COMMENTS ON THE OPTIMUM CMOS TAPERED BUFFER PROBLEM

Two recent papers, one by Li et al.1 and the other by Prunty and Gal, on the optimum CMOS tapered buffer problem are commented on. These papers claim that an equivalent ''short-circuit'' current capacitance should be added to the output capacitance of an inverter to account for the increased propagation delay obtained because of the short-circuit current that flows when a real input waveform is considered instead of an input step voltage. Their reasoning results in an optimum tapering factor that is dependent on the waveform rise and fall times. However, the propagation delay only depends on the load capacitance and the ratio of the input to output transition times. In a fixed-taper buffer these transition times are equal making the optimum tapering factor independent of the ''short-circuit'' current. We also suggest an improved method of determining the optimum tapering factor in practical situations based on circuit simulations.