Research On Optimal Design Of Analog Circuits Based On Operating-point Driven

As the sustained development of integrated circuits currently, it has come to the era that the system chips and the age-chip mixed-signal system chips been used skillfully, but the analog IC design is still the main reason for its development constraints, so it is a imminent task to improve their level of design automation. And it is an inevitable way to use optimized measures in analog IC design. This article researches a new design method named operating-point driven formulation in analog integrated circuit design process.General design of traditional optimization design select the size of the circuit device as the design variables, while the operating-point driven formulation select a group of independent device voltage and current on DC operating point as the design variables. This method stem from the manual design process. Due to the changes in the circuit design space, so it is easily to express constraints on the working area of the circuit design(i.e., the constraint equation) and allow the mapping relationship(i.e., the target function) between design variables and design goals easier, besides, it is helpful in optimizing the convergence of the algorithm. Also, as it is familiar to the traditional manual design, so it makes the designers better on understanding the design process and the design results, as a result, it provides possibility for the optimization process comb ined with traditional design experience.The primary problem should be solved in operating-point driven formulation is that how to get the size and a variety of electrical parameters of the device from the DC operating point of the voltage and the current. As the simple first-order model can not meet the accuracy requirements in the design of small MOS transistors in most cases, so this article propose a method using fuzzy logic molding to accomplish this task. The method has two steps, the first step is to obtain a set of sample data using the internal precision MOS model in the Hspice circuit simulator, and the second step is to build the relationship between the DC operating point of the voltage and the current and the device size using the T-S fuzzy logic systems. The T-S fuzzy logic has many advantages such as strong adaptability, without iteration and better applications.We study the MOS transistor circuits using operating-point driven formulation and the multi-objective optimization design using genetic algorithm on the work mentioned above. It is the simply best way to solve the problem in circuit design that the designers select the best choice from a set of conflicting design specifications in the multi-objective optimization Pareto front curve. But the multi-objective optimization requires a lot of computation based on genetic algorithm NSGA-II, especially when we need to call the circuit simulator to calculate the circuit performance. So in this article, we use an method to solve this problem. The first step is establishing a single transistor’s small signal parameters model that is on BSIM3v3 level, the second step is using equation-based device sizing method, in this way, we can quickly obtain a more accurate multi-objective optimal solution and reduce the amount of work.Two optimization example of common CMOS op amp circuit is given in this article, and the results indicates that the solution proposed in this article is valid.