We present algorithms for reducing large circuits, described atnSPICE-level detail, to much smaller ones with similar input-outputnbehavior. A key feature of our method, called time-varying Pade (TVP),nis that it is capable of reducing time-varying linear systems. Thisnenables it to capture frequency-translation and sampling behavior,nimportant in communication subsystems such as mixers andnswitched-capacitor filters, Krylov-subspace methods are employed in thenmodel reduction process. The macromodels can be generated in SPICE-likenor AHDL format, and can be used in both time- and frequency-domainnverification tools. We present applications to wireless subsystems,nobtaining size reductions and evaluation speedups of orders of magnitudenwith insignificant loss of accuracy. Extensions of TVP to nonlinearnterms and cyclostationary noise are also outlined
展开▼
