考虑工艺波动影响的RLC互连统计延时

摘要

Analysis of RLC Statistical delay considering process fluctuation is presented in this paper. Construction of parasitic parameters and moments with process variation is first given, and then a statistical delay model based on Weibull distribution is achieved. The proposed method is also applied to the other available delay such as Elmore, equivalent Elmore and D2M. For the statistical delay model based on Weibull distribution, compared with HSPICE, results show that the maximum error of 50% delay is 0.11%, the maximum error of mean and the average in Monte Carlo analysis is 2.02%.%该文提出了一种考虑工艺波动的统计RLC互连延时分析方法.文中首先给出了考虑工艺波动的寄生参数和矩的构建方法,然后基于Weibull分布给出了RLC互连的统计延时模型.所提方法同样适用于已有的延时模型如Elmore模型,等效Elmore模型和D2M模型.通过对几种模型的比较,表明,基于Weibull分布的RLC互连的统计延时模型是最精确的,和HSPICE相比,50%延时误差最大0.11%,蒙特卡洛分析中的均值和平均偏差误差最大2.02%.