Dual-Band Transmitters Using Digitally Predistorted Frequency Multipliers for Reconfigurable Radios
MetadataShow full item record
The objective of the proposed research is to develop simplified reconfigurable transmission systems with frequency multipliers for the transmission of complex modulated signals. Because they rely on nonlinear properties, frequency multiplier-based transmission systems require proper linearization techniques and accurate modeling of the signal transfer function. To accomplish these two goals, the author has developed techniques to model and linearize frequency multipliers and to digitize feedback signals for nonlinear characterization. First, adaptive predistortion techniques and zonal transfer theories have been developed for modeling and linearization. The predistortion system has been verified by applying an IS-95B signal to various frequency multipliers built by the author. Second, because the output signals at higher harmonic zones occupy wider frequency bandwidths than the signal in the fundamental zone does and thus make it harder to use traditional sampling techniques, a simplified but effective method called the sub-Nyquist sampling rate was developed and verified. Third, two methods for reconfigurable transmitters using frequency multipliers in conjunction with digital predistortion linearizers were developed. Both methods make it possible to transmit complex signals via frequency multipliers by using dual-band transmission systems that incorporate frequency multipliers that are based on linearization techniques. One of these methods uses a circuit topology that can be switched between a fundamental-mode in-phase combined amplifier and a push-push frequency doubler using input phasing. The second suggested method uses a fundamental-frequency power amplifier followed by a varactor multiplier that can be bypassed with an RF switch. This work will contribute to the development of low-cost and size-effective reconfigurable transmission systems because it requires fewer transmitting components and needs less sampling of the feedback networks.