【摘 要】
：
Parasitic capacitances associated with overhangs of the T-shape-gate enhancement-mode(E-mode)GaN-based power device,were investigated by frequency/voltage-dependent capacitance-voltage and inductive-load switching measure-ments.The overhang capacitances i
【机 构】
：
High-Frequency High-Voltage Device and Integrated Circuits R&D Center,Institute of Microelectronics,
论文部分内容阅读
Parasitic capacitances associated with overhangs of the T-shape-gate enhancement-mode(E-mode)GaN-based power device,were investigated by frequency/voltage-dependent capacitance-voltage and inductive-load switching measure-ments.The overhang capacitances induce a pinch-off voltage distinguished from that of the E-mode channel capacitance in the gate capacitance and the gate-drain capacitance characteristic curves.Frequency-and voltage-dependent tests confirm the instability caused by the trapping of interface/bulk states in the LPCVD-SiNx passivation dielectric.Circuit-level double pulse measurement also reveals its impact on switching transition for power switching applications.
其他文献
针对目前大多数人脸识别算法参数多、计算量大,难以部署到移动端和嵌入式设备中的问题,提出了 一种基于改进MobileFaceNet的人脸识别方法.通过对MobileFaceNet模型结构的调整,将bottleneck模块优化为sandglass模块,改良深度卷积和逐点卷积的相对位置,适当增大sandglass模块的输出通道数,从而减少特征压缩时的信息丢失,增强人脸空间特征的提取.实验结果表明:改进后的方法在LFW测试数据集上准确率达99.15％,模型大小和计算量分别仅为原算法的61％和45％,验证了所提方法
针对多个固态激光雷达协同工作时,需要准确地进行外参标定的实际需求,提出一种基于点云配准的多固态激光雷达自动标定算法.该标定算法由标定物分割、初始配准和精确配准三个阶段构成.在标定物分割阶段,首先通过叠加多帧非重复扫描数据制作标定点云,再使用半径滤波和体素下采样滤波分割出纹理特征明显的 目标点云.在初始配准阶段,使用3D-HARRIS算法提取关键点,并使用方向直方图(SHOT)特征描述子进行特征描述,然后匹配对应点并使用采样一致算法完成初始配准;在精配准阶段使用迭代最近邻(ICP)算法进行精确配准,从而获得
为提高单应性估计的准确性和解决真实标注难获取的问题,提出一种具有修正功能的无监督单应性估计算法.该算法采用级联结构,其思想类似于迭代,其中每一级网络都保持相同的层数和参数量,下一级网络输出的单应性矩阵为真实矩阵与之前输出单应矩阵和的残差.考虑到模型复杂度和实时性的需求,文章采用两级网络级联.通过在COCO数据集中的5 000张图片上进行验证,结果表明,相比传统方法和其他基于深度学习的方法,所设计的级联无监督算法具有更准确的估计能力,其在测试集中的平均像素误差为0.54,较传统方法下降95.38％,运行速度
During last several years,electron acceptors for organic sol-ar cells(OSCs)have experienced three major innovations.The first invention was a fused-ring electron acceptor(FREA),ITIC,reported by Zhan et al.in 2015,which consists of an in-dacenodithienothio
在相机位姿估计任务中,参考点位置的测量容易受测量仪器或者所处环境影响,出现一定程度的测量误差,其中误差较大的点对最终求解精度的影响很大,被称为野值点.文章针对现有算法抗野值点能力较弱的问题,提出一种基于测量误差不确定性加权的快速相机位姿估计算法.该算法以经典的正交迭代算法为基础,引入M估计方法,对目标函数根据参考点的测量误差赋予相应大小的权值,最大程度地将野值点的影响排除,并引入Kronecker积对计算过程进行优化,减少了迭代过程中的冗余计算,提高运算速度.仿真实验结果表明,改进后的算法提升了原算法的抗
Artificial intelligence(AI)processes data-centric applications with minimal effort.However,it poses new challenges to system design in terms of computational speed and energy efficiency.The traditional von Neumann architecture cannot meet the requirements
A proposed inductive-phase-compensation ultra wideband CMOS digital T-type attenuator design based on an analys-is of minimising phase errors is presented in this letter.In a standard CMOS technology,the proposed attenuator is analytically demonstrated to
Recently,all-inorganic perovskites have attracted atten-tion due to good thermal stability[1-12].Among them,CsPbl3 has the most desirable optical bandgap(～1.7 eV)for applica-tions in optoelectronic devices[13-16].In general,making black-phase CsPbl3 film
The electrical properties of cubic perovskite series,CaCu3-xTi4-xFe2xO12 with x=0.0,0.1,0.3,0.5,and 0.7,have been studied by employing current density as a function of electric field characteristics registered at different temperatures and thermal variati
Silicon Hall-effect sensors have been widely used in industry and research fields due to their straightforward fabrica-tion process and CMOS compatibility.However,as their material property limitations,technicians usually implement complex CMOS circuits t