Research Projects

Highly linear, low-noise, and tunable bandpass filters are the holy grail of RF engineering.However, traditional on-chip active bandpass filters suffer from fundamental noise and nonlinearity issues due to active components like opamps and transconductors, making power-efficient on-chip implementations impractical. It turns out that the noise and intermodulation distortion of an active bandpass filter are both directly proportional to the quality factor Q. Making narrow band filters, therefore, where the Q is very large, is so difficult that researchers have abandoned them. As a result, RF-quality bandpass filters are typically realized using off-chip passives; offering excellent linearity but at the cost of bulkiness and lack of tunability. As is well known, superheterodyne receivers were ubiquitous in discrete radios due to their immunity to dc offsets and flicker noise, and improved channel selection due to multiple stages of filtering. Using the superheterodyne architecture in integrated radios, however, is challenging due to the need for high-Q bandpass filters. Since active bandpass structures cannot be used on account of their noise and linearity limitations, the only choice is to employ off-chip passive filters. This not only increases pin count, but is also power hungry due to the low impedance levels of these filters. Due to all these reasons, integrated RF receivers have abandoned this otherwise robust radio architecture, and gravitated to the direct conversion architectures. This is in spite of direct conversion receiver’s well-known challenges, including 1/f noise, LO reradiation, and IQ imbalance. Regardless of architecture, modern receivers ultimately require digitization of their output so that sophisticated DSP algorithms can demodulate the signal appropriately. In this proposal, we aim to combine active bandpass filtering with analog-to-digital conversion in a way as to achieve low noise and high linearity operation (like in a passive bandpass filter), but with the compactness and tunability advantages of an active filter. Why is this feasible? We have experience with combining low pass filtering and analog-to-digital conversion. We have shown both theoretically and experimentally that this approach enables the realization of filters with lower noise and distortion for a given power. We propose an allied approach: combining bandpass filter realization with the ADC itself. This breakthrough paves the way for a new generation of high-performance, power-efficient RF front-end and IF-section designs. The aim of this project is to realize bandpass continuous-time ADCs, where narrow-band filtering at RF center frequencies and analog-to-digital conversion are simultaneously achieved. This area is new, and we hope to make fundamental and long-lasting contributions to this challenging and industrially-relevant subject. We also see commercialization opportunities for this research.