What is the Curtice model for a GaAs FET and when is it used?

The Curtice model is one of the earliest nonlinear transistor models developed specifically for GaAs MESFETs and FETs. It models the drain current using polynomial or cubic equations: (1) Curtice quadratic model: I_DS = beta × (V_GS - V_TO)² × (1 + lambda × V_DS) × tanh(alpha × V_DS). Where: beta = transconductance parameter, V_TO = threshold voltage, lambda = channel-length modulation parameter, and alpha = linear region parameter. This is a simple, computationally efficient model. The quadratic dependence produces a linear g_m (constant IM3 vs amplitude). (2) Curtice cubic model: I_DS = (A0 + A1×V1 + A2×V1² + A3×V1³) × tanh(gamma × V_DS). Where V1 = V_GS(t - tau) (delayed gate voltage, capturing the transit time), and A0-A3 are polynomial coefficients. The cubic term captures the g_m compression (the transconductance decreases at high V_GS). (3) Statz model: a variation of the Curtice approach with improved linear-to-saturation transition. When to use: (a) Simple GaAs MESFET circuits (switches, simple amplifiers at < 10 GHz). (b) Quick estimates where a full Angelov extraction is not justified. (c) Legacy designs where Curtice models have already been validated against hardware. When NOT to use: (a) GaN HEMTs (the Curtice model does not include dispersion/trapping effects essential for GaN). (b) Accurate harmonic simulation (the Curtice charge model is less sophisticated than Angelov, leading to errors in harmonic prediction). (c) Wideband design (the simple charge model in Curtice does not capture the frequency-dependent capacitance behavior accurately). (d) Any design requiring high-fidelity nonlinear prediction (PA optimization for P_out, PAE, ACLR). In modern RF design: the Angelov model has largely superseded the Curtice model for all HEMT applications. The Curtice model is still found in older simulation libraries and textbooks, but new designs should use Angelov or the foundry-provided model.