Automated Transformer Selection for RFIC Design: Accelerating Development with a Comprehensive Database

Abstract

The design of transformers, a key component of radio frequency integrated circuits (RFICs), is traditionally carried out through an iterative process involving extensive electromagnetic simulations. While process design kits (PDKs) offer tools based on interpolation or fitting equations to simplify parameter estimation, these tools are restricted to standard geometries, leaving designers to manually simulate and optimize custom designs. This approach is inefficient and resource intensive. This paper proposes an automated process to generate a database containing the physical and electrical parameters of a wide range of transformers. This database is part of a tool designed to efficiently identify the desired transformer. To evaluate the tool's effectiveness in reducing the time required for design, a millimeter-wave (mm-Wave) 69.4-74.2 GHz differential low-noise amplifier (LNA) is designed using GlobalFoundries 45 nm silicon-on-insulator (SOI) technology. This circuit demonstrates a noise figure (NF) of 4.1 dB, a gain of 10.1 dB, an input third-order intercept point (IIP3) of -10.78 dBm, and a power consumption of 4.7 mW from a 0.406 V DC supply. Moreover, the simulated performance achieves these specifications within a highly compact area of 0.12 mm2. The transformer selection process for the circuit takes only a few seconds, whereas the conventional method of manual transformer design and electromagnetic simulation would require a significantly greater amount of time.