Variability Analysis of Crosstalk among Differential Vias using Polynomial-Chaos and Response Surface Methods
The variability of crosstalk due to changes in differential and ground via configurations is studied in this paper. The polynomial-chaos method and the response surface method are adopted to mathematically model the variability. One goal of the work is to exploit the obtained models to locate the optimal response that both meets the performance requirement and remains robust to geometry variations due to manufacturing tolerances. Both methods correlate well with simulations and show great capabilities in practical applications for optimization and sensitivity analysis. The other goal of the work is to compare the two methods in terms of mathematical theories, sampling schemes, postprocessing, accuracy, efficiency, and limitations. Details of the comparison is given through this paper and a summary table is included in Section V.
Y. Wang et al., "Variability Analysis of Crosstalk among Differential Vias using Polynomial-Chaos and Response Surface Methods," IEEE Transactions on Electromagnetic Compatibility, vol. 59, no. 4, pp. 1368 - 1378, Institute of Electrical and Electronics Engineers (IEEE), Aug 2017.
The definitive version is available at https://doi.org/10.1109/TEMC.2017.2682169
Electrical and Computer Engineering
Electromagnetic Compatibility (EMC) Laboratory
National Science Foundation (U.S.)
Keywords and Phrases
Crosstalk; Surface properties; Geometry variations; Manufacturing tolerances; Mathematical theory; Optimal response; Performance requirements; Polynomial chaos; Response surface methods (RSM); Variability analysis; Sensitivity analysis; Differential vias; Latin hypercube (LH); Optimization; Polynomial chaos (PC); Worst-case crosstalk
International Standard Serial Number (ISSN)
Article - Journal
© 2017 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.
01 Aug 2017
This work was supported by the National Science Foundation (NSF) under Grant IIP-1440110 and Grant 1332152.