Problem Solving 101 with Gunter Kompa

At Artech House, we ask our authors what problems their books can help readers solve. In this series, we share what our authors aim to do in their writing. Read on to see what Gunter Kompa, who wrote Parameter Extraction and Complex Nonlinear Transistor Models:

  • For the first time it is shown how to treat microwave and RF transistors with distributed parasitic networks. Such complex nonlinear models appear with large gate-periphery transistors, typically high power multifinger AlGaN/GaN devices, comprising interelectrode capacitive coupling. Complex models are also relevant for small-size transistors (number of gate fingers ≤ 2) because of increasing stray capacitance coupling with increasing frequency at about 80 GHz or above.
  • The book offers practical guidance to determine the model parameters of a distributed parasitic network of a nonlinear transistor model.
  • The book introduces in measurement techniques to characterize microwave and RF transistors up to 100W. Characterization covers single-/two tone and IMD measuements. Measurement setups are based on modular design. Source- and load-pull facilities for baseband (zone 0) and higher zones are described, realized and validated.
  • The book gives a comprehensive survey of optimization methods commonly used for model parameter extraction. The algorithms are described and evaluated on the basis of previous experiences. The author presents the “leaping Simplex” which was developed over years in his research group. Leaping Simplex is an extension of the classical local nonlinear Simplex algorithm and is able to excape from local minima.
  • Estimation of parameters values used as starting model parameter value vector in an optimization process is suggested on two ways. Well defined parameter values can be obtained on the basis of available device physical data or from low-frequency measurements.
  • The book points out that the model parameters correlate with each other. This means that the starting parameter values for optimization should not be independently chosen (using e.g. different bias points) but derived from a set of measurements in a single bias point.
  • A new “cold reverse” measurement method is proposed to take care for the distributed parasitic equivalent network.
  • Various standard FET models have been tested on the basis of measured data of a large gate-periphery AlGaN/GaN power FET. Capabilities and limitations are discussed. Most recent advances in GaN-based transistors are emphasized.

For more information and to order, click here.

Leave a Reply

Your email address will not be published. Required fields are marked *