“…Let us first consider the case that the circuit parameters lead to a saturation-like characteristic for i nl (v), as shown in Figure 4(a). Thus, in absence of injection, that is, v in = 0, we have i(v,v in ) = i nl (v) % À I 0 sgn (v) [16][17][18][19], where I 0 can be estimated starting from the measured or simulated voltage-current characteristic of the active part of the ILFD, that is, i(v,v in ). Then, if a signal v in 6 ¼ 0 is applied into the gate of the tail device, the voltage-current characteristic is modified and the function g(v) defined in (5), which takes into account the dependence of the ILFD nonlinearity from the input signal, can be estimated approximating the derivative through a finite difference between two measured or simulated voltage-current characteristics obtained for small and constant values of v in , that is,…”