Analysis of a radio frequency class D amplifier architecture with bandpass sigma-delta modulation

This thesis analyzes an amplifier architecture that combines a RF class D amplifier with a bandpass sigma-delta modulator, broadening the utility of class D amplification to include signals with envelope variation. An integrated design methodology is presented that incorporates the coding efficiency and average pulse transition frequency of the encoded pulse train into classical RF class D amplifier design equations. The equations are used to predict the power efficiency of a complementary voltage switched class D amplifier design with CMOS, pHEMT, and MESFET switches. Simulated results are compared with the analysis and verify the design methodology. The power efficiency analysis shows a direct link between modulator coding efficiency and the output power of the amplifier; therefore, a modulator with high coding efficiency is desirable. It is shown that coding efficiency depends significantly on the order of the modulator loop filter as well as the carrier oversample ratio employed in the design. The variation with carrier oversample ratio is not monotonic for second and fourth order modulators, and some oversample ratios are more optimal than others. Bandpass sigma-delta modulation synthesizes a pulse train with synchronous zero-crossings, and the coding efficiency limitations of encoding a binary amplitude pulse train with constrained zero-crossings is analyzed. The analysis and characterization of other encoder designs shows that bandpass sigma-delta modulation is remarkably efficient. The analysis is extended to pulse train upconversion employing Manchester encoding. Upconversion reduces the difficulty of implementing highly selective noise shaping resonators at RF frequencies, and the impact of upconversion in terms of coding efficiency and average transition frequency is shown.