The current thesis is about investigation of new methods and techniques to be able to utilize the switched mode amplifiers, for linear and efficient applications. Switched mode amplifiers benefit from low overlap between the current and voltage wave forms in their output terminals, but they seriously suffer from nonlinearity. This makes it impossible to use them to amplify non-constant envelope message signals, where very high linearity is expected. In order to do that, dithering techniques are studied and a full linearity analysis approach is developed, by which the linearity performance of the dithered amplifier can be analyzed, based on the dithering level and frequency. The approach was based on orthogonalization of the equivalent nonlinearity and is capable of prediction of both co-channel and adjacent channel nonlinearity metrics, for a Gaussian complex or real input random signal. Behavioral switched mode amplifier models are studied and new models are developed, which can be utilized to predict the nonlinear performance of the dithered power amplifier, including the nonlinear capacitors effects. For HFD application, self-oscillating and asynchronous sigma delta techniques are currently used, as pulse with modulators (PWM), to encode a generic RF message signal, on the duty cycle of an output pulse train. The proposed models and analysis techniques were applied to this architecture in the first phase, and the method was validated with measurement on a prototype sample, realized in 65 nm TSMC CMOS technology. Afterwards, based on the same dithering phenomenon, a new linearization technique was proposed, which linearizes the switched mode class D amplifier, and at the same time can reduce the reactive power loss of the amplifier. This method is based on the dithering of the switched mode amplifier with frequencies lower than the band-pass message signal and is called low frequency dithering (LFD). To test this new technique, two test circuits were realized and the idea was applied to them. Both of the circuits were of the hard nonlinear type (class D) and are integrated CMOS and discrete LDMOS technologies respectively. The idea was successfully tested on both test circuits and all of the linearity metric predictions for a digitally modulated RF signal and a random signal were compared to the measurements. Moreover a search method to find the optimum dither frequency was proposed and validated. Finally, inspired by averaging interpretation of the dithering phenomenon, three new topologies were proposed, which are namely DLM, RF-ADC and area modulation power combining, which are all nonlinear systems linearized with dithering techniques. A new averaging method was developed and used for analysis of a Gilbert cell mixer topology, which resulted in a closed form relationship for the conversion gain, for long channel devices.
|Kwalificatie||Doctor in de Filosofie|
|Datum van toekenning||7 mrt 2012|
|Plaats van publicatie||Eindhoven|
|Status||Gepubliceerd - 2012|