Light micro-energy harvesting in standard CMOS technologies

Resumo

Micro-energy harvesting is the process of scavenging small amounts of energy from the environment. This energy can come from different sources, like light, thermal gradients, vibrations, etc. In conjunction with low-power budget strategies, this energy is used to power systems and extend their battery life. Implantable devices, wearable computing or smart dust wireless sensors are examples of electronic systems that need energy harvesting to operate. This thesis addresses the issue of light micro-energy harvesting with both the energy transducer and the Power Management Unit (PMU) that boosts its input voltage and runs the Maximum Power Point Tracking (MPPT) on the same silicon substrate in standard CMOS technologies. The voltage generated by on-chip solar panels (photodiodes) is too low to power the electronics within the chip, making necessary the use of a voltage step-up converter. DC-DC converters are the solution to solve this challenge. Capacitive DC-DC converters, also called charge pumps, are particularly attractive because they can be fully integrated with a relative small form factor. Different models to study the performance of the charge pumps have been reported in the literature. Nevertheless, these models do not account for all the phenomena affecting the charge pumps. Therefore, new accurate models were developed in this thesis to help in the design process of charge pumps. On the one hand, a reliable, simple and reproducible model for the transient analysis of two-phase charge pumps including the charge reusing approach was developed. The model provides both charge and voltage time responses at every flying capacitor and at the output for both current and capacitive loads in the slow switching limit (SSL) region. The model accounts for both top and bottom flying parasitic capacitances and it is demonstrated that parasitic capacitances from the switches can be included as part of the top and bottom parasitic capacitances improving the accuracy of the model. The model was validated through circuit-level simulations and experimental results and compared with the main models in the literature, featuring better accuracy and saving up to 10000x in computation time. On the other hand, an accurate and simple model for the transient analysis of the joint effect of the photodiode and the charge pump driven by two clock signals was also developed in this thesis. The model also accounts for both top and bottom flying parasitic capacitances of the charge pumps. A classical model for the photodiode whose photogenerated current was extracted from device- level simulations was assumed. The joint model was verified by circuit-level simulations achieving high accuracy and computation time savings of up to 1700x. Finally, a new architecture of an energy harvesting system to improve the current solutions was proposed. The architecture of this micro-energy harvesting system consists of a PMU powered by a 1 mm2 on-chip solar cell to rise up the harvested voltage to an output voltage higher than 1.1 V, suitable for powering low-power circuits. The chip was fabricated in 0.18 μm CMOS technology achieving a form factor of 1.575 mm2. The PMU includes a MPPT block which performs an open-loop and continuous MPPT without disconnecting the photodiode from the PMU, handling an input power range from nW to μW and overcoming state-of-the-art solutions. The PMU is able to start up from a harvested power of 2.38 nW without any external kick off or control signal achieving a peak efficiency of 57% during normal operation. This approach can be of interest for applications such as implantable devices, which demand long-time maintenance cycles along with the capability of handling low and wide input power ranges.