PhD abstract

Differential circuits are widely used in the design of high frequency components mainly because of their better noise immunity. These circuits can be characterized using mixed-mode S parameters (differential- and common-mode S-parameters and cross-mode terms). Furthermore, the trend toward miniaturization and integration of microwave devices increases the need for planar or coplanar microwave integrated circuits such as micro-strip lines or coplanar waveguides. The ungrounded coplanar waveguide structure with all the conductors located on the same side of the substrate eliminates the need for via-holes, and thus simplifies manufacturing and prevents the appearance of some parasitic elements. From the viewpoint of electrical metrology, it is necessary to establish the traceability of the mixed-mode S-parameter measurements to the International System of Units (SI). The Multimode Thru-Reflect-Line (TRL) calibration method, derived from the commonly-used TRL calibration for S-parameter measurements of single-ended circuits, is particularly well suited for this purpose as the standards are traceable via dimensional measurements. The characteristic impedance, which defines the reference impedance of the measurement system, can be achieved from the propagation constants determined during the Multimode TRL calibration and the capacitances per unit length of the transmission line.

We present the first design and realization of Multimode TRL calibration and verification kits using coupled coplanar lines in the “Ground - Signal - Ground - Signal – Ground” configuration on quartz (SiO2), the low-loss substrate, for on-wafer mixed-mode S-parameter measurements from 1 GHz to 40 GHz.

Measurements are performed using two methods: the “one-tier” technique, based on the Multimode TRL calibration procedure, determines and corrects all systematic errors. The “two-tier” approach, in which the Multimode TRL is applied at the second-tier, is applied to measurement data that were partially corrected by the first calibration. The feasibility and the validation of the methods are demonstrated by measurements of matched, mismatched and unbalanced lines and T-attenuators showing good agreement between simulated and measured results.

The propagation of uncertainty can be derived by the calculation of partial derivatives using the Metas.Unclib tool or by the numerical approach based on the Monte Carlo technique. The accuracy of on-wafer S-parameter measurements depends on sources of influence attributed to the measurements and to the imperfections of the standards such as the VNA noise and non-linearity, the cable stability, the measurement repeatability, and the sensitivity in calibration standards’ realization. We focus, first and foremost, on the propagation of uncertainties related to the repeatability of the standards and the device under test measurements to the corrected mixed-mode S-parameters of the mismatched line. The results show that the partial derivatives approach based on an approximation of the first-order Taylor series cannot be accurately used due to the significant influences of non-linear functions in the Multimode TRL algorithm. The Monte Carlo method is then more precise although it requires very long computation time.

Key words

four-port vector network analyzer, differential circuit, multimode TRL calibration, coupled coplanar waveguide, mixed-mode scattering parameters, measurement uncertainty