Jan Michal

Electronic circuit design using multiobjective optimization

This paper presents a variation and extension of a previously existing method for multiobjective optimization known as goal attainment method (GAM). The method GAM is in this research combined with a mechanism that automatically provides a set of parameters (weights, coordinates of the reference point) for which the method generates noninferior solutions uniformly spread over a suitably chosen part of the Pareto front. The resulting set of solutions is then presented in a graphic form to the designer so that the solution representing the most satisfactory tradeoff can be easily chosen. The whole algorithm was implemented as a program and tested on two RF design examples (an LNA and a power amplifier), whose optimization results are also presented in the paper.

Multiobjective optimization with an asymptotically uniform coverage of Pareto front

This paper suggests an enhancement of an existing method for the multiobjective optimization known as GAM (goal attainment method). In our proposal, the GAM algorithm is combined with a mechanism that automatically provides a set of parameters (weights, coordinates of the reference point) for which the method generates noninferior solutions uniformly spread over a suitably selected part of the Pareto front. The resulting set of solutions is then presented in a suitable graphic form so that the solution representing the most satisfactory tradeoff can be easily chosen. The whole algorithm was implemented as a program and tested on several RF design examples (video, low-noise, and power amplifiers), whose optimization results are also presented. For a comparison, the first design example was also solved by another method known as WMM (weighted metrics method).

Multi-objective optimization of a low-noise antenna amplifier for multi-constellation satellite-navigation receivers

Although the major parts of function blocks for the satellite navigation receivers are fully integrated in a CMOS chip in most cases, it is convenient to create an antenna preamplifier as a separate circuit based on a low-noise pHEMT. Such an RF front end can be strongly optimized to attain a trade-off between the noise figure and transducer power gain. Furthermore, as all the principal navigation systems (GPS, GLONASS, Galileo, and Compass) work in similar frequency band (roughly from 1.1 to 1.7 GHz), it is reasonable to create this low-noise preamplifier for all of them. In the paper, a sophisticated method of the amplifier design is suggested based on multi-objective optimization. First, an extraction of pHEMT model parameters was performed, including comparisons among several models. The extraction was carried out by our original three-step robust identification procedure based on a combination of meta-heuristic and direct optimization methods. Second, a substantial improvement of a standard method for the multi-objective optimization is outlined. Third, the equations of passive elements of the circuit (including transmission lines and T splitters) were carefully defined using frequency dispersion of their parameters as Q, ESR, etc. Fourth, an optimal selection of the amplifier operating point and essential passive elements was performed using the previously improved goal attainment method. Finally, the s-parameters and noise figure of the proposed preamplifier were measured, and the third-order intermodulation products were also checked.

Multiobjective optimization for RF low noise amplifier design

A modified semiautomatic multiobjective optimization method based on an asymptotically uniform coverage of reference set in combination with goal attainment method was proposed. The method was successfully tested on the multiobjective optimization of an RF low-noise antenna preamplifier.

A new assessment method of pHEMT models by comparing relative errors of drain current and its derivatives up to the third order

At present, there are relatively more precise pHEMT models available for computer-aided design, and they are frequently compared to each other. However, such comparisons are mostly based on absolute errors of drain-current equations and their derivatives. In the paper, a novel method is suggested based on relative root-mean-square errors of both drain current and its derivatives up to the third order. Moreover, the relative errors are then relativized to the best model in each category to further clarify obtained accuracies of the drain current and its derivatives. Furthermore, one our older and two newly suggested models are also included in the comparison with the traditionally accurate Ahmed, TOM2 and Materka ones. The assessment is performed using measured characteristics of a 110 GHz pHEMT. Finally, a usability of the models of the higher-order derivatives is illustrated using an IP3 computation/measurement of a multi-constellation receiver for a satellite navigation with ATF-54143.

A new criterion for stability assessment of the microwave pHEMT-based low-noise amplifiers

At present, the foremost low-noise amplifiers are based on usage of the pseudomorphic high-electron-mobility transistors (pHEMTs). However, in many cases, the amplifiers are often constructed at the limits of absolute stability or even in the region of potential instability because only such solutions give requested circuit properties as a trade-off between the transducer power gain and noise figure. The stability properties are mostly checked by the classical criteria such as the Rollett conditions, μ-factor etc. In the paper, a novel additional very efficient criterion is proposed, finding the most critical couple of poles and evaluating the ratio of imaginary and real part. The efficiency of the method is demonstrated on a low-noise antenna preamplifier for a multi-constellation satellite-navigation receiver based on an ATF-54143 pHEMT.

Multiobjective optimization of input low noise amplifier for common GPS/Galileo/GLONASS/Compass satellite navigation system receiver

As all the four main navigation systems (GPS, Galileo, GLONASS, and Compass) work in similar frequency bands, it is reasonable to create a common input low noise amplifier for all of them. Although the whole chip including a lot of correlators and other digital circuits is quite complicated, a common low noise antenna preamplifier operating at the frequencies from 1.1 to 1.7 GHz could be quite simple and efficient. We have proposed finding a compromise between the amplifiers amplification and noise figure under several natural constrains by multiobjective optimization. Moreover, we have utilized our enhancement of known optimization algorithm (the goal attainment method) to improve its efficiency, which led to finding a very good design tradeoff for the amplifier.

Multi-objective optimization of radio-frequency front-ends

This paper proposes an enhancement of a standard method for the multi-objective optimization known as GAM (Goal Attainment Method). In our modification, the GAM algorithm is combined with a procedure that automatically provides a set of parameters (weights, coordinates of the reference point) for which the method generates non-inferior solutions uniformly spread over a suitably selected part of the Pareto front. The resulting set of the solutions is then presented in a convenient graphic form so that the solution representing the most satisfactory trade-off can easily be chosen. The whole algorithm was implemented as a program and tested on various circuit design examples. A typical one was a four-dimensional multi-objective optimization of an RF video front-end. Therefore, this task is thoroughly described in the paper as a representative demonstration of the capability of the program, including a sophisticated quasi four-dimensional graphic for selecting the trade-off. Moreover, a comparison of our modification with WSM (Weighted Sum Method) is also performed.

An implementation of the circuit multiobjective optimization with the weighted sum strategy and goal attainment method

A robust implementation of the weighted sum strategy and goal attainment method is described, and a comparison of both methods is performed. While the weighted sum strategy is preferable if the problem is known to be convex, or if a smaller number of controlling parameters is advantageous, the goal attainment method provides a better control over accessed parts of the Pareto front, but only at the cost of higher demands on the designer as well as on the used computational power. The robustness is increased by means of a two-stage modified single-objective method. The use of both methods is demonstrated on a practical example of a video amplifier design involving a four-dimensional optimization. Obtained results are evaluated and discussed regarding average correlation and inferiority. Furthermore, an interpolative approach to generating graphical output suitable for multidimensional problems is suggested and used for the four-dimensional objective space of the optimized circuit.

An Accurate Sparse-Matrix Semisymbolic Algorithm for Analyzing Distributed Microwave Circuits

An optimal pivoting strategy for the reduction algorithm transforming the general eigenvalue problem to the standard one is presented for both full- and sparse-matrix procedures. The algorithm increases the precision of the semisymbolic analysis especially for the large-scale microwave circuits. The accuracy is furthermore increased using longer numerical data. First, the long double precision is utilized. Further, the application of a suitable multiple-precision arithmetic library is suggested. Finally, using the longer numerical data to eliminate possible imprecision of the multiple eigenvalues is evaluated. The algorithm is demonstrated by estimating the frequency of a distributed microwave oscillator, and by estimating the bandwidth of a distributed microwave amplifier.