The Structure Function Method Optimization And Data Processing Technology For Thermal Resistance Measurement Of Semiconductor Devices

Power semiconductor devices and integrated circuits are widely used in the fields of microwave communication, power electronics, military defense and other fields. But the increase of current density will result in the operating temperature rose and rose in active region and the thermal resistance becomes larger and larger when the device is working, which seriously affects the electrical characteristics and reliability of the device. Therefore, it is very important to improve the thermal design and reliability of device by studying a method to measure operating temperature rise and thermal resistance precisely.In this paper, the operating temperature rise is measured using temperature sensitive electrical parameter method(electrical method). Based on the thermal resistance tester hardware equipment independently developed by lab, related control software will be developed and resistance algorithm will be optimized. The research work includes the following aspects:1. Based on the previous research, improving fitting method of original discrete data of transient temperature rise curve. Traditional multi-exponentials fitting is replaced by a nonparametric fitting algorithm called locally weighted scatterplot smoothing(LOWESS) in order to describe temperature variation more precisely. Thermal time-constant spectra of transient temperature response curves obtained by two fitting methods are extracted respectively and are compared to analysis the heat conduction system structure, which can prove the advantage of LOWESS algorithm.2. Using standard deviation of data before power as the judgment of smoothing scale of original transient temperature response curve, and forward backstepping 1ns~1?s temperature change trend by utilizing the combination of exponential fitting and LOWESS smoothing method. Consequently, a differential structure function curve containing the real thermal resistance constitutes of device will be obtained.3. Based on the existing technology, realizing the software and hardware optimization design of a high and low voltage LED thermal resistance tester. In hardware part, separate testing current of high and low voltage circuit, replace 16 bits DAC chips, use programmable power to provide working current and automatically adjust reference voltage. In software part, mainly rewrite controlling algorithm of programmable power.4. Complete massive data processing and interface control software development of Si C thermal resistance tester with eight channels and VDMOS & diode with two channels integration tester. Using Microsoft Visual Studio 2010 as the integrated development environment, design many controlling algorithm, including data sampling of USB-4600 card, data save path, multi-channel measurement, curves display, power model select of eight channels, constant current model select of VDMOS and so on, which realize software and hardware conjointly intelligent measurement.