Comparison of vision systems between Fuji and Siemens mounters

The vision system has now become an important part of the high-precision mounting machine. This paper first introduces the image processing and structure of the placement machine. Based on the comparison of the vision systems of FUJI and SIEMEMS placement machines, it briefly describes how the modern high-precision SMT equipment adapts to the placement of micro spaced pieces. Abstract The vision processing system is the critical portion of the high – placement ma-chine.  This paper firstly introduces the principle of vision processing and structure of the placement machine.  Based on the contrast of vision processing system between fuji and siemems placement machine, we expound how the modem SMT machine meet the requirement of placing the micro pitch parts.  keywords Image Processing Gray value method Patch precision Image Processing Algorithms, As a result, the SMT's key equipment - SMT placement machine has put forward higher requirements for placement accuracy. This paper makes a detailed comparison between the vision systems of FUJI (mainly IP3, CP6) and SIEMEMS (S80F) mounters from the perspective of application, so that relevant technicians engaged in SMT can better understand how the image processing technology of today's high-precision mounters meets the requirements of high-precision mounting of devices. The principle of SMT machine vision system is a computer based image observation, recognition and analysis system. It mainly uses the camera as the sensing component of the computer sense, or the detection component. The camera senses the light intensity distribution of the object within a given field of vision, and then converts it into analog electrical signals. The analog electrical signals are digitized into discrete values through the A/D converter. These values represent the average light intensity of a given point within the field of vision. The digital image thus obtained is covered by regular spatial grids, and each grid is called a pixel. Obviously, the target image occupies a certain number of grids in the pixel array, as shown in Figure 1. The computer processes the above pixel array containing the digital image of the target, compares the image features with the reference image input into the computer in advance, analyzes and judges, and the computer gives instructions to the actuator according to the results. Digital Image Video Image Figure 1 Image digitization in machine vision system gray resolution. The gray value method uses the multi-level brightness of the image to represent the resolution. The gray resolution is specified at what discrete value is the measurement light intensity given by the machine. The smaller the light intensity to be processed, the higher the gray resolution. The vision system is composed of vision hardware and software. Hardware is generally composed of image detection, image storage and processing, and image display. The hardware functions of the vision system are shown in Figure 2. Hardware processor memory CPU CCD A/D D/A monitor actuator Fig. 2 Vision system principle Camera is the sensing part of the vision system, and the vision used for the mounting machine adopts solid-state camera and CCD camera. The main part of the solid-state camera is an integrated circuit. A CCD array composed of many small photosensitive elements is made on the integrated circuit chip. The electrical signal output by each photosensitive element is inversely proportional to the corresponding reflected light intensity on the observed target. This electrical signal is recorded as the gray degree of a pixel. The coordinates of the image element determine the position of the point in the image. A large amount of information obtained by the camera is processed by a microprocessor. The processing results are displayed by industrial television. There are communication cables between cameras and microprocessors, and between microprocessors and actuators and displays. Generally, RS232 serial communication interfaces are used. 3 The main factors that affect the accuracy of the vision system are the number of pixels of the camera and the optical magnification. The more pixels of the camera, the higher the accuracy; The higher the magnification of the image, the higher the accuracy. Because the larger the optical magnification of the image is, the more pixels of the junction area are, so the higher the accuracy is. On the IP3 of FUJI, precision is required for devices with 0.15mm foot width. However, if the magnification is too large, it is more difficult to find the device, and it is easy to lose parts, which reduces the mounting rate. Therefore, appropriate optical magnification should be selected according to actual needs. 4 Comparison between FUJI and SIEMENS vision systems 1. PCB positioning The IP and CP of FUJI have a dedicated MARK CAMERA, which is used to obtain the position, size and shape of the marker points on the PCB and read the center position. When positioning a PCB, at least two punctuation points (based on the X and Y TABLE level,) are required on the PCB to search around the center of each landmark point in a certain range. If no target is found, the search range will be expanded (can be set in the program). After determining the location of the marker point, compare it with the coordinates in the program to determine the deviation, which is reflected in the three values of X, Y and Q, and then correct the mounting data. Siemens is about the same. 2. Device detection and centering FUJI uses one large camera and one small camera to identify and align different components, and perform the detection function at the same time. Different irradiation methods are used for different devices. The J type feet (PLCC, SOJ, BGA) are illuminated by front lights, and the others are illuminated by back lights. The nozzle on the mounting head absorbs the device at the FERDER position specified in the program, and the absorption should be at the center of the device as much as possible, especially for larger devices such as the PLCC84, which is necessary, otherwise, it will not pass the image processing. After getting the shape image of the component from a certain position, the edge data is obtained through a special algorithm (guided by the component) to get the center position. Compared with the data in the program, the deviation values of X, Y and Q are obtained. At the same time, the correction data is given, Perform the following detection functions: whether the actual device has any deviation from the device described in the PART DATA (package: including the number of pins, pin position, pin length, and external size), whether the pin is bent, coplanarity of the pin, and polarity detection. When the chip mounter performs the detection function, it compares the features of the detected device with the stored packaging device. If the detection fails, the device may be packaged incorrectly, or the material may be loaded incorrectly, or the device may be defective, and the system will order the mounting head to send the device to the scrap tape. In practical application, if the above problems occur, the specific reasons should be carefully analyzed. FUJI provides the image of the industrial CRT display viewable device. Through the on-site console on the machine, it can be manually operated to obtain the image of the real device. There are many ways to check the packaging and actual differences in the device program. CRT can prompt where there is an error (BUG). When there is an error, the screen also provides an error code, which is convenient for analyzing the cause of the error and providing suggestions for modification. In the vision software (PART DATA), there are different VISION TYPES for different devices, which are also different image processing algorithms. There are different gray scale solutions for different device pins, different exposure sequences for pins, which can verify the number of pins, and polarity detection can be performed for polar devices, reflecting the adaptability of the mounter. Fuji's IP3 image processing adopts a variety of advanced technologies. The pixel of the gray scale processing system has doubled compared with the previous machine (BINARY), and it can paste BGA, FLIP CHIP, CONNECTORS and a variety of special-shaped devices. Siemens 80F4 is also a multi-function machine. It has two sets of chip heads, namely, rotary head and IC head. The rotary head is composed of 12 patch heads, which can be used to mount PLC C44. The IC head can be mounted to 55mmX55mm devices. Siemens SMT has three CAMERAs, namely PCB CAMERA, COMPONENT CAMERA and IC CAMERA. PCB CAMERA is mainly used to map the marker points on the machine and PCB. COMPONENT CAMERA is located above the rotating head, which is used for optical alignment of small devices and adjustment of placement position. IC CAMERA focuses on optical alignment of large devices. Siemens SMT has three main photography methods, namely, block devices (such as general CHIP components, SO (including PLCC) devices, and BGA. In the optical alignment of CHIP components, only the parallel light is available, and only the edge of the device is confirmed, so as to find the center of the device and calculate the error to be adjusted when mounting. While optical alignment of SO devices is carried out, the relative position of each pin shall be detected. If the pin is not in the standard position, it will also be judged as unqualified devices. Since BGA is a spherical solder ball, its photographic processing is different, so each solder ball should be tested. The location and brightness of the solder ball are the testing contents. If there is any substandard device, it will be discarded as a substandard device. Siemens has obvious advantages over FUJI in image processing of PLCC. The main reason is that the light source of FUJI is a directional light, and the processing results for J-pin are the same. Only the lower end of J-pin is reflected. Comparatively speaking, Siemens light source has side light and also has reflection image on the inclined plane of the J-pin, which can perform a relatively comprehensive optical detection on PLCC. Siemens also has obvious advantages over FUJI-IP3 in mounting PLCC, and side light plays an important role in optical detection of BGA. 5 Conclusion In the comparison of the visual systems of FUJI and SIEMENS, we have a deeper understanding of the image processing technology of the placement machine. It can be seen that the high-precision placement machine integrates a variety of modern high-tech technologies such as computer, light, electronics, automatic control, etc. With the rapid development of these technologies, the placement opportunity is developing towards a higher speed, higher precision, and stronger functions. We have written the experience of using SMT placement machine into this article, hoping to communicate more with colleagues engaged in SMT placement work to achieve the goal of mutual improvement.