IMS 2015 Workshops

Direct Extraction of FET Circuit Models from Microwave and Baseband Large-Signal Measurements for Model-Based Microwave Power Amplifier Design


Patrick Roblin, The Ohio State University
David E. Root, Keysight Technologies



Leveraged by the availability of nonlinear vector network analyzers (NVNA), the efficient extraction of circuit-based FET models including memory effects directly from large-signal microwave measurements is now possible and has been successfully demonstrated. There is now a comprehensive body of work published on this emerging topic. This workshop will bring together many of the world experts in this field to present the results already demonstrated by this new approach and outline the outstanding challenges remaining ahead. This workshop will start with a review of the progress in the field, and include tutorials on key extraction topics. The focus of this workshop is to report on advanced techniques for the extraction and validation of circuit-based device-models using large-signal microwave measurements with NVNAs. Model-based nonlinear embedding design technique that enables efficient power amplifier (PA) designs starting from a desired intrinsic PA mode will also be reviewed as an emerging application.,As wireless communication keeps expanding and various RF front end technologies are competing for these new markets, continuing challenges are arising for the development of circuit-based device models for power microwave field-effect transistors (FET) that are able to predict the device performance under realistic large-signal operating conditions. Specifically, device models must account for the actual device response when the FETs are excited by modulated RF communication signals exhibiting a high peak to average power ratio. Under such conditions some memory effects such as (1) self-heating and (2) traps in GaN HEMTs, which are intermittently activated by high instantaneous voltages, are observed to profoundly modify the device performance at low average power.,Circuit-based device models based on a physical topology remain of great importance as they provide the means to monitor the load-lines at the current-source reference planes in order to verify the transistor's mode of operation in power amplifiers (PA). They also play an essential role in the promising model-based nonlinear embedding design technique that enables efficient design starting from a desired intrinsic PA mode. However, reasonable, accurate and realistic models are required for successful model-based PA design. Given the high peak to average power ratio of modern RF communication signals, various low-frequency memory effects (self-heating, trapping) are known to be excited by intermittent high instantaneous voltages at peak power. These memory effects can in turn substantially modify the subsequent response of the device under large-signal operation at average power. Much effort has been placed in finding new ways to efficiently characterize and model these effects. This includes direct model extraction from (1) modulated large-signal microwave measurements with NVNAs and (2) various low-frequency large-signal measurement techniques. This workshop will review these various advanced measurement and modeling techniques for the efficient extraction of realistic transistor models as well as present examples of application to power amplifier design.


  1. Direct extraction of nonlinear FET IV and QV functions from time-domain large-signal measurements
    Paul Tasker, - Cardiff University, Cardiff UK

  2. Parasitic extraction and the intrinsic-element bias-dependence of GaN devices
    Apolinar Reynoso-Hernandez - CICESE, Baja de California, Mexico

  3. Extraction of FET nonlinear models by the dynamic-bias measurement technique
    Dominique Schreurs - K. E. Leuven, Belgium

  4. Extraction of ANN model for SOS MOSFET from real-time active-load-pull measurements
    Patrick Roblin - The Ohio State University, Columbus, OH

  5. Advanced GaN and GaAs transistor evaluation and transistor modeling using combined small- and large-signal VNA & NVNA microwave measurements
    Iltcho Angelov - Chalmers University of Technology, Sweden

  6. NVNA-based artificial neural network (ANN) model generation for III-V FETs including traps and thermal memory effects: DynaFET
    David E. Root - Keysight Technologies, Santa Rosa

  7. Self-heating and trap characterization and simulation for large signal GaN transistor modeling
    Raymond Quéré - XLIM Institut de Recherche, Limoges, France

  8. Characterization of Dynamic Intrinsic Cells for Scalable FET Circuit Models
    Antony Parker - Macquarie University, Sydney