BGA植球机三维计算机辅助模块化设计

球栅阵列(BGA)封装能很好地满足IC芯片引脚数越来越多、IC尺寸越来越小的要求,成为新的IC封装技术主流。分析BGA植球机的工艺过程和工作原理,完成该设备的机械部分设计。在对BGA全自动植球机进行运动仿真后,可以确定各个机构运动正常。所设计的STAR-04Z型BGA全自动植球机产品样机能够满足市场的功能需求。     

BGA全自动植球机视觉检测技术及应用

在线缺陷检测是球栅阵列（BGA）植球机研发中的一个难题.本文介绍了视觉检测技术在BGA植球机上的应用.设计了一种图像采集和图像处理并行进行的工艺过程方案,提高了系统的检测速度.通过图像预处理之后提取图像的边缘特征,采用模板匹配法进行识别检测,并在此过程中融合了序贯相似度检测算法（SSDA）,以提高模板匹配的速度.试验证明,该视觉系统的工艺和检测过程均为可行且行之有效的方案.     

BGA植球的实验研究与工艺方法

为了节约生产成本,保障生产进度,实现BGA器件的再次利用,文中开展了BGA植球工艺的研究。通过理论分析确立了BGA植球质量的4个关键因素:植球方式、助焊剂、植球钢片、回流温度曲线,并对4个因素进行了一系列对比工艺实验,根据实验结果确立了最佳的工艺参数,形成了较完善的植球工艺。重植球BGA的性能检测表明,重植球BGA芯片二次利用时能正常工作且性能稳定。验证实验结果证实了该植球工艺的可行性,可有效应用于BGA返修中。     

BGA/CSP植球机供球机构的设计及应用

研究设计BGA/CSP植球机的供球机构,采用翻转预埋方式供球,锡球直径0.15~0.76 mm,解决市场上常见植球机振动盘供球方式易造成锡球缺失和粘连问题。针对该供球方式水平定位精度要求高,对真空植球头和供球机构之间的定位误差进行分析,得出供球机构传动链回程误差对植球头吸球效果影响较大,设计了气缸夹紧机构进行位置补正。对气缸夹紧机构进行受力分析和力矩分析来选用合适的气缸。最后通过设备调试,供球机构铺球效果良好,植球头能够完全吸取锡球,为高植球精度打下基础。     

基于机器视觉的BGA植球后缺陷检测技术的研究

为了对BGA植球后的缺陷进行检测,运用了机器视觉的检测方法,利用图像处理技术对原始图像进行灰度化、二值化和噪声抑制,将焊球和背景基板图像进行有效的分离,突出了焊球的边缘信息,再运用圆的Hough变换对焊球边缘进行拟合,计算出每个焊球的圆心位置和直径,最终通过粗检测和精检测有效地分别识别出了焊球的各种可能缺陷,提高了检测效率和正确率。试验证明该方法是行之有效的方案。     

BGA连接器植球工艺研究

球栅阵列（Ball Grid Array,BGA）封装具有体积小、引脚密度高、信号完整性和散热性能佳等优点，因而广泛应用于大规模集成电路的封装领域。植球工艺作为BGA封装（连接器）生产中的关键工艺会直接影响器件与电路的性能及可靠性，现从植球工艺路线、BGA连接器设计要求、植球工艺参数及关键技术、试验及检测要求等几个方面，阐述了影响BGA连接器植球工艺实现的各种因素，借以提高BGA连接器产品的可靠性及稳定性。     

滚动轴承的缺钉、缺球及注脂定位检测系统

针对脂润滑滚动轴承装配过程中缺钉、缺球及注脂位置检测问题,研究了可一次提取轴承图像并先后完成2种检测的图像检测系统。系统采用Lab VIEW虚拟仪器技术,完成对轴承图像的采集、缺钉缺球缺陷特征识别提取、注脂头转动角度计算等图像处理的工作,并结合下位机单片机对整个检测流程和注脂头电动机进行控制。实践表明,该系统检测准确,效率高,工作稳定可靠。     

基于改进U-Net的球栅阵列气泡缺陷检测方法

针对现有基于深度学习的图像分割算法在球栅阵列(BGA)焊点气泡检测中检测效率较低,无法满足工业生产中实时性的检测需求,提出了一种基于改进U-Net的球栅阵列缺陷识别方法。该方法在现有的U-Net经典网络的基础上提出用深度可分离卷积与密集连接结合的轻量密集连接单元替换常规的卷积单元,同时添加多尺度跳跃连接减少编解码特征之间的差异,实现针对BGA焊点气泡的精确分割和提取。采用自建数据集对该方法的有效性进行实验,结果表明,改进的U-Net模型网络在减少U-Net网络计算复杂度的同时提升了网络性能,能够增加BGA焊点气泡的检测效率。     

基于计算机视觉技术的BGA芯片精准焊接定位的研究

针对BGA芯片的特点,本文介绍一种基于计算机视觉技术的单摄像头双目成像焊接定位系统,本文介绍了单CCD视觉系统的构成及原理,并采用经典算子和利用灰色系统理论对BGA芯片图像进行处理。实验结果证明灰色系统理论是一种有效的、具有可调功能的边缘检测新算法,能够较为准确地检测出有用的边缘信息,具有一定的抗噪声能力,进而有效的提高BGA芯片定位精度。     

WLP微球植球机的重力植球法研究与应用

晶圆级封装(WLP)是目前主流的高端半导体封装技术之一,如何提高焊锡植球精度是当前的各大厂商关注的热点问题。通过介绍焊锡植球的基本流程,引出并比较了两种常见的焊锡球搭载方式。并对重力植球法的关键技术进行进一步阐述,包括焊锡球供球与回收装置和机器视觉对准装置。最后通过实验对植球过程中可能产生的缺陷进行检测和观察,与半自动植球机比较,证明通过参数调整能满足实际生产所需要的植球要求。     

A fast full-field 3D measurement system for BGA coplanarity inspection

Ball-Grid-Array (BGA) has become one of the most popular packaging techniques in the electronics industry. Coplanarity inspection of BGA solder balls is critical for process and quality control in BGA manufacturing. Currently, the 3D measurement systems for BGA coplanarity inspection are mainly based on laser scanning techniques. However, they suffer from low inspection speed due to the physical line-scanning process. In this paper, a fast and cost-effective 3D measurement system for coplanarity inspection of BGA solder balls is proposed. The proposed system uses an LCD-based phase measuring technique to perform full-field 3D measurement of BGA solder balls with high accuracy. Experiments have shown that the coplanarity measurement of BGA solder balls is very efficient and effective with the proposed system. The measurement repeatability is in the micrometer range. The processing time of the proposed 3D measurement system for an image of 640×480 pixels is less than 2 s on a typical personal computer.     

BGA焊球视觉检测算法及系统设计

为了实现对BGA焊球的自动检测,建立了自动视觉检测系统。对系统所采用的焊球特征进行提取及缺陷识别,基于高斯混合模型的分类器对检测算法进行研究。根据焊球的形状和尺寸特征设计了焊球缺陷识别和分类算法,并以锡多、锡少和毛刺缺陷为例,分析典型缺陷的识别算法。以焊球形状的圆度和特征区域的面积等特征参数为评价标准,构建二维特征空间。在二维特征空间线性组合的基础上,构建基于高斯混合模型的分类器。构建了训练样本集,并对该分类器进行训练,根据训练结果并结合应用实际修正了模型,并采用测试集对该分类器进行测试验证。实验结果表明,焊球缺陷检测算法的准确度为97.06%,漏判率为0%,检测可靠度为100%。该视觉检测系统满足了工程运用中对识别准确度、稳定性、可靠性等方面的要求。     

双图像传感器的球表面展开方法

钢球作为广泛应用的轴承组件,其表面质量直接影响轴承性能.在钢球表面缺陷自动检测中,钢球表面展开技术是核心,球表面能否充分展开直接决定了表面质量检测的准确性.目前发展的钢球表面展开技术仍存在机械结构复杂、展开不充分及随机性强的问题.基于计算机视觉技术研究了基于双图像传感器的球表面展开方法.首先,介绍基于双图像传感器的展开原理及装置.其次,建立数学分析模型,解算出完全展开整个球面所需要满足的参数条件;最后,设计实验完成全覆盖球表面的图像采集,验证理论模型结果.实验结果表明,在比例系数k为0.97,2个相机45.倾斜拍摄图像时至少需要拍摄4幅图像实现全覆盖展开,与理论模型分析结果一致.此方法在精确有效展开的同时,机械结构从复杂的二维运动简化为一维运动,提高了球表面展开的可靠性,具有广泛的应用前景.     

全自动平衡机系统组成与控制

介绍了一种全自动平衡机系统的组成及体系结构,并结合这种体系结构给出了系统的控制过程,分析了控制过程中的关键技术及实现方法,提供了去重模型和高层串行通信协议,在此基础上研制了样机并给出了试验结果及结果分析。试验及实践结果表明,研制的全自动平衡机系统能较好地实现对不平衡转子的动平衡修正。     

A Rotation-Invariant and Non-Referential Approach for Ball Grid Array (BGA) Substrate Conducting Path Inspection

The aim of this paper is to locate and classify boundary defects such as open and short circuits, mousebites, and spurs on ball grid array (BGA) substrate conducting paths using machine vision. Boundary defects are detected by a boundary-based corner detection method using covariance matrix eigen-values. Detected defects are then classified by discrimination rules derived from variation patterns of eigenvalues and the geometrical shape of each defect type. Real BGA substrates with both synthetic and real boundary defects are used as test samples to evaluate the performance of the proposed method. Experimental results show that the proposed method achieves 100% correct identification for BGA substrate boundary defects under a sufficient image resolution. The proposed method is invariant with respect to the orientation of the BGA substrates, and it does not require prestored templates for matching. This method is suitable for various types of BGA substrate in small-batch production because precise positioning of BGA substrates and the prestored templates are not necessary.     

Ball array calibration on a coordinate measuring machine using a gage block

The distances between the balls of a ball array used in machine geometry calibration have to be very accurate. These distances can be calibrated using a laser measurement system, which requires specially designed optical devices and measuring probes. In this paper, a new and economical alternative method for calibrating the ball array is described. A single gage block is used for measuring the standard distance at the starting position. Then, the exact distances between the balls can be obtained by using the coordinate measuring machine (CMM) probe motion. This method does not depend on the accuracy of the CMM. Also, this method does not require expensive instruments or devices, but a CMM and a gage block. A simple “parallel-plane” bracket, mounted on to the measuring end of a CMM probe, is used to determine the centers of the balls automatically and accurately.     

大型精密球体机械化研磨的新方法

本文介绍大型4轴自动球面研磨机的工作原理、机械结构和微机控制系统。它完全改变了传动的手工研磨方法,高效率、高精度、低成本地实现了大型球体的机械化研磨。对于整球加工精度已超过超精密机床加工水平。     

一种新型自动化蓝莓采摘机的设计

针对小浆果种植领域中蓝莓收获难的问题,基于计算机软件和可编程序控制器,设计了一种新型自动化蓝莓采摘机。介绍了这一蓝莓采摘机的结构与工作原理,分析了控制方案,制作了样机,并进行了采摘试验。应用这一新型自动化蓝莓采摘机,可以满足蓝莓收获效率高、破损率低等要求。     

Volumetric distortion assessment of a five-axis machine by probing a 3D reconfigurable uncalibrated master ball artefact

In this paper, a method is proposed that uses an artefact assembled in situ and exploiting the on-machine probing capability to perform a rapid volumetric distortion assessment of a five-axis machine. The 3D reconfigurable uncalibrated master ball artefact (RUMBA) is an assembly of the machine pallet and an adjustable number of master balls located within the machine working and probing envelop. The artefact design philosophy allows flexibility in the number and positions of the balls thus conferring the reconfigurability attribute. Reconfigurability allows adaptation of the artefact to the machine topology and geometry and its construction around a fixture and workpiece to reduce disruption to machine production. On the down side, the artefact is uncalibrated as its construction concept hinders precise knowledge of its geometry. During a test, combinations of all the machine axes are used to move and probe the master balls. A mathematical model is developed in order to identify the setup errors of each master ball and the probe. Simultaneous identification of the rotary axes’ linear offset with the setup errors is proposed to improve the parameter estimation and volumetric distortion prediction. The machine contribution to the volumetric distortion is then computed from the ball centre data excluding the setup errors. Tests are conducted in a laboratory on a horizontal machining centre. The results show that the proposed mathematical procedure is capable of removing the influence of the uncalibrated artefact geometry, excluding scale, and that the proposed design is suited to the machine environment.