Architectures and Circuit Techniques for High-Performance Field-Programmable CMOS Software Defined Radios

Next-generation wireless communication systems put more stringent performance requirements on the wireless RF receiver circuits. Sensitivity, linearity, bandwidth and power consumption are some of the most important specifications that often face tightly coupled tradeoffs between them. To increase the data throughput, a large number of fragmented spectrums are being introduced to the wireless communication standards. Carrier aggregation technology needs concurrent communication across several non-contiguous frequency bands, which results in a rapidly growing number of band combinations. Supporting all the frequency bands and their aggregation combinations increases the complexity of the RF receivers. Highly flexible software defined radio (SDR) is a promising technology to address these applications scenarios with lower complexity by relaxing the specifications of the RF filters or eliminating them. However, there are still many technology challenges with both the receiver architecture and the circuit implementations. The performance requirements of the receivers can also vary across different application scenario and RF environments. Field-programmable dynamic performance tradeoff can potentially reduce the power consumption of the receiver. In this dissertation, we address the performance enhancement challenges in the wideband SDRs by innovations at both the circuit building block level and the receiver architecture level. A series of research projects are conducted to push the state-of-the-art performance envelope and add features such as field-programmable performance tradeoff and concurrent reception. The projects originate from the concept of thermal noise canceling techniques and further enhance the RF performance and add features for more capable SDR receivers. Four generations of prototype LNA or receiver chips are designed, and each of them pushes at least one aspect of the RF performance such as bandwidth, linearity, and NF. A noise-canceling distributed LNA breaks the tradeoff between NF and RF bandwidth by introducing microwave circuit techniques from the distributed amplifiers. The LNA architecture uniquely provides ultra high bandwidth and low NF at low frequencies. A family of field-programmable LNA realized field-programmable performance tradeoff with current-reuse programmable transconductance cells. Interferer-reflecting loops can be applied around the LNAs to improve their input linearity by rejecting the out-of-band interferers with a wideband low in- put impedance. A low noise transconductance amplifier (LNTA) that operates in class-AB-C is invented to can handle rail-to-rail out-of-band blocker without saturation. Class-AB and class-C transconductors form a composite amplifier to increase the linear range of the input voltage. A new antenna interface named frequency-translational quadrature-hybrid (FTQH) breaks the input impedance matching requirement of the LNAs by introducing quadrature hybrid couplers to the CMOS RFIC design. The FTQH receiver achieves wideband sub-1dB NF and supports scalable massive frequency-agile concurrent reception.