A 3D fan-out packaging method for the integration of 5G communication RF microsystem and antenna is studied.First of all,through the double-sided wiring technology on the glass wafer,the fabrication of 5G antenna arra...A 3D fan-out packaging method for the integration of 5G communication RF microsystem and antenna is studied.First of all,through the double-sided wiring technology on the glass wafer,the fabrication of 5G antenna array is realized.Then the low power devices such as through silicon via(TSV)transfer chips,filters and antenna tuners are flip-welded on the glass wafer,and the glass wafer is reformed into a wafer permanently bonded with glass and resin by the injection molding process with resin material.Finally,the thinning resin surface leaks out of the TSV transfer chip,the rewiring is carried out on the resin surface,and then the power amplifier,low-noise amplifier,power management and other devices are flip-welded on the resin wafer surface.A ball grid array(BGA)is implanted to form the final package.The loss of the RF transmission line is measured by using the RF millimeter wave probe table.The results show that the RF transmission loss from the chip end to the antenna end in the fan-out package is very small,and it is only 0.26 dB/mm when working in 60 GHz.A slot coupling antenna is designed on the glass wafer.The antenna can operate at 60 GHz and the maximum gain can reach 6 dB within the working bandwidth.This demonstration successfully provides a feasible solution for the 3D fan-out integration of RF microsystem and antenna in 5G communications.展开更多
针对高频超声检测倒装芯片缺陷的精度易受噪声影响以及高频超声信号维度高的问题,提出一种基于K-奇异值分解(K-Singular value decomposition,K-SVD)训练局部字典的高频超声信号稀疏去噪方法。采用K-SVD训练字典来减小信号与字典中原子...针对高频超声检测倒装芯片缺陷的精度易受噪声影响以及高频超声信号维度高的问题,提出一种基于K-奇异值分解(K-Singular value decomposition,K-SVD)训练局部字典的高频超声信号稀疏去噪方法。采用K-SVD训练字典来减小信号与字典中原子之间的误差,并针对K-SVD不能训练高维度字典的问题,将高频超声信号分段,在低维度字典上对局部信号进行稀疏分解,从而降低训练字典和稀疏分解的计算复杂度;利用信号的全局最大后验概率(Maximum a posteriori probability,MAP)估计重构信号,消除因局部处理带来的信号跳变,实现高频超声信号的去噪。仿真和试验结果证明,提出的方法能够有效的去除高频超声信号中的噪声,与在全局字典上进行高频超声信号的稀疏分解相比,采用局部训练字典对信号进行稀疏分解在保证去噪性能的同时降低了计算复杂度。展开更多
This paper proposes a novel nondestructive diagnostic method for flip chips based on an improved semi-supervised deep extreme learning machine(ISDELM)and vibration signals.First,an ultrasonic transducer is used to gen...This paper proposes a novel nondestructive diagnostic method for flip chips based on an improved semi-supervised deep extreme learning machine(ISDELM)and vibration signals.First,an ultrasonic transducer is used to generate and focus ultrasounds on the surface of the flip chip to excite it,and a laser scanning vibrometer is applied to acquire the chip’s vibration signals.Then,an extreme learning machine-autoencoder(ELM-AE)structure is adopted to extract features from the original vibration signals layer by layer.Finally,the study proposes integrating the ELM with sparsity neighboring reconstruction to diagnose defects based on unlabeled and labeled data.The ISDELM algorithm is applied to experimental vibration data of flip chips and compared with several other algorithms,such as semi-supervised ELM(SS-ELM),deep ELM,stacked autoencoder,convolutional neural network,and ordinary SDELM.The results show that the proposed method is superior to the several currently available algorithms in terms of accuracy and stability.展开更多
文摘A 3D fan-out packaging method for the integration of 5G communication RF microsystem and antenna is studied.First of all,through the double-sided wiring technology on the glass wafer,the fabrication of 5G antenna array is realized.Then the low power devices such as through silicon via(TSV)transfer chips,filters and antenna tuners are flip-welded on the glass wafer,and the glass wafer is reformed into a wafer permanently bonded with glass and resin by the injection molding process with resin material.Finally,the thinning resin surface leaks out of the TSV transfer chip,the rewiring is carried out on the resin surface,and then the power amplifier,low-noise amplifier,power management and other devices are flip-welded on the resin wafer surface.A ball grid array(BGA)is implanted to form the final package.The loss of the RF transmission line is measured by using the RF millimeter wave probe table.The results show that the RF transmission loss from the chip end to the antenna end in the fan-out package is very small,and it is only 0.26 dB/mm when working in 60 GHz.A slot coupling antenna is designed on the glass wafer.The antenna can operate at 60 GHz and the maximum gain can reach 6 dB within the working bandwidth.This demonstration successfully provides a feasible solution for the 3D fan-out integration of RF microsystem and antenna in 5G communications.
文摘针对高频超声检测倒装芯片缺陷的精度易受噪声影响以及高频超声信号维度高的问题,提出一种基于K-奇异值分解(K-Singular value decomposition,K-SVD)训练局部字典的高频超声信号稀疏去噪方法。采用K-SVD训练字典来减小信号与字典中原子之间的误差,并针对K-SVD不能训练高维度字典的问题,将高频超声信号分段,在低维度字典上对局部信号进行稀疏分解,从而降低训练字典和稀疏分解的计算复杂度;利用信号的全局最大后验概率(Maximum a posteriori probability,MAP)估计重构信号,消除因局部处理带来的信号跳变,实现高频超声信号的去噪。仿真和试验结果证明,提出的方法能够有效的去除高频超声信号中的噪声,与在全局字典上进行高频超声信号的稀疏分解相比,采用局部训练字典对信号进行稀疏分解在保证去噪性能的同时降低了计算复杂度。
基金supported by the fellowship of China Postdoctoral Science Foundation(Grant No.2021T140279)the National Natural Science Foundation of China(Grant Nos.51705203,51775243 and 11902124)“111”Project(Grant No.B18027)。
文摘This paper proposes a novel nondestructive diagnostic method for flip chips based on an improved semi-supervised deep extreme learning machine(ISDELM)and vibration signals.First,an ultrasonic transducer is used to generate and focus ultrasounds on the surface of the flip chip to excite it,and a laser scanning vibrometer is applied to acquire the chip’s vibration signals.Then,an extreme learning machine-autoencoder(ELM-AE)structure is adopted to extract features from the original vibration signals layer by layer.Finally,the study proposes integrating the ELM with sparsity neighboring reconstruction to diagnose defects based on unlabeled and labeled data.The ISDELM algorithm is applied to experimental vibration data of flip chips and compared with several other algorithms,such as semi-supervised ELM(SS-ELM),deep ELM,stacked autoencoder,convolutional neural network,and ordinary SDELM.The results show that the proposed method is superior to the several currently available algorithms in terms of accuracy and stability.