The power supply chain of data centers from the medium voltage (MV) utility grid down to the chip level voltage consists of many series connected power conversion stages and accordingly shows a relatively low efficiency. Solid-State Transformers (SSTs) could improve the efficiency by substantially reducing the number of power conversion stages and/or directly interfacing the MV AC grid to a 400 V DC bus, from where server racks with a power consumption of several tens of kilowatts could be supplied by individual SSTs. The recent development of SiC MOSFETs with a blocking voltage of 10 kV enables the realization of a simple and hence highly reliable two-stage SST topology, consisting of an AC/DC PFC rectifier and a subsequent isolated DC/DC converter. In this context, an isolated 25 kW, 48 kHz, 7 kV to 400 V series resonant DC/DC converter based on 10 kV SiC MOSFETs is realized and tested in this paper. To achieve zero voltage switching (ZVS) of all MOSFETs, a special modulation scheme to actively control the amount of the switched magnetizing current on the MV and LV side is implemented. Furthermore, the design of all main components and especially the electrical insulation of the employed medium frequency (MF) transformer is discussed in detail. Calorimetric efficiency measurements show that a fullload efficiency of 99.0 % is achieved, while the power density reaches 3.8 kW/L (63 W/in 3 ).
This paper analyzes the potential of Artificial Neural Networks (ANNs) for the modeling and optimization of magnetic components and, specifically, inductors. After reviewing the basic properties of ANNs, several potential modeling and design workflows are presented. A hybrid method, which combines the accuracy of 3D Finite Element Method (FEM) and the low computational cost of ANNs, is selected and implemented. All relevant effects are considered (3D magnetic and thermal field patterns, detailed core loss data, winding proximity losses, coupled loss-thermal model, etc.) and the implemented model is extremely versatile (30 input and 40 output variables). The proposed ANN-based model can compute 50 000 designs per second with less than 3 % deviation with respect to 3D FEM simulations. Finally, the inductor of a 2 kW DC-DC buck converter is optimized with the ANN-based workflow. From the Pareto fronts, a design is selected, measured, and successfully compared with the results obtained with the ANNs. The implementation (source code and data) of the proposed workflow is available under an open-source license.
Due to their extremely high energy demand, data centers are directly supplied from a medium voltage (MV) grid. However, a significant part of this energy is dissipated in the power supply chain, since the MV is reduced step-by-step through multiple power conversion stages down to the chip voltage level. In order to increase the efficiency of the power supply chain, the number of conversion stages must be substantially reduced. In this context, Solid-State Transformers (SSTs) are considered as a possible solution as they could directly interface the MV AC grid to a 400 V DC bus, whereby server racks with a power consumption of several tens of kilowatts could be directly supplied from an individual SST. With a focus on the lowest system complexity, the SST ideally should be built as simple two-stage system consisting of an MV AC/DC PFC rectifier stage followed by an isolated DC/DC converter. Accordingly, this paper focuses on the design and realization of a 25 kW, 3.8 kV single-phase AC to 7 kV DC PFC rectifier unit based on 10 kV SiC MOSFETs. By simply adding an LC-circuit between the switch nodes of the well-known full-bridge-based PWM AC/DC rectifier, the integrated Triangular Current Mode (iTCM) concept is implemented, which only internally superimposes a large triangular current ripple on the AC mains current and therefore enables zero voltage switching (ZVS) over the entire AC mains period. Special attention is paid to the realization of the MV inductors and their electrical insulation, the AC-input LCL-filter to limit EMI emissions, and the challenges arising due to cable resonances when connecting the SST to the MV grid via a MV cable. Despite the large insulation distances required for MV, the realized 25 kW MV PFC rectifier achieves an unprecedented power density of 3.28 kW/L (54 W/in 3 ) and a full-load efficiency of 99.1 %, determined using a calorimetric measurement setup, which is discussed in detail in the Appendix.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.