Abstract The modern wireless communication techniques are aiming on increasing bandwidth and the number of carriers for higher data rate. This sets challenging linearity requirements for RF power amplifiers (PAs). Unfortunately, high linearity can only be obtained at the cost of efficiency. In order to improve the performance of the PA, in-depth understanding of nonlinear behaviour is mandatory. This calls for techniques that can give componentwise information of the causes of the distortion. The aim of this thesis is to develop a technique that can provide such information. This thesis proposes a detailed distortion analysis technique that is based on frequency domain fitting of polynomial models. Simulated large-signal spectra are used for fitting as these contain the necessary information about the large-signal bias point and amplitude range. Moreover, in the frequency domain the delays are easy to compensate, and detailed analysis to any fitted tone can be performed. The fitting procedure as such is simple but becomes difficult in multi-dimensional nonlinearities if the controlling voltages correlate strongly. In this thesis the solvability and reliability of the fitting procedure is increased by numerical operations, model-degree reduction and by using different excitations. A simplified Volterra method is used to calculate the distortion contributions by using the fitted model. The overall distortion is analysed by calculating the voltage response of the contributions of each nonlinearity to the terminal nodes of the device by the use of linear transfer functions of the circuit. The componentwise analysis is performed by phasor presentation enabling the cancelling mechanisms to be seen. The proposed technique is implemented on top of harmonic balance simulation in an APLAC circuit simulator in which extensive distortion simulations are performed. The technique relies on the existing device model and thus the fitted model can be only as accurate as the particular simulation model. However, two different RF PAs are analysed that show a good agreement between measurements and simulations. The proposed technique is verified with several test cases including amplitude dependent amplitude and phase distortion, intermodulation distortion sweet spots, bandwidth dependent memory effects and impedance optimization. The main finding of the detailed analysis is that the distortion is a result of several cancelling mechanisms. In general, cubic nonlinearity of transconductance is dominating the in-band distortion but is cancelled by the 2nd-degree nonlinearity that is mixed to the fundamental band from envelope and 2nd harmonic bands that is usually the main cause of memory effects.

As wireless devices and radar systems become more ubiquitous, there is a growing need for wideband multi-standard RF-SOCs. To enable the advantages of multi-standard systems, reconfigurable RF front ends are needed. Because of the large number of RF signals in wideband systems, tunability and linearity become important parameters. Prior work has shown tunable LC filters to be advantageous in the microwave regime. A balanced-to-unbalanced (balun) transformation circuit is required to support the differential nature of a tunable LC filter. An active balun that also performs as a transconductor to drive the LC tank would relax the design requirement for the LNA and remove a passive balun that would have to precede the LNA. This thesis discusses the linearization of active baluns and presents a comparison between two 4 to 7 GHz tunable BPF designs with active baluns implemented in 0.18 μm SiGe BiCMOS technology. Fourth order filtering is achieved by subtracting two 2nd order LC-tanks. This approach allows 3-dB bandwidth to be tunable from 10% to 20%. In each design, a linearized input active balun is employed to drive the LC-tanks from a single-ended input while preserving noise figure and IIP3 performance. Two different linearization techniques are applied for the balun designs. Simulated NF ranges from 7.5 to 13 dB and IIP3 averages about 5 dBm with the peak value of 21 dBm. / Master of Science / As wireless devices and radar systems become more ubiquitous, there is a growing need for Radio Frequency (RF) integrated circuits that can support multiple frequency bands and standards. Because of the large number of RF signals, robust tunability and power handling of the electronics become important parameters. Power handling is important because the amplifier and the filter can generate distortions if the power going through them becomes too high. Prior work has shown integrated tunable inductor-capacitor (LC) resonance based filters to be advantageous in the microwave frequency regime compared to integrated switched capacitor based filters. A balancedto-unbalanced (balun) conversion of the RF signals is needed to support the differential nature of the LC resonators. This thesis discusses transistor-based balun designs that can be integrated into front-end LC filter chips. The goal is to reduce distortion in the filter under the present of large number of RF signals and to keep noise of the circuit in reasonable range. The designs are implemented in 0.18 μm SiGe BiCMOS integrated circuit technology and simulated in commercial computer aided design software; predicted performance is competitive with the state of the art. The fabricated chips will be characterized in future work.

This thesis covers the development of a series of new methods and the application of adaptive filter theory which are combined to produce a generalised adaptive filter system which may be used to perform such tasks as pattern recognition. Firstly, the relevant background adaptive filter theory is discussed in Chapter 1 and methods and results which are important to the rest of the thesis are derived or referenced. Chapter 2 of this thesis covers the development of a new adaptive algorithm which is designed to give faster convergence than the LMS algorithm but unlike the Recursive Least Squares family of algorithms it does not require storage of a matrix with n2 elements, where n is the number of filter taps. In Chapter 3 a new extension of the LMS adaptive notch filter is derived and applied which gives an adaptive notch filter the ability to lock and track signals of varying pitch without sacrificing notch depth. This application of the LMS filter is of interest as it demonstrates a time varying filter solution to a stationary problem. The LMS filter is next extended to the multidimensional case which allows the application of LMS filters to image processing. The multidimensional filter is then applied to the problem of image registration and this new application of the LMS filter is shown to have significant advantages over current image registration methods. A consideration of the multidimensional LMS filter as a template matcher and pattern recogniser is given. In Chapter 5 a brief review of statistical pattern recognition is given, and in Chapter 6 a review of relevant connectionist models. In Chapter 7 the generalised adaptive filter is derived. This is an adaptive filter with the ability to model non-linear input-output relationships. The Volterra functional analysis of non-linear systems is given and this is combined with adaptive filter methods to give a generalised non-linear adaptive digital filter. This filter is then considered as a linear adaptive filter operating in a non-linearly extended vector space. This new filter is shown to have desirable properties as a pattern recognition system. The performance and properties of the new filter is compared with current connectionist models and results demonstrated in Chapter 8. In Chapter 9 further mathematical analysis of the networks leads to suggested methods to greatly reduce network complexity for a given problem by choosing suitable pattern classification indices and allowing it to define its own internal structure. In Chapter 10 robustness of the network to imperfections in its implementation is considered. Chapter 11 finishes the thesis with some conclusions and suggestions for future work.