Future work

While the results show a consistent line of improvements when compared to the currently used manual design process, there are still opportunities for further improvements. Some of the smaller and practical ones have been discussed in the previous section, this section will focus on the bigger ones that might need additional research.

A first possible improvement is already touched upon in the previous section, namely FEA accuracy. In order to comply with the IPC/JEDEC standards, one needs to verify each board. If one can show that the FEA results are always in line with the real world results, the time taken in this verification process could be reduced. Some FEA improvements might be needed, think of a 3D instead of a 2D analysis, and discrimination between surface mount and through hole components.

A second possible improvement is extending the research to larger DUTs. A time limiting factor in the optimization process are the FEA computations, which have a quadratic time complexity.

Hence, generating an optimized fixture for large DUTs can quickly become computationally costly.

In case of a panelized DUT, it might be beneficial to first generate a fixture for a single panel, then use this result as a seed for the panelized fixture.

A third possible improvement is in the contact placing process. The current contact placing strategy, discussed in Section7.2, adds contact to the nearest free spot. In case of board cutouts, it can be desirable to use instead closest distance over the PCB as placement criteria. This might overcome placing contacts at the opposite side of a cutout, as frequently happened in the P1SMD benchmark.

A forth possible improvement is the use of an optimization controller algorithm that temporar-ily allows worse results to get to better final results, i.e. is less prone to finding local optima. One controller algorithm candidate is the simulated annealing optimizer, discussed in Section 4.1.4.

Time did unfortunately not allow testing the quality of this optimization controller.

A fifth possible improvement focuses on the topic of fixture manufacturing inaccuracies. So far, push fingers are assumed to have exactly the same length. In practice, manufacturing inaccuracies may cause push fingers ends to deviate from the assumed flat plane. A first research topic would involve measuring such deviations and measuring the influence of such deviations in terms of additional PCBA strain. If this research shows significant strain deviations, additional research could be done on improving the FEA model and/or the optimization controller, minimizing the real board strain.

A sixth possible improvement extends the focus on DUT strain and fixture costs to DUT stability. In some specific cases, DUT strain might still be within safe limits, but large deformations may cause wrong alignments with the fixture. Hence, a third optimization goal could be the maximum absolute deviation in Z-direction.

CHAPTER 9. CONCLUSION

9.3 Conclusion

Section5.3stated four research questions, bundled together in one main question. In this section, these will be answered one by one.

1) What algorithms exists to generate a new, or adjust an existing fixture design?

How do such algorithms behave? No literature has been found discussing an algorithmic approach on improvement of fixture designs. Sections6.1to6.7discuss newly designed algorithms.

The behaviour of these algorithms is discussed in Sections8.3to8.7.

2) What optimization algorithms can use such algorithms to generate an optimal fixture? How do such algorithms behave? Several generic optimization algorithms are discussed in the literature. Chapter 4 presents five of these, together with a case study of an optimization algorithm applied to a related field. Section 6.8 discusses a practical optimization algorithm implementation, of which the results are presented in Section8.8.

3) How good are the generated fixture designs in comparison to manually created fixture designs? Quite good. Table 8.3compares the automatically generated and optimized fixtures with manually created fixtures. On average, strain is reduced by approximately 50%, while costs are reduced by approximately 45%.

4) How good are the generated fixture designs in comparison with the IPC/JEDEC guidelines? This question can only be partially answered. According to the IPC/JEDEC guide-lines, both strain and strain rate need to be measured in order to decide whether or not a fixture design is acceptable. Since strain rate is not analyzed nor otherwise measured, it is not possible to fully answer this question. But, taking a look at Table8.3, two benchmark DUTs are expected to be in line with the IPC/JEDEC guidelines, two might be, and one is probably not.

Main research question: In what way can a fixture design be automatically proposed based on a DUT design file and a test probe mapping file, such that both DUT strain and fixture costs are minimized? An initial fixture design can be generated according to the algorithm discussed in Section6.1. The resulting fixture can be further optimized using the hill climb optimizer discussed in Section 6.8, combined with the different step algorithms discussed in Sections6.2 to 6.7. A proof-of-concept tool has been build, of which the details are discussed in Chapter7. The test results, summarized in Table8.3, show very good results when compared with manually designed fixtures.

Bibliography

[1] E. Andreassen, A. Clausen, M. Schevenels, B. S. Lazarov, and O. Sigmund,

“Efficient topology optimization in matlab using 88 lines of code,” Structural and Multidisciplinary Optimization, vol. 43, no. 1, pp. 1–16, Jan 2011. [Online]. Available:

https://doi.org/10.1007/s00158-010-0594-7 viivii, viivii, viivii,20,21

[2] A. Bansal, S. Yoon, and V. Mahadev, “Flexural strength of bga solder joints with enig substrate finish using 4-point bend test,” in Proceedings of the SMTA Pan Pacific Microelec-tronics Symposium, Jan 2005, pp. 1–8. 14

[3] A. Bansal, S. Yoon, J. Xie, Y. Li, and V. Mahadev, “Comparison of substrate finishes for flip chip packages,” in Proceedings Electronic Components and Technology, 2005. ECTC’05.

IEEE, May 2005, pp. 30–37. [Online]. Available: https://doi.org/10.1109/ECTC.2005.

1441241 3

[4] M. P. Bendsøe, “Optimal shape design as a material distribution problem,” Structural optimization, vol. 1, no. 4, pp. 193–202, Dec 1989. [Online]. Available: https:

//doi.org/10.1007/BF01650949 19

[5] T. Blackwell and J. Kennedy, “Impact of communication topology in particle swarm optimization,” IEEE Transactions on Evolutionary Computation, pp. 1–1, 2018. [Online].

Available: https://doi.org/10.1109/TEVC.2018.2880894 19

[6] R. Eberhart and J. Kennedy, “A new optimizer using particle swarm theory,” in MHS’95.

Proceedings of the Sixth International Symposium on Micro Machine and Human Science.

IEEE, Oct 1995, pp. 39–43. [Online]. Available: https://doi.org/10.1109/MHS.1995.494215 19

[7] A. P. Engelbrecht, “Particle swarm optimization: Global best or local best?”

in 2013 BRICS congress on computational intelligence and 11th Brazilian congress on computational intelligence. IEEE, Sep 2013, pp. 124–135. [Online]. Available:

https://doi.org/10.1109/BRICS-CCI-CBIC.2013.31 19

[8] C.-H. Fu, L.-Y. Hung, D.-S. Jiang, C.-C. Chang, Y. Wang, and C. Hsiao, “Evaluation of new substrate surface finish: Electroless nickel/electroless palladium/immersion gold (enepig),”

in 2008 58th Electronic Components and Technology Conference. IEEE, May 2008, pp.

1931–1935. [Online]. Available: https://doi.org/10.1109/ECTC.2008.4550246 3

[9] J. Goldstein, “Strain gage testing: Predicting and preventing brittle fracture of bgas-brittle fracture of components on printed circuit boards (pcb) during assembly, test or shipping operations has been a,” SMT-Surface Mount Technology-Tulsa, vol. 18, no. 6, pp. 44–51, 2004. 3

[10] Guidance for Strain Gage Limits for Printed Circuit Assemblies, IPC, July 2011. viivii, 3, 14,15,77

[11] IPC/JEDEC-9704A: PRINTED WIRING BOARD (PWB) STRAIN GAGE TEST GUIDELINE, JEDEC, 3000 Lakeside Drive, Suite 309S Bannockburn, Illinois, Jan 2012.

viivii,2,14,77

[12] D. Kehmeier, Intermediate Data Format V3.0, Mentor Graphics Corporation, 1001 Ridder Park Drive San Jose, CA 95131, Oct 1996. 31

[13] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proceedings of the IEEE international conference on neural networks, vol. 4, Nov 1995, pp. 1942–1948. [Online].

Available: https://doi.org/10.1109/ICNN.1995.488968 19

[14] K. Meyyappan, A. McAllister, M. Kochanowski, and I. Hsu, “Effects of glue on the bend performance of flip chip packages,” IEEE Transactions on Components and Packaging Technologies, vol. 31, no. 3, pp. 670–677, May 2008. [Online]. Available:

https://doi.org/10.1109/TCAPT.2008.922006 3

[15] H. Mlejnek, “Some aspects of the genesis of structures,” Structural optimization, vol. 5, no.

1-2, pp. 64–69, Mar 1992. [Online]. Available: https://doi.org/10.1007/BF01744697 19 [16] THE NASTRAN THEORETICAL MANUAL, NASA, Jan 1981. 13

[17] O. Sigmund, “A 99 line topology optimization code written in matlab,” Structural and multidisciplinary optimization, vol. 21, no. 2, pp. 120–127, Apr 2001. [Online]. Available:

https://doi.org/10.1007/s001580050176 20

[18] F. Song, K. Newman, C. Yang, and S. R. Lee, “Investigation of lead-free bga solder joint reliability under 4-point bending using pwb strain-rate analysis,” in 2009 11th Electronics Packaging Technology Conference. IEEE, Dec 2009, pp. 863–868. [Online]. Available:

https://doi.org/10.1109/EPTC.2009.5416422 3

Appendix A