基于区域化的CMOS模型参数提取方法

A New Extraction Method for CMOS Model Parameter Based on Binning

根据CMOS的可缩放特性,提出了一种基于区域化(Binning)的模型参数提取方法。本方法首先按器件的尺寸划分区间,然后计算参数的缩放系数,最后根据缩放公式确定同一区间内不同尺寸器件的参数值。而传统的CMOS参数提取方法是在一定的区间内提取一个典型的参数值,用它代表整个区间内器件的参数值。与传统方法相比,本方法提高了模型参数的精度。针对一组0.35μmCMOS器件,利用本方法获得的缩放系数计算了区间内几个单管的模型参数,所得到的结果与针对每个单管提取参数的结果吻合得很好,平均误差小于6%。

According to the scalable characteristic of CMOS, a method for extracting CMOS model parameters based on binning is presented in this paper. First, the devices are split into bins according to device size, and then the coefficients of scalable parameters are calculated, parameters for different devices in the bin are decided from the scalable formulas. The traditional method extracts a typical parameter value for a device bin, which is used for the whole bin. Compared with the traditional method, the proposed method improves the accuracy of model parameters. For a group of 0.35μm CMOS devices, parameters for several devices in the bin are computed from scalable parameters through proposed method, which match well with the parameters extracted from single device, the average error is less than 6%.