A Measurement‐Based 1‐ to 100‐GHz Lumped Model for On‐Chip Transformers With 3.2% Average Error

ABSTRACT With the rapid development of wireless systems and applications, the area of mm‐Wave (millimeter‐wave) and sub‐THz (terahertz) IC (integrated circuit) has been extensively studied and has achieved remarkable progress. The abundant frequency band resource of mm‐Wave and sub‐THz frequency band has promised superior capability of IC operation. However, mm‐Wave and sub‐THz ICs suffer from a severe gap between theoretical and measured performance. Normally, mm‐Wave and sub‐THz IC design relies heavily on EM (electromagnetic) simulation, which leads to ineffective iterative design due to the complex nature and massive parameters related to secondary effects of devices. Transformers, as one of the most commonly used passive components, have been widely exploited to support various advanced circuit techniques. Therefore, the fast and accurate iterative design of transformers is essential to high‐performance mm‐Wave and sub‐THz IC design. However, on‐chip transformers suffer from inadequate modeling of secondary effects including distribution, coupling, and self‐resonation. In addition, the geometry and surroundings of on‐chip transformers vary from one to another and pose subtle influences on circuit performance, making piecewise EM simulation time‐consuming and ineffective. To surmount design issues related to transformers, this paper proposes an enhanced lumped model accompanied by a measurement‐based parameter extraction and optimization procedure. The proposed model includes notable coupling and distribution effects in the mm‐Wave and sub‐THz bands while keeping a compact structure. An on‐chip transformer is fabricated using 65‐nm CMOS technology to verify the proposed model and procedure. According to measurement results, the proposed model achieves S‐parameter modeling error of less than 3.5% and NF (noise figure) modeling error of less than 0.4% up to 100 GHz.