NEW TUNER CHARACTERIZATION AND GAIN COMPENSATION TECHNIQUES FOR ON-WAFER MICROWAVE NOISE MEASUREMENT

Abstract

<p>Accurate characterization of a noisy device starts with an accurate measurement system. Measurement uncertainty and error continues to be a challenging subject as technology advances. The conventional method to noise characterization of on-wafer devices is to determine its noise parameters. To extract the noise parameters of an unpackaged device involves a sophisticated measurement system and calibration procedure. This thesis presents a new automated on-wafer noise measurement system based on Labview 8.5.1 which is used to examine measurement uncertainty for noise parameter extraction. The software program can be used and customized for a wide range of on-wafer noise measurements. This thesis covers the design and operation of the measurement system, which is then used to analyze measurement uncertainty.</p> <p>Measurement uncertainty can be due to various sources from environmental surroundings to instrument settings and the components of the system itself. In many scenarios, inaccuracies are random and cannot be completely resolved. In this thesis, a new tuner characterization technique that improves source tuner characterization is presented. Additionally, a new gain compensation technique is applied to measured noise powers that attempt to improve noise parameter extraction accuracy is proposed. The tuner characterization technique is evaluated against a current industry solution and the affects of the gain compensation technique is evaluated using a newly developed figure of merit. This research work concludes that a direct noise power correction is valid and necessary to further improve noise parameter accuracy. However, the proposed technique when applied resulted in minimal change to the overall noise parameter data. It is found that that source termination selection and total points used for fitting continue to be the major source of uncertainty for noise parameter accuracy.</p>