A scalable linear model for FETs

Getting to the Heart of the Matter: Considerations for Large-Signal Modeling of Microwave Field-Effect Transistors

“…Except in regions of significant heating and trapping, which will be discussed later, it is possible to choose and refine the access network model and de-embedding process to leave intrinsic active FET parameters that are the same at all frequencies [8], [10]. This is the remarkable feature of the data in Figure 3, which includes surfaces over a wide range of frequency measurements that sit on each other.…”

“…However, a frequency-dispersive element, such as a time delay, will introduce a nonconservative and, hence, nonphysical reactive component. Fortunately, it is possible to extract parameters for a smallsignal model with frequency-independent capacitance and conductance terms, such as those in Figure 2(b), Figure 3, [8], and [10]. One must, as we will explore below, allow for the fact that there will be regions of significant frequency dependence, which result from heating and trapping effects, and then select data from the other bias regions, as shown in Figure 3, from which to extract and verify a dispersion-free smallsignal model.…”

“…For the most part, it is possible to refine or optimize the de-embedding process to achieve this [10]. That is, extract a smallsignal model with frequency-independent capacitance and conductance terms, such as those in Figures 2(b) and 3.…”

History and Evolution of Millimetre-Wave MMICs for Point-to-Point Radio

“…High linear power over a wide bandwidth presents a major challenge for physically smaller LNAs, power amplifiers and mixers. Better linear and non-linear models are needed, with ability to scale parameters so that the geometry of active devices is part of the optimisation process and coupling between more compact structures is taken into account [5]. Cycle times demand first-time success at higher frequencies, putting pressure on modelling and simulation tools.…”