Modeling And Parameters Extraction Of On-chip Spiral Transformers For RFICs Design

With the rapid development of millimeter-wave circuits, CMOS RF IC design technology is constantly improving. Currently, System on chip (SOC) of Very Large Scale Integration(VLSI) has become a major trend in circuit design. Single-chip passive components, including inductors, capacitors, transformers and baluns, have been widely used in RF IC design, especially on-chip spiral transformer, it is used to implement functions such as impedance conversion, resonant load, low-noise feedback, bandwidth enhancement, and differential-to-single conversion. Therefore, it is the main goal to design high performance and low cost on-chip spiral transformers.Although manufacture process of monolithic transformer has been made great progress, the model of on-chip spiral transformer has been largely overlooked. Existing technology libraries rarely have equivalent circuit models for on-chip spiral transformer, so lots of time is wasted on optimization and design of transformer with great inaccuracy. Therefore, it is significant to establish an equivalent circuit model suitable for RFIC on-chip spiral transformer.The geometric characteristics, equivalent circuit model and parameter extraction method of on-chip spiral transformer are researched in this paper, the main contents include:1) the theory and frrequently-used performance parameters of on-chip spiral transformer; 2) analyze the merits and demerits of threee different structures on-chip spiral transformer and the circuit requirements of each other; 3) research the influence of four geometric parameters including metal width, pitch, diameter and number of laps on the performance of on-chip spiral transformer; 4) the basic process of modeling the on-chip spiral transformer and the de-embedding technology in parameters extraction process; 5) improve the equivalent circuit model of on-chip stacked transformer and present a improved parameter extraction method, which combines the analytical approach and the empirical optimization procedure, and overcomes the drawbacks of the both methods while retaining their advantages.