Switch-factor based loop RLC modeling for efficient timing analysis

Yu Cao, Xiaodong Yang, Xuejue Huang, Dennis Sylvester

Research output: Contribution to journalArticle

6 Scopus citations

Abstract

Timing uncertainty caused by inductive and capacitive coupling is one of the major bottlenecks in timing analysis. In this paper, we propose an effective loop RLC modeling technique to efficiently decouple lines with both inductive and capacitive coupling. We generalize the RLC decoupling problem based on the theory of distributed RLC lines and a switch-factor, which is the voltage ratio between two nets. This switch-factor is also known as the Miller factor, and is widely used to model capacitive coupling. The proposed modeling technique can be directly applied to partial RLC netlists extracted using existing parasitic extraction tools without advance knowledge of the return path. The new model captures the impact of neighboring switching activity as it significantly affects the current return path. As demonstrated in our experiments, the new model accurately predicts both upper and lower delay bounds as a function of neighboring switching patterns. Therefore, this approach can be easily implemented into existing timing analysis flows such as max-timing and min-timing analysis. Finally, we apply the new modeling approach to a range of activities across the design process including timing optimization, static timing analysis, high frequency clock design, and data-bus wire planning.

Original languageEnglish (US)
Pages (from-to)1072-1078
Number of pages7
JournalIEEE Transactions on Very Large Scale Integration (VLSI) Systems
Volume13
Issue number9
DOIs
StatePublished - Sep 1 2005

Keywords

  • Clock
  • Data-bus
  • Loop inductance
  • RLC model
  • Slew rate
  • Static timing analysis
  • Switch-factor

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Switch-factor based loop RLC modeling for efficient timing analysis'. Together they form a unique fingerprint.

  • Cite this