Digital control is gaining popularity in switching converters because of its flexibility and re-configurability to implement various control functions without hardware modification. It is also less susceptible to aging, noise, process and parameter variations. Nonetheless, a major challenge with digital control is bandwidth limitation which is primarily caused by the phase delay in the ADC and the digital feedback loop. This research reviews the existing digital control strategies. A digital controller with adaptive prediction is proposed which increases the unity-gain bandwidth even with a modest 2x oversampling ratio. By introducing an additional low-frequency pole-zero pair, the DC gain can be further increased which significantly improves the line transient dynamics by reduc...[ Read more ]

Digital control is gaining popularity in switching converters because of its flexibility and re-configurability to implement various control functions without hardware modification. It is also less susceptible to aging, noise, process and parameter variations. Nonetheless, a major challenge with digital control is bandwidth limitation which is primarily caused by the phase delay in the ADC and the digital feedback loop. This research reviews the existing digital control strategies. A digital controller with adaptive prediction is proposed which increases the unity-gain bandwidth even with a modest 2x oversampling ratio. By introducing an additional low-frequency pole-zero pair, the DC gain can be further increased which significantly improves the line transient dynamics by reducing the maximum overshoot and undershoot. An FPGA-based hardware for a digitally-controlled single-output buck converter has been experimentally verified, demonstrating the effectiveness of the proposed digital control scheme.

Digital control can also be used to regulate single-inductor multiple-output (SIMO) switching converters. As the LED technology continues to evolve, there is a stronger need for higher power efficiency and lower cost which motivates us to investigate a new SIMO-based driver topology. Instead of connecting all the strings to a common output bus as in conventional driver topologies, the output bus voltage in each LED string can be individually optimized in the proposed driver. The voltage headroom per string is minimized which reduces the overall power loss. Since each string is driven independently, same-colored LEDs from neighboring bins with larger forward-voltage variations can be tolerated, hence lowering the LED cost. Another important design consideration is the scalability of a SIMO-based driver. The proposed scalability models define the theoretical maximum achievable number of strings in a general SIMO architecture. They enable converter designers to quickly determine the maximum possible number of parallel strings in a SIMO-based driver required for a particular application. A single-inductor dual-string driver has been implemented in an FPGA-based PCB prototype. The experimental results are shown to be consistent with the corresponding ones from simulation. Cross-regulation has always been a major design challenge for SIMO converters. A modified quasi-V² pseudohysteretic control for SIDO driver is proposed in order to reduce the cross-regulation between the two independently-driven LED channels. The key features include a fast discharge path for the quasi-V² sense node via an additional pull-down switch and a feedback path from the unchanged output to the first-order RC filter. Similar to the inductor current in DCM, the quasi-V² feed-forward voltage is returned to zero at the end of each switching phase which significantly suppresses the cross-regulation. The simulation results show that the cross-regulation is substantially reduced with the modified quasi-V² control scheme, compared with the original one. The experimental results demonstrate that no noticeable cross-regulation is observed for a load current step of 120 mA.