Signal integrity analysis of high-speed interconnects by using nonconformal domain decomposition method

Yang Shao, Zhen Peng, Jin Fa Lee

Research output: Contribution to journalArticlepeer-review

Abstract

3-D interconnect techniques in multilayer very large scale integration design, such as stacked layers and various chip-stacking systems, have paved the way for the improvement of integrated circuit density and operation speed. However, these techniques often accompany with impedance discontinuities that induce signal integrity (SI)/power integrity and electromagnetic interference effects. Consequently, SI analysis serves as one of the key guidelines of chip-package-board design in deep submicrometer technology. However, full-wave numerical analysis of 3-D high-speed and high-density interconnects is also considered to be a great challenge. In this paper, a nonconformal domain decomposition method with second-order transmission condition for SI analysis of high-speed interconnects on a multiscale multilayer printed circuit board is presented. The accuracy and robustness of the proposed method are first witnessed by modeling a six-layer two-trace differential pair. Then we study a 14-layer 16-trace interconnect model to demonstrate the efficiency of the method. Eye diagrams of the two models are also presented in the time-domain SI analysis.

Original languageEnglish (US)
Article number6093738
Pages (from-to)122-130
Number of pages9
JournalIEEE Transactions on Components, Packaging and Manufacturing Technology
Volume2
Issue number1
DOIs
StatePublished - Jan 2012
Externally publishedYes

Keywords

  • Domain decomposition method
  • Maxwell's equations
  • second-order transmission condition
  • signal integrity

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Industrial and Manufacturing Engineering
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Signal integrity analysis of high-speed interconnects by using nonconformal domain decomposition method'. Together they form a unique fingerprint.

Cite this