Nos tutelles

CNRS UPS Enseeiht


Accueil > Actualités

Contribution to the Distributed Fault Tolerant Control of Multiphase Converters Dedicated to the Supply of Microprocessors for Automotive Applications

par Laurence Laffont - publié le , mis à jour le

Miguel MANNES-HILLESHEIM’s thesis defense, intitled "Contribution au contrôle Distribué tolérant à la panne des convertisseurs multiphases dédiés à l’alimentation de microprocesseurs pour application automobile"(Contribution to the Distributed Fault Tolerant Control of Multiphase Converters Dedicated to the Supply of Microprocessors for Automotive Applications), will be taking place on Monday, December 13th, 2021 at 10.00 am, at the ENSEEITH’s salle de thèse, building C.

Thesis’ work has been realised within the CS group of LAPLACE, in partnership with NXP Semiconductors - Toulouse.

Jury :

Mr. Eric LABOURÉ, GeePs - CentraleSupelec, rapporteur,
Mr. Yves LEMBEYE, G2Elab - Grenoble Alpes University, rapporteur,
Ms. XUEFANG Lin-Shi, Ampere - INSA Lyon, external reviewer,
Mr. Cyrille GAUTIER, , SATI – Centrale Supelec, external reviewer,
Mr. Nicolas ROUGER, LAPLACE, internal reviewer,
Mr. Eric ROLLAND, NXP Semiconductors, guest,
Mr. Marc COUSINEAU, thesis’ supervisor, internal reviewer,
Mr. Guillaume GATEAU, thesis’ co-supervisor,internal reviewer.

Abstract :

Advanced Driver-Assistance Systems can take control of the vehicle in case of danger. Redundancy and reconfiguration ability in case of fault occurrence are fundamental to ensure mission continuity. The supply of computer cores is made usually with multiphase converters that offer naturally multiple (redundant) current paths to the processor thanks to its parallel topology. However, converter controller is a weakness in terms of availability usually implemented in a centralized way. NXP Semiconductors has initiated a partnership with the LAPLACE laboratory to finance this thesis for the study of a fault tolerant distributed control mode. This thesis focuses on the use of a distributed control method, presenting no single point of failure, to develop a modular and redundant control signal interleaving management unit, as well as to control its operation, in terms of response time and stability.
Chapter I presents the context and state of the art. The context presents the NXP safety path ; a survey of requirements and techniques is developed to understand up-to-date processors power supply ; trends of workload and power consumption increases are highlighted ; and the adoption of multiphase converter topology is verified. The state of the art covers mainly the modular distributed techniques applied on voltage regulation and current balance, and then on interleaving techniques, needed to multiphase converter operation.
Chapter II and III present the theoretical studies and evaluations of two preexisting distributed interleaving techniques. The same modeling technique is applied in each chapter to develop a multiple inputs multiple outputs model compliant with the specificities of each approach and then reveal the stability concern. Subsequently, a technique of choice of corrector for the optimization of the transient response is proposed. The singularities of each approach are discussed and simulations are provided for verification.
Chapter IV presents the conception of an improved original interleaving approach for control signal interleaving, as well as its analysis. It is a mixed-signal approach, consisting of a few digital logic devices and an analogue core. The theoretical study and evaluation of the new interleaving technique is presented, as well as simulation results. A proof of concept was developed using standard SMD components on a printed circuit board. The dynamic response of this interleaving approach is verified for modal disturbances, cases of start-up and reconfiguration, where a module is inserted/removed during operation. The behavior during communication fault event are also shown, comprising the cases of a line shorted to ground and line-to-line short circuit.

Keywords : Fail-operational, fault-tolerant, distributed control, reconfiguration, mission continuity, interleaving, modular, scalable, circular-chain.