In this paper, we propose an approach for the analysis of power supply noise in the frequency domain. Power supply network is modeled as a linear time invariant (LTI) system comprising of the parasitic resistance (R), inductance (L), and capacita.nce (C). Each gate (or cell) is modeled as a time-varying current source. Voltage fluctuation caused by t he switching events is calculated based on the effective impedances seen by the corresponding current sources and the spatial correlation between the nodes of the power network. Superposition is applied to the LTI system to obtain the overall noise spectrum at any node of the power supply network. Inverse Fast Fourier Transformation (IFFT) is then performed on the frequency domain noise spectrum to obtain the time domain noise waveform. Given the POWER/GROUND network and the time-varying current sources derived from an event-driven simulation based on primary inputs, our approach can produce accurate noise waveforms while being two orders of ma.gnitude faster than HSPICE.
Date of this Version