Several reliability issues and solutions for LED lighting system

Abstract

This thesis presents a study on several reliability issues for LED lighting systems. Firstly, a full survey on exsiting LED ballast has been conducted, and critical design challenges are classified on power level basis (low/ middle/ high power). Specifically, reliability issues have been highlighted, and three major factors have been stipulated: issue of electrolytic capacitor; issue of current imbalance in parallel LED strings; issue of LED junction temperature. The information revealed in the whole survey provide important design criteria for existing LED system designs and guidance for further research directions by pointing out the critical design problems.

Two possible solutions for Electrolytic-Capacitor-Less LED Ballasts are proposed regarding the first reliability issue. A series of novel passive LED ballasts are proposed. They are found to be suitable for outdoor applications, such as street lighting applications, where the ability to withstand extreme weather conditions are of major concern. When compared with those in switched mode power supplies, these passive ballasts have good power factor performance and comparatively high efficiency. In addition, an active solution has been developed for indoor applications. Its circuit topology is derived from existing differential inverter topologies and inherits same merits such as simple structure, reduced size, and low cost.

Self-configurable current-mirror techniques have been derived and developed afterwards to cope with the current imbalance issue for system with parallel LED strings. In contrast with traditional current sharing methods (either linear type or switched type), the proposed techniques offer a simple solution without the need of independent current references, complicated controllers and auxiliary power supplies. These features are favored by outdoor applications and such re-configurable mirror circuits are originally designed for passive LED ballast as post-current regulators. The techniques are further extended with the ability to tolerate possible circuit failure, such as short circuit and open circuit fault.

Then, a new non-contact method for the measurement of both junction-to-case thermal resistance and junction temperature in a LED device has been proposed, with respect to the third reliability issue. Traditionally the direct measurement of junction temperature in LED is not easy without the help of sophisticated methods such as laser or expensive equipment like TeraLED Transient Thermal Tester system. In contrast, the proposed method requires only the external LED power, luminous flux, and heatsink temperature information. The method can be further adopted as a simple tool by engineers to check the internal temperature states in a practical LED system for regulation and evaluation purpose. Finally, a thermal design methodology has been developedfor an LED street lamp system powered by a weakly regulated renewable small power grid. It has been successfully incorporated in the proposed passive LED ballast, such that the LED system can provide the least output luminous fluctuation against line voltage variation. It is envisaged that, with the proposed design methodology, such lighting system will not only provide high reliability, with potential lifetime exceeding 10 years, but with a proven feature of reduced light fluctuation, furthermore, it is found that the passive LED system can act as a smart load and has the potential of reducing the energy storage requirement for smart grids. These merits are attractive to public lighting systems powered by future renewable power grids.