Modeling and optimization of power distribution network using response surface methodology for power integrity analysis

Abstract

Complex semiconductor packages face Power Integrity (PI) issues due to high speed operation and denser geometries. DC and AC analysis of the power distribution network is critical to ensure that the electrical response parameters are within the specified tolerance range. Failing this can lead to device malfunction and therefore require multiple design re-spins. This paper proposes a novel methodology for predictive analysis of power integrity (PI) performance of power distribution networks (PDN). In the proposed technique, Plackett Burman Design of Experiments (PB-DoE) and Response Surface Modeling (RSM) are used to develop an empirical model for predicting response parameters such as DC drop voltage, current density and impedance of the PDN at the early stage of the design cycle. The empirical model developed in this work is validated using Mentor Graphics HyperLynx PI tool for the developed test case with identical input design parameters. The response parameters obtained from the analytical model and the simulator give close agreement with an error of 6%. The process improvement optimization method is successfully implemented using the response surface model to improve the performance of the system. The suitability of the proposed technique is verified over the frequency range of 1 MHz to 1 GHz.