A CAD-oriented non-quasi-static approach for the transient analysis of MOS ICs

A CAD-compatible non-quasi-static MOSFET model

“…5 -12, our model consistently yields close agreement with the numerical solution, while the models of [l] and [3] exhibit several problems besides being inaccurate. The model of [3] shows prominent bump features; the model of [I] produces zero Ib; and Id's of both [l] and [3] models rise sharply in comparison with the numerical solution and turns-on much faster at an earlier time. A tabulation of the rms errors of the calculated currents as a function of the ramp rate is shown in Figs.…”

“…In particular, the rms errors in I, and Id in this model remain at -3% and IO%, respectively, when the input voltage waveform is changed from a slow ramp to an abrupt step. Under the same situation, the rms errors become dramatically larger for other models: 20 % and 140 % for BSIM3 111, and 20 % and 80 % for the Turchetti model [3].…”

“…the improved accuracy of our model in predicting all four terminal currents. Most importantly, the time delay before the drain current becomes nonzero is accurately determined by this model, while the models of [l] and [3] show premature turn-ons. Figs.…”

“…Again, while our model displays close agreement, ref. [3] model yields wrong charge profiles which can actually be negative at times and converge to a wrong profile as time approaches infinity. Fig.…”

“…The approximate solution of c(x,t), however, fails to converge to the @ function in steady state, and can be negative although physically it should be positive. In a fast transient in which the NQS effects are important, the model of [3] yields significant error in all four terminal currents. The model of [3] also produces bump in the source current and wrongly predicts the timing of the turn-on of the drain current when the transistor is biased from cutoff to saturation.…”