Xin Li Carnegie Mellon University

Robust Analog/RF Design with Projection-Based Posynomial Modeling

As IC technologies are scaled to finer feature sizes and circuit applications move to higher frequency bands, analog/RF circuit design faces several new challenges. First, device models have become increasingly complex in order to characterize the physical behavior of transistors at high frequencies. Secondly, at these frequencies, parasitic couplings become more important and more complex. Finally, but perhaps most importantly, with sub-wavelength photo-lithography, process variations become a critical issue and significantly impact the overall circuit performance.

In this talk we describe a RObust Analog Design tool (ROAD) for post-tuning analog/RF circuits. Starting from an initial design derived from hand analysis or analog circuit synthesis, ROAD automatically extracts posynomial performance models based on transistor-level simulations, and optimizes the circuit by geometric programming. A key feature of ROAD is the formulation of all design constraints with consideration of process variations. As such, ROAD can converge to a robust design and facilitate the tradeoff between yield and performance. In order to reduce the computation cost, we propose a projection-based approach for posynomial modeling. We show that quadratic posynomials are invariant (i.e. remain posynomials) under the rank-one projection. In addition, a novel implicit power iteration algorithm is developed to find the optimal projection space and extract the posynomial coefficients with robust convergence. The efficacy of ROAD is demonstrated by several circuit examples in IBM BiCMOS 6HP 0.25µm process.

Bio
Mr. Xin Li received the B.S. and M.S. degrees in electronics engineering from Fudan University, Shanghai, P.R. China, in 1998 and 2001, respectively. He is currently working toward the Ph.D. degree at Carnegie Mellon University, Pittsburgh, PA. His research interests include synthesis, modeling and simulation for RF/analog and mixed-signal systems.