The proposed thesis aims to develop innovative isolated DC-DC power converter topologies that enhance efficiency and scalability for sustainable energy technology applications. The research begins by optimizing the parallel differential power processing (PDPP) architecture in photovoltaic (PV) systems by replacing two dual active bridge (DAB) converters with a single optimized DAB. This first phase focuses on simplifying the PDPP architecture, reducing hardware redundancy, and improving system cost-effectiveness while maintaining high performance in mitigating mismatch losses among PV strings. The insights gained from this optimization will inform the second phase by adapting the DAB topology. In this phase, the conventional full-bridge design of the DAB will be replaced with a Packed U-Cell (PUC5) converter to meet the bidirectional power flow and high-power density requirements. This unified project bridges renewable energy domains by leveraging the fundamental principles of efficient power processing while addressing the unique challenges of sustainable energy applications. Through detailed real-time simulations or experimental validations and performance evaluations, the thesis will contribute to developing scalable, cost-effective, and high-performance solutions that advance sustainable energy technologies. My thesis aligns closely with my study programme in Sustainable Energy Technology, particularly in Electric Mobility Systems. It integrates key aspects of power engineering, snd energy storage, by focusing on advanced power converter architectures. The academic aspect of my thesis lies in advancing the state of the art in power converter technology through theoretical analysis, real-time simulation, or experimental validation. It contributes to existing knowledge by proposing novel designs—replacing dual DAB converters with a single optimised DAB in PV systems and introducing a new DAB converter using the PUC5 topology.

A modular approach to high-precision current amplifiers benefits lithography machines by reducing size and increasing efficiency. This research proposes two promising controller structures for a modular topology: a parallel controller structure and a cascaded controller structure. The study evaluates and compares their performance, introduces Delta-Sigma (DS) Analog-to-Digital Converters (ADCs) as a cost-effective alternative, and investigates their limitations in the proposed controller structures. A Feed Forward (FF) strategy is introduced to enhance the frequency response performance of both controllers.
Two methods to increase efficiency, particularly in amplifiers that are using interleaved Pulse-Width Modulation (PWM), are introduced: Phase-Shedding (PS) and Zero-Voltage Switching (ZVS). By estimating power losses it is shown that PS increases power efficiency for lower current setpoints, while ZVS increases efficiency for higher current setpoints. The study investigates how these methods impact controller stability and output current performance.
A hardware implementation validates the operation of the two controller structures. Stability issues with the cascaded controller structure become apparent and are resolved by a method that sacrifices bandwidth. Various measurements are conducted, and their results are compared to analytical expectations. The best frequency response is obtained using the parallel controller structure with a complex FF strategy, which gives a magnitude response of -0.3 dB and a phase shift of -3.6° at a 400 Hz setpoint.
When comparing the two controller structures, each has its own advantages and disadvantages. Choosing between the structures should be based on specific application requirements. If the application requires large bandwidth with high stability, the parallel controller should be selected. If the application requires good modularity, low quantization noise, and disturbance rejection performance, the cascaded controller should be selected.

The global shift towards sustainable energy sources, together with rapid technological advancements, has driven a significant increase in the demand for electric vehicles (EVs) over recent years. This surge in EV adoption highlights the critical need for robust supporting infrastructure, particularly in the deployment of efficient and accessible EV chargers. Gallium Nitride (GaN) is a wide bandgap semiconductor material that offers better performance characteristics compared to traditional silicon-based components in medium-power and high-frequency application. Its higher electron mobility, saturation electron velocity and breakdown electric field make GaN an ideal candidate for high-frequency power conversion applications in EV chargers. To explore the potential of GaN technology in EV charging applications, an 11kW GaN based isolated bidirectional multi-port EV charger with a two-stage architecture has been designed to interconnect bipolar DC grid and electric vehicles. The two stages of the charger consist of a three-level triple active bridge converter and a three-level interleaved buck converter. The TAB converter is an extension of conventional dual active bridge (DAB) converter, offering capability to direct power transmission among three ports. However, the addition of a third port introduces complexity in managing power flow. The TAB converter can operate under various modulation schemes, including three-level and five-level modulation. An analysis of the switching modes associated with each modulation scheme has been
erformed. In order or derive the soft switching range of the converter, the current expression
of series inductors of each port are derived based on generalized-harmonic approximation method and equivalent circuit models. A detailed analysis of the commutation process during critical switching transitions has been conducted, leading to the derivation of the soft
witching criteria for both three-level and five-level operations of the TAB converter. To validate the design and verify the soft switching range, simulations have been carried out. Moreover, the prototype of the 11 kW multi-port EV charger is designed based on calculation and simulation results. The prototype employs a top-side cooled GaN high-electron-mobility transistor (HEMT) to enhance thermal dissipation. Testing of the prototype indicates that the converter can transmit 1.2 kW from the primary to the secondary side at the rated voltage
of 700 V.

This report proposes a design of a 1-megawatt data center that utilizes hydrogen as the primary fuel for power and cooling. Currently, data centers rely on utility grids for their operation. The growing global efforts to achieve sustainability introduce intermittent renewable energy sources, such as solar and wind, into the power grid, which goes against the reliability principles of the data center industry. As computer technologies further permeate our everyday lives, their energy consumption and environmental impact become greater.

The feasibility study aims to investigate the potential of using hydrogen as a primary energy carrier in data center applications in terms of reliability, efficiency, electrical and hydrogen topology, infrastructure layout, and control strategies. A comprehensive review of commercially available and emerging hydrogen technologies related to storage, transport, electricity generation, and cooling was conducted. Based on this literature review, two designs for a hydrogen data center were developed. A modular approach to the design was proposed, where each module is self-contained and responsible for providing the servers with 1 megawatt of electrical power and cooling.

The first low-temperature design utilizes a hydrogen pipeline with pressure swing adsorption purification supplying hydrogen gas to a cogeneration system. This system combines a low-temperature Proton-Exchange Membrane fuel cell and a direct-fired hydrogen single-stage absorption refrigeration cycle into a single unit to deliver stable power and cooling to the servers. The findings indicate that the overall energy efficiency of this design configuration is high, and the technologies are mature enough to be used today. However, the study identified various technological limitations of purification systems that must be addressed with further research and development.
The second high-temperature design eliminates the need for a purification system, relying instead on high-temperature Proton-Exchange Membrane fuel cells coupled to a double-effect absorption refrigeration system. Unfortunately, this design is not feasible in 2024 because the high-temperature fuel cells are in the early stages of development.

A cost analysis was conducted for both designs in the current (2024) and projected (2030) scenarios. The results indicate that at the current level of hydrogen technology development and manufacturing, the operational costs of the low-temperature configuration were comparable to the traditional data centers. However, if hydrogen technology continues to develop according to the projected trends, the proposed designs can achieve up to 50% cost savings compared to the grid-powered data centers. This research showcases the potential for hydrogen as an energy carrier in data center applications, encouraging further research and exploration of hydrogen technologies.

Multilevel inverters are widely used in industries for higher power applications due to low harmonic pollution and high efficiency. On the other hand, the phase-shift switching technique is well-known for its harmonic mitigation performance. However, phase-shift switching has not been regularly implemented on multilevel inverters due to complexity at a higher number of levels and difficulty in voltage balancing of the auxiliary capacitors in the configuration. In this project, the five-level Packed U-Cell inverter (PUC5) will be investigated to design and implement a phase-shift switching technique with an integrated voltage balancing algorithm. The developed technique will be implemented on the real-time simulator and hardware setup.

A literature review will be done on phase-shift switching algorithms and multilevel inverters. Afterward, a phase-shift switching algorithm with an integrated voltage balancing technique will be developed for the PUC5 inverter and simulated in Simulink. Hardware-in-the-loop (HIL) will be done using Opal-RT devices.

The focus of this report is the design and implementation of a universal gate driver for half-bridge circuits. This design includes a programmable dead-time controller and adjustable properties to ensure efficient and reliable switching on a variety of high-voltage applications. Furthermore, this work includes a comparative understanding of all design choices, such as the options considered for the dead-time controller, the gate driver component and the circuit of the output phase. The designed system is expected to receive as input a PWM signal and generate two output signals to drive a halfbridge configuration. The output pulses must be generated with a programmable dead-time and at the input switching frequency. The end product of this design will be used in the TU Delft Power Electronics Lab. The specific requirements are determined in order to meet the lab’s needs and this report further describes the process from acquiring the technical requirements, the circuit design, and finally, the prototype evaluation.