Uncertainties have become a major concern in integrated circuit design. Inorder to avoid the huge number of repeated simulations in conventional MonteCarlo flows, this paper presents an intrusive spectral simulator forstatistical circuit analysis. Our simulator employs the recently developedgeneralized polynomial chaos expansion to perform uncertainty quantification ofnonlinear transistor circuits with both Gaussian and non-Gaussian randomparameters. We modify the nonintrusive stochastic collocation (SC) method anddevelop an intrusive variant called stochastic testing (ST) method toaccelerate the numerical simulation. Compared with the stochastic Galerkin (SG)method, the resulting coupled deterministic equations from our proposed STmethod can be solved in a decoupled manner at each time point. At the sametime, ST uses fewer samples and allows more flexible time step size controlsthan directly using a nonintrusive SC solver. These two properties make ST moreefficient than SG and than existing SC methods, and more suitable fortime-domain circuit simulation. Simulation results of several digital, analogand RF circuits are reported. Since our algorithm is based on genericmathematical models, the proposed ST algorithm can be applied to many otherengineering problems.
展开▼
