Title :
Evidence for the existence of a useful periodic steady-state algorithm for nonlinear circuits
Author :
Rumin, Nicholas C.
Author_Institution :
Dept. of Electr. Eng., McGill Univ., Montreal, Que., Canada
fDate :
11/1/1988 12:00:00 AM
Abstract :
Four algorithms for computing the periodic steady state of nonlinear circuits have been compared with respect to their convergence properties. In particular, the gradient and extrapolation algorithms were compared by implementing them in the same circuit-analysis program. The results obtained on nonautonomous test circuits containing between two and fourteen nonparasitic storage elements indicate that the extrapolation algorithm is more efficient. Comparison of the experimental data with usable published results on the Newton algorithm suggests that it is the most efficient one. However, its memory requirements cast doubt on its usefulness for large circuits. Insufficient data is available on the recently proposed Newton-like two-frequency algorithm; however, it appears that, for highly nonlinear periodic circuits, its convergence properties may be comparable to those of the extrapolation algorithm
Keywords :
circuit analysis computing; convergence; nonlinear network analysis; Newton algorithm; circuit-analysis program; convergence properties; extrapolation algorithm; gradient algorithm; large circuits; memory requirements; nonlinear circuits; periodic steady-state algorithm; two-frequency algorithm; Analytical models; Circuit analysis; Circuit simulation; Circuit testing; Convergence; Equations; Extrapolation; Nonlinear circuits; Steady-state; Time domain analysis;
Journal_Title :
Circuits and Systems, IEEE Transactions on