L-2L de-embedding method with double-T-type PAD model for millimeter-wave amplifier design

Abstract

For millimeter-wave CMOS circuit design, accurate device models are necessary. Especially an accurate de-embedding method is very important. Hence, precise deembedding of pad parasitics is the first and valuable step to achieve accurate device models. In this work, a new pad modeling based on an L-2L de-embedding is proposed. The pad model is derived with an assumption that characteristic impedance of transmission line becomes constant at high frequency. Every device used in an amplifier is characterized with the proposed de-embedding method, and simulation and measurement results well agree with each other up to 110 GHz.

Analysis of the parameter extraction for on-chip transmission lines

“…A set of homogeneous equations can be derived from (8), as follows We can get γ using (6), but there are still four unknowns in (9), including Z c , l 11 , l 12 , l 22 . It can be proved that the exact solution of Z c cannot be obtained [24] according to the homogeneous equations (9). This also proves that the direct use of the measurement data of two (or more) transmission line chips cannot directly solve the characteristic impedance mathematically.…”

“…The traditional open-short de-embedding method has been proven to be unsuitable to ultra-high frequencies due to its incapability for dealing with distributed nature in millimeter-wave frequency range based on assumptions of lumped equivalent circuit and ideal dummy patterns [16,17]. On the other hand, based on cascading distributive assumption, multi-line methods, including L-2L [18,19,20,21], and thru-based approaches such as thru-only [22], have been widely used for on-chip transmission line de-embedding [23,24]. The targets of the transmission line de-embedding include the determination of complex propagation constant and characteristic impedance.…”