Context-dependent transformation of Pareto-optimal performance fronts of operational amplifiers

Abstract

Abstract:The use of Pareto-optimal performance fronts in emerging design methodologies for analog integrated circuits is a keystone to overcome the limitations of traditional design methodologies. However, most techniques to generate the fronts reported so far neglect the effect that the surrounding circuitry (such as the load impedance) has on the Pareto-front, thereby making it only realistic for the context where the front was generated. This strongly limits the use of the Pareto front because of the strong dependence between the key performances of an analog circuit and its surrounding circuitry, but, more importantly, because this circuitry remains unknown until the Pareto-optimal front is being used. Since performance front generation is a costly process, this paper proposes that performance fronts for a new context of use of a given circuit can be obtained from fronts that were previously generated under some different conditions. Towards this goal, a transformation methodology for performance objectives of operational amplifiers has been developed. Experimental results for a folded-cascode and a Miller-compensated operational amplifiers show that this is a promising approach to reuse the fronts in multiple contexts.

Implementation issues in the hierarchical composition of performance models of analog circuits

“…The PoFs of the opamps are obtained by transforming a previously generated PoF to these loading conditions using the technique described in [7]. This PoF was generated considering as design variables the widths and lengths of the transistors, the values of the compensation capacitor and resistor and the value of the bias current.…”

“…To generate the PoF of the LP-CT filter, the elements defining the design space are: (a) The design variables characterizing all the passive devices and (b) a previously generated, load-independent PoF of the opamp (following the methodology explained in [7]). At each iteration of the optimization process, the following critical steps are taken: 1) During the mutation of the indexing variables/coordinates, the PoFs of the opamps must be transformed to the load conditions imposed by the passive elements of the w1 w3 w5 w10 w12 w1c w3c w5c lbn l1 l10 l14 l1c l3c filter, which are also variables.…”

“…At each iteration of the optimization process, the following critical steps are taken: 1) During the mutation of the indexing variables/coordinates, the PoFs of the opamps must be transformed to the load conditions imposed by the passive elements of the w1 w3 w5 w10 w12 w1c w3c w5c lbn l1 l10 l14 l1c l3c filter, which are also variables. This is done by using the approach presented in [7]. 2) Once transformed, the indexing variables/coordinates can be established according to the new performances of the opamps.…”

“…If these conditions change when the opamps of the PoF are used in the filter, the PoF must be re-generated as their performances will be different or not valid any more. In [7] a solution is proposed to transform PoFs of opamps to different loading conditions using a hybrid-2 parameter characterization.…”

Application of artificial immune system approach to develop an algorithm for optimizing multi objective problems

“…proposed a paper that solve distributed autonomous robotics system problem using artificial immune system technique [17]. proposed a paper describing that the robustness of a induction motor speed regulation system is weak due to uncertainty of parameters [18].…”