Skip to Main Content
A new statistical parameter extraction methodology which translates actual process variations into SPICE model parameter variations is presented. This methodology uses E-T data to extract SPICE model parameters and guarantees that its extraction results match measured variations. We have applied this methodology to an industrial 0.5 /spl mu/m process. Excellent, overall I-V curve fit for multiple device geometries is achieved. A compact statistical circuit design technology that improves upon the typical/worst/best case methodology is also presented.