren L, Lv W, Jiang S (2015) Machine prognostics based on sparse representation model. J Intell Manuf 1-9  https://doi.org/10.1007/s10845-015-1107-8

Copyright information

© Springer-Verlag London Ltd., part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of Industrial Engineering and ManagementNational Taipei University of TechnologyTaipeiRepublic of China

Personalised recommendations