A comparative analysis of mTRL and 2X-Thru de-embedding methods

Abstract

In high-frequency measurements, accurate characterization faces challenges arising from parasitic effects introduced by the measurement setup. These effects, which include artifacts from test fixtures, cables and connectors, can significantly impact the precision of measurements. The standard method for S-parameter characterization performed by a Vector Network Analyzer (VNA) in the frequency domain. Typically, a Device Under Test (DUT) is mounted on a Printed Circuit Board (PCB) to establish a connection with the VNA. It is crucial to eliminate any artifacts introduced by the test fixture to obtain the actual performance of the DUT. The de-embedding process aims to eliminate these parasitic effects and reveal the intrinsic electrical properties of the DUT. This study comprehensively analyzes two distinct de-embedding techniques: the Multiline Thru-Reflect-Line (mTRL) method and the 2X-Thru fixture removal. The methods are compared regarding calibration standard requirements, de-embedding methodology, fixture removing accuracy, and de-embedding results. The comparison reveals that the accuracy of the mTRL method strictly depends on the impedance variation between calibration standards. Conversely, the 2X-Thru with impedance correction, on the other hand, is more susceptible to impedance variation since it relies on a single calibration standard, which also eases the measurement routine and reduces the complexity. Self-test fixture de-embedding was performed for both calibration methods to validate the calibrations, with subsequent checks on the transmission coefficients of residuals. In both calibration methods, the de-embedding results of DUT were compared by evaluating transmission reflection coefficient parameters and the TDR responses of the de-embedded structure. The calibration was verified using the test structure recommended by the IEEE P370 Electrical Characterization of Printed Circuit Board and Related Interconnects at Frequencies up to 50 GHz standard, and the fabrication tolerances of the implemented de-embedding structures were evaluated. One important aim of this study is to contribute to the development of open-source de-embedding tools since the validity and accuracy of commercial tools cannot be thoroughly evaluated for academic purposes. In this context, the study also shares results obtained from commercial applications. In the IC characterization, it is essential to address the parasitic effects arising from bond wires. Beyond on-wafer measurements, a product application involves packaging, necessitating consideration of the impact of bond wires on signal integrity. In that respect, utilizing both the mTRL and 2X-Thru de-embedding methods, a GSG bond wire model was developed. The extraction and application results of the proposed bond wire model are presented. As a summary, this thesis provides a detailed explanation of the methodologies employed in the mTRL and 2X-Thru methods, followed by a comprehensive comparison of the de-embedded results obtained from both approaches. Furthermore, a comparison has been made between results obtained from open-source tools and those from commercial applications. Finally, a GSG bondwire model has been developed using the both mTRL and 2X-Thru de-embedding methods.