ABSTRACT. The power consumption of world wide data centers has been growing exponentially for the past 15 years. Robust power delivery network needs to be co-designed with voltage regulating devices, to ensure proper operation of high-performance microprocessors (xPUs, FPGA, ASIC etc.). In this talk we will talk about power delivery and thermal challenges of high performing compute devices that run cloud services in the data centers. We will explore the risks and opportunities in this area and share our considerations in practical designs.

ABSTRACT. Power-electronic control and modeling have gone through fundamental shift in its approach with time slowly but surely replacing reduced-order-manifold-based approaches proposed decades back. With the revolution in embedded processors and advancements in multi-objective optimization, stability theory, hybrid systems, and communication/information theory, radically new multi-scale spatio-temporal approaches are being developed and implemented that are showing unprecedented promise for plurality of power-electronic applications and changing the mindset regarding the control and modeling of such hybrid dynamical switching-power systems. At the actuation level, the advent of rapid switching wide-bandgap devices is enabling the accelerated penetration of such next-generation controls across plurality of voltage and power levels encompassing radically improved, new, and complex power-electronic systems. This keynote will begin with an outline on the role of control in traditional power-electronic systems and networks and how they shape the behavior of such hybrid dynamical systems. Subsequently, an overview of the traditional power-electronic control, analysis, and modeling approaches will be provided along with brief discussions on their strengths and limitations. That leads to the future of controls in power electronics and what should and could be done beyond traditional power-electronic control that addresses existing, evolving, and future applications needs encompassing wide temporal and spatial scales? This talk will provide some insights on how and what radically new ideas may need to be synthesized that reach far beyond historical and conventional power-electronic control needs with wide power-conversion applications.

ABSTRACT. Micro Fulfillment Centers (MFC) are being developed in order to get closer to customers, shorten the delivery time and reduce delivery expenses. Fabric, established in 2015, developed a robotic based MFC. The solution uses a Lift robot to retrieve a tote from the shelving unit and a Ground robot to move it within the area. A fleet of hundreds of robots, managed by Warehouse Control System (WCS), enables the fulfillment of multiple item orders in a few minutes. The robot uses a LiIon battery pack and is controlled over a WIFI communication system. A robotic arm, developed to replace the human operator, is already operational and reduces labour costs in the MFC.

ABSTRACT. Virtual synchronous machines (VSM) are a promising technology to integrate renewable energy sources into power grids. The VSM is a power converter that emulates, towards the grid, the behavior of a synchronous machine, providing the grid services which are necessary to operate a power system in a stable manner. When more VSMs are connected to the same grid, sub-synchronous oscillations between them (and between VSMs and other generators) may occur if damping coefficients and inertias are not properly tuned. For this purpose, virtual friction (VF) has been proposed to provide high damping without a strong coupling of frequency deviation and power output, unlike for frequency droop. VF applies a damping torque to the virtual rotor of the VSMs proportional to the deviation of the rotor frequency from the center of inertia (COI)-frequency of the grid. To the best of the authors knowledge, this technique has only been validated theoretically and in simulations for isolated microgrids. The goal of this paper is to demonstrate the effectiveness of VF both in isolated and grid connected operation by a theoretical and experimental assessment.

ABSTRACT. The application of droop control techniques without inner current control loops to doubly-fed induction generator (DFIG) based wind turbines does not allow to provide a stable response for every operating point. After modeling the system dynamics and analyzing the causes of instability, this paper proposes a control strategy that allows to stabilize the system response at all possible operating points in terms of rotational speed, active and reactive power. Simulation results performed in MATLAB/Simulink validate the proposed control strategy proving its effectiveness.

ABSTRACT. This paper presents a set of self-adaptive methods to detect and damp power system resonances. Unlike previously reported methods that rely on impedance analysis to determine the resonance frequency and amount of damping required, the proposed methods detect system resonances directly based on voltage and current measurements. Both time- and frequency-domain techniques are presented for this purpose. Once a resonance is detected, damping is automatically applied in a narrow frequency range centered at the resonance frequency to avoid untended consequences at other frequencies. The amount of damping to be applied is self-adjusted, and the damping function is automatically deactivated if the resonance has disappeared, e.g. due to changes in network configuration or system operation conditions.

ABSTRACT. We propose a circulating-current control scheme for parallel three-phase voltage-source inverter (VSI) that can be applied in general to any space-vector modulation (SVM) scheme and experimentally demonstrate its effectiveness for two commonly used SVM schemes (symmetrical SVM and bus-clamped SVM). The proposed control strategy relies on radio frequency (RF) based wireless communication to exchange circulating-current information among the parallel converter modules. In general, such a control scheme may lead to more redundant control implementation of distributed power systems and could also be used as a back-up for wire-based control schemes to provide fault tolerance.

ABSTRACT. This paper concern with the design of phase compensation of multi-resonant current controllers for AC converters based on system stability analysis for precise tracking and disturbance rejection. Phase compensation method is used to enhance the system stability in order to compensate the system total sampling and transport delay, since it can improve the phase response of resonant controller and consequently system phase margin. However, the phase compensation open loop system results in large anti-peaks around resonant frequencies, which may cause tracking error and system instability. This paper proposes a simpler design of phase compensated multi-resonant controller based on compensation of the phase lag of only the last resonant term of the multi-resonant controller in order to eliminate the anti-peaks affect and improve the system robustness and disturbance attenuation.

ABSTRACT. This paper presents a novel autotuning method that intends to improve controller performance in mobile device battery chargers. The mobile device chargers have a wide range of possible operating conditions and high converter parameter variations, making the design of fixed-coefficient controllers a challenging task. This work aims to tackle this issue using a comprehensive, yet simple, self-tuning algorithm with the goal to maximize bandwidth and to guarantee stability. The novel approach uses a single-tone injection, a single-node measurement, and generates orthogonal components for extracting in-phase and quadrature projections of the response, while requiring modest hardware resources. The algorithm enables significant savings of computational resources thus making it practical. Effectiveness of the proposed solution has been experimentally verified using an FPGA-based controller and a monolithic battery charger prototype.

ABSTRACT. Piezoelectric resonator DC-DC converters have been demonstrated with high efficiency and power-density, offering a promising prospect for power converter miniaturization. However, the operation of these converters relies on precise switch timings to achieve zero-voltage-switching and soft charging of the resonator’s input capacitance, creating challenges for control at high frequencies. This paper presents a closed-loop control method that regulates voltage through forward-zero cycle modulation. The technique uses a fixed switching frequency, requires only one sensing operation per resonant period, and enables as fast transient response.

ABSTRACT. This paper introduces a highly-miniaturized soft-switched power-supply suitable for advanced IoT applications. It incorporates a HB-LLC resonant converter operating in the MHz range and a new hybrid PFM-PWM digital controller that facilitates tight output voltage regulation carried out through a simple voltage loop controller hardware. The switching frequency and the dutycycle are adjusted simultaneously and with full continuity and monotonicity. Governed by two PI compensators to guarantee smooth and fast recovery from load-transient or input voltage aberrations the controller accounts for the non-monotonicity attribute of asymmetrically driven LLC resonant converters. The operation of the controller is experimentally verified on a 3-7V input to 1.8V output HB-LLC converter, demonstrating excellent voltage regulation capabilities and significant reduction of the passive component’s stress during start-up.

ABSTRACT. A hardware-efficient controller obtaining minimumdeviation response for both direct and indirect energy transferdc-dc converters and not requiring complex calculations isintroduced. Its operation is verified with a VIN ∈ [3 V to 20V], 100 W, 250 kHz prototype, producing up to 20 V, for bothbuck and boost converter cases. Practically the smallest possiblevoltage deviation resulting in the theoretically smallest outputcapacitance value is demonstrated.