Research On Junction Temperature Measurement Of Distributed High-power White Light-emitting Diode Arrays Based On The Spectrum Methods

As a new type of solid state lighting source, high-power white light-emittingdiode (LED) has a lot of advantages in energy-saving, environmental-protecting,small-cubage, excellent anti-shake ability and rapidly start. However, the opticalefficiency and service lifetime of high-power white LEDs decrease rapidly withincreasing junction temperature. As that interferes the application and developmentof LED in illumination. Especially, distinguishing features of the distributed LEDarray that constructed with tens or hundreds of LED chips are high powerconsumption, heat concentrated and high junction temperature. The junctiontemperature (JT) of each chip in the LED product has been the key factor thatinfluences the reliability and service lifetime of LED product. Thus, it is extremelymeaningful to test the junction temperature of each chip accurately in the LEDproduct.Junction temperature measurement of distributed high-power white LED arraysbased on the spectrum methods is mainly discussed in the thesis. The research isconsisted of two parts: one part is junction temperature measurement based onspectrum methods, and the other is the development of junction temperaturemeasurement system for high-power white LED array. Such details are as follows:(1) To analyze the variation in temperature-sensitive optical parameters (TSOPs)property with changing JT, the electroluminescent (EL) spectrum is tested atdifferent JTs. It is found that the blue peak wavelength of the spectrum decreaseswith increasing JT. Thus, it is possible to test the JT using peak wavelength shift forhigh-power white LEDs.(2) The feasibility of junction temperature measurement for GaN-basedhigh-power white LEDs by the peak-shift method is proved by simulation andexperiment. It is concluded that selecting an appropriate pulse width is useful todecrease the test error.(3) The JTs of LED are measured when the LED is driven by constant current atdifferent ambient temperatures by the peak-shift method. It is concluded that the testerror of the peak-shift method is less than±4°C. Then the JTs of LED are measuredrepeatly when the LED is driven by different-amplitude constant current. It is concluded that peak-shift method has little repetitive error.(4) Spectral crosstalk occurs when the optical fiber is used directly to acquirethe LED spectrum in a LED array. To analyze this spectral crosstalk, a distributedhigh-power LED array spectral crosstalk model is first constructed in the spectralsimulation software. By analyzing, it is concluded that adding a diaphragm is usefulto prevent the stray light from entering into the optical fiber, and reducing thespectral crosstalk. When with a diaphragm, JT of the central LED in the distributedLED array is measured by the peak-shift method. The accuracy is accurate shownfrom the testing results.(5) The influence of testing angle between the optical fiber port and LEDemitting center changes on peak wavelength is analyzed. The near-fieldmeasurement system (SIG-400system) is used to test the emitting spectrum withdifferent testing angles. It is concluded that the changes brings a huge influence tothe test result of peak wavelength. Therefore, the testing angle should keep the sameduring the calibration and test process.(6) With a flat field imaging grating as the spectrophotometry element, ajunction temperature measurement system for LED array is developed. The systemcan acquire the spectrum of the muti-channel at the same time separately.(7) The optic correction results of the system shows that spectrum of the sixLEDs’spectrum can be well distinguished on the CCD. Thus, the system can acquirethe spectrum of all the six channels at the same time separately. Then, the calibrationof the wavelength and the CCD responsivity is finished. What is more, severaltechnical parameter of the system is tested.(8) Based on peak-shift method, junction temperatures of six LEDs are measuredusing the system. The test result shows that the spectrum shift red and the peakwavelength increases with the increasing JT. Moreover, the test error is less than±6℃.