CAD for microelectronics
Modeling of analog circuits and devices
The electronics industry increasingly demands circuits operating in extreme conditions regarding power consumption, operating frequency and supply voltages, requiring the electronic devices to operate near their technological limits. The performance of analog and RF circuits is extremely dependant on the behavior of these devices. Therefore, in order to design circuits able to comply with the industry requirements, it is necessary to know their physical behavior as well as to have accurate and efficient computational models. this research line focuses on the development and implementation o semiconductor devices for circuit simulation as well as the modeling of basic analog circuits. Model implementation is carried out using hardware description languages, allowing their use with most circuit simulators currently in the market.
Integrated passive devices
In addition to solutions in terms of topology and physical implementation, the performance of RF blocks depends largely on the quality of the integrated passive components. Traditionally, the design kits of silicon-based technologies provide models for resistors, capacitor and inductors. For inductors, however, the modeled geometries and the realizable range of values are relatively limited. Additionally, transmission lines structures and transformers, which may be essential in certain types of circuits integrating matching networks, feedback loops, single-ended to differential mode conversion (balun) or power combining. It is therefore often the responsibility of the designer to develop, model and size such components, in order to obtain efficient solutions in terms of performance and silicon area. Therefore, this research line covers the study, design, electromagnetic simulation and modeling of passive components such as transmission lines, inductors, transformers, baluns and power combiners in advanced silicon technologies.
Numerical methods for RF and microwave circuit analysis
Computer aided design of modern mobile communication systems demand numerical simulations which impose severe restrictions on traditional numerical methods for electronic circuit analysis. While the harmonic balance method is not suitable for this application, since it is limited to steady state analysis (and thus for periodic excitations), the transient simulation method is capable of performing this simulation. However, it requires an enormous computational effort. In this context, several techniques known as the envelope method have been recently proposed, which, through a clever combination of analyzes in time and frequency domains, exploit the bandpass behavior of the signals involved and thus are able to perform the necessary simulations with a much lower computational effort compared to the transient method, based solely on the time domain. The objective of this research line is the comparative study of different envelope methods available in the literature and their co-simulation with systems simulators.
Numerical and analytical methods for RF system simulation
There is a number of numerical and analytical methods for the definition, specification and validation of RF transceiver systems that represent different levels of abstraction and performance metrics. Such systems can be seen as a sequence of blocks (black boxes) which act on a signal carrying information and whose quality is degraded along the system. Metrics in the frequency domain, such as signal to noise ratio (SNR) and signal to noise and distortion (SNDR), and metrics in the time domain, such as the binary error rate (BER) and packet error rate (PER) are defined at different levels of abstraction and determine the maximum deterioration to which the signal can be submitted. Analytic techniques (Friss formula for equivalent noise figure, theoretical curves of BER x Eb / N0) allow a first sizing of the blocks. Optimization techniques may be implemented aiming, for example, at minimizing energy consumption for a given performance. Subsequently, computational tools for system simulation are necessary to obtain more accurate results. The computational complexity of these tools can be reduced by employing behavioral models for the representation of functional blocks and exploting the bandpass characteristic of RF systems.This translates, in the time domain, to the complex envelope description of the signals, and, in the frequency domain, using the notion of power spectral density.This research line aims to study and implement different methods for RF system simulation, as well as their with co-simulation with circuit simulators applying the complex envelope signals. The different levels of abstraction can communicate creating top-down and bottom-up mechanisms, since behavioral models evolve with simulation results and new system specifications are required when the models are updated.