In the circuit board industry, an increasing number of parts and boards are proving to be difficult to inspect with automated optical inspection (AOI) because the solder is invisible. Furthermore, high-quality requirements such as bonding strength of the automobile industry and full surface inspection of the solder are increasing. To address these needs, Omron has introduced new technology for accomplishing inspections within the required inline take time (the rate at which a product must be completed to meet customer demand). This has been one of the most challenging requirements for computed tomography (CT) X-ray automatic inspection equipment. For continuous imaging technology, highly accurate positioning control and high-speed image sensing are required.
The Case for a New Inspection Method
In recent years, remarkable technological advancements have been made in electric vehicles (EVs), advanced driver-assistance systems (ADASs) and even automated driving. For the world of circuit board mounting, this means a move towards further densification, while an increasing number of parts and PCBs have visually inaccessible soldered joints that cause difficulties in visual inspection. Typical examples of these include fillet-less chips and ball grid arrays (BGAs) with solder joints arranged on the underside of the package.
The automotive industry imposes particularly stringent quality assurance requirements to protect consumers, and suppliers are often required to perform inline full-surface circuit board inspections (rather than sampling inspections) and measure solder shapes and inspect down to bonding strength. Compounding this is the problem of line worker shortages, which is partly responsible for the current rapid increase in demand for high-precision, high-quality automated inspections.
Hence, events in the mounting industry like circuit board quality issues and production line stoppages can pose serious risks to customers. An outflow of defective circuit boards would immediately lead to a crisis that could threaten the safety of people and society. For this reason, it is more important than ever to provide a mechanism that prevents any outflow of defective circuit boards to the market.
In response to these trends, Omron developed its AXI (automated X-ray inspection) system, which has become widely used in surface mount technology (SMT) production lines thanks to its ability to inspect visually inaccessible items like solder joints on the underside of parts. Because of the problem with takt time, however, a conventional model has been used mainly for offline sampling inspections or for inline inspections of key parts only.
This article presents an outline of the technologies employed for the VT-X750 Series automated inline CT X-ray inspection system ( Figure 1) to improve this problem and achieve speeds sufficient for inline use in automotive circuit board mounting processes, thereby allowing quality assurance of circuit boards in large quantities.
Achieving High Image Quality with CT-Based AXI
The major types of X-ray-based diagnostic imaging methods include two-dimensional (2D) X-ray, tomosynthesis, and computed tomography. The 2D X-ray method is used to obtain one image per shot with an X-ray source, a workpiece, and an X-ray camera arranged vertically ( Figure 2). The image projected by this method is recorded as two-dimensional data. While capable of image acquisition in a shorter time, this method is inferior to the other methods in terms of image quality because the amount of data that it handles is small.
The tomosynthesis method is used to obtain a certain number of images of a workpiece in a relative position to an X-ray source or X-ray camera within a limited angular range. This method allows for the acquisition of tomographic images with the desired heights highlighted ( Figure 3). Although it is more time-consuming than the 2D X-ray method, tomosynthesis enables faster image acquisition than the CT method and is superior to the 2D X-ray method in terms of image quality. It must be noted that if tomographic images are captured at a far enough distance from the focus position of the X-ray source or camera, they tend to be blurrier than CT images.
The CT method is used to obtain a number of images of a workpiece in a relative position to an X-ray source or camera during a 360-degree rotation and reconstruct them into three-dimensional (3D) data. This method handles a larger volume of data than the other methods and therefore provides the best image quality. Its strength is that it enables the extraction and use of not only horizontal planar direction data, but also height direction data from the restructured 3D data. Even when captured far from the focus position of the X-ray source or the X-ray camera, a tomographic image using this method will have a clear, low-blur image quality. On the other hand, this method takes more time for image acquisition and usually delivers a higher dose to the workpiece.
The AXI Solution
Omron adopted a new inspection method that can identify the desired points in 3D data and perform image-based diagnosis to accurately inspect the shape of each solder joint surface. The Omron AXI solution takes advantage of the CT method and enables high-precision inspections that are free from circuit board underside restrictions. Its major technical components consist of hardware that is capable of safe, high-precision sensing, along with software that allows for high-speed control with excellent responsiveness.
The hardware consists largely of mechanical, electrical, and imaging components. Therefore, the design parameters — such as electromechanical safety, shielding, axis motion accuracy, control responsiveness, image quality, and imaging rate — play an important role in ensuring system performance. The software part of the system consists of assembly optimizer for machine difference corrections, a main application for inspection program development, a reconstruction process for turning captured images into 3D data, and an algorithm used to perform the inspections of the obtained 3D data. These technical components are related to each other in a complex manner and must work together seamlessly within each function module for high-precision, high-speed inspections. This is particularly important for high-quality CT image acquisition, which is the core of this technology and provides the basic performance of the imaging devices, high-precision geometry design and control, and robust correction processing and inspection algorithms.
The following sections take a look at each of these capabilities.
1. Basic Performance of Imaging Devices (FPD and X-ray Source)
The flat panel detector (FPD) is a camera that first converts X-rays into light via a fluorescent emitter called a scintillator. The light is then converted into electrical signals to obtain digital images, which can acquire high sharpness and sensitivity via pixel-by-pixel loading. The AXI system is equipped with a complementary metal oxide semiconductor (CMOS) type of FPD to help obtain high-definition images of the object. Each parameter is designed to optimize the images’ contrast to suit the part or object that is being isolated for inspection purposes.
X-ray sources can be generally categorized into two types: open tube and closed tube. An open-type X-ray source has a few disadvantages: (a) it must be installed along with a vacuum pump and other associated equipment outside it, (b) it has a high running cost due to such factors as a short-life filament, and (c) the radiation source itself has a large weight. Meanwhile, in the case of a closed tube, the X-ray generator is constantly kept in a vacuum in a hermetic glass container. Therefore, this type of radiation source features a compact body and does not need a pump installed outside its tube. The system makes use of a micro-focus closed-tube radiation source featuring a lightweight body and a small focal diameter.
2. High-Precision Geometry Design and Control
For the geometry design of the inline AXI system, a parallel table type has been adopted instead of a rotary table type ( Figure 4), since the former type can obtain CT images by changing the physical position of the X-ray source or X-ray camera (FPD) with respect to the object. A rotary table type has a round, narrow field of view for 3D data imaging, which can lead to images being blurred at their edges and limitations on high-speed imaging due to the rotation speed limit 1. What is of particular importance about this parallel table type is the positioning accuracy of the XY-axis rotation trajectory. That said, the Z-direction axis positioning accuracy also matters because this system is driven in the Z-axis direction when switching the inspection resolution or when tracing the bow of the workpiece under measurement by a displacement gauge.
A high-precision guide is provided for each axis to achieve a positioning accuracy on the order of micrometers using proprietary motor control technology. The XY-axis rotation accuracy in particular helps to obtain increasingly better CT images with the increase in roundness of the rotation trajectory. Serving as the foundation are high-quality mechanical parts mounted at the required degree of parallelism/straightness and the control technology based on the programmable logic controller (PLC) or servomotor that enables high-precision synchronous driving. In addition, the imaging devices are each supported with a high-rigidity frame and designed with a ruggedized architecture to keep all terminal parts unaffected by vibrations.
Thus, to obtain clear CT images, it is important not only to pursue the above-mentioned basic performance of the imaging devices or parameters associated with image quality, but also to develop a design that takes into consideration parameters associated with the mechanical components ( Figure 5). These parameters relate to the mechanical rigidity or weight balance of the system, the geometry design specifying how to turn the imaging devices, and the system architecture for realizing high-precision axis positioning control.
3. Robust Correction Processing and Algorithm
The AXI performs rapid processing and control of sensed information — not only in its hardware but also in its software application developed from proprietary reconstruction processes and algorithms and a deep knowledge of visual inspection systems — to achieve a high degree of solder shape reproducibility and improve the inspection of solder joint surfaces. Whenever a workpiece is brought into the system for inspection, there will be variations in the stopping position. For this reason, the AXI system is equipped with a visible light source and a camera as well as a means of correcting variations or rotational shifts of the feed stop position on the workpiece conveyor using images obtained in visible light.
To accurately perform algorithmic processing, sufficiently high precision is required to extract the correct tomographic positions in a reconstructed image. The AXI system is therefore equipped with a displacement sensor and controller that serve the purpose of measuring the bow of the workpiece or the amount of deflection thereof and correcting the Z-axis height position ( Figure 6).
Inline Full-Surface Inspection
Among the major challenges for the realization of inline full-surface inspections are improving the takt time and system hardware reliability. The latter relates to the weight and size of the enclosure, the dose to a workpiece under inspection, and the ease of maintaining it. The following sections describe each of these challenges.
1. Takt Time Problems
While capable of providing high-precision image quality, the CT imaging method must perform imaging during a 360-degree rotation and takes a large amount of takt time to do so. As a result, this method has a drawback of failing to meet the inline takt time. As the result of many engineering validation tests, it turned out that mechanical moving speeds, and especially the XY-axis rotation speed, needed a major improvement.
The so-called “stop-and-go” method to obtain images with the axes held stopped during rotation on a 1PJ basis (1PJ = projection: a single, pre-reconstruction-phase image obtained at an angle during rotation). Hence, it needs a large takt time per rotation. This stop-and-go method is resistant to image blurring because image acquisition occurs only after the complete cessation of the axes’ motion. The rotating hardware unit, however, experiences strong vibrations and impacts, so the X-ray source must be of a fixed type. Therefore, Omron adopted a configuration in which the stage and the FPD rotate instead. This resulted in an FPD with a large rotation radius and hence in a mechanism that took a large amount of CT image acquisition time per rotation.
2. Enclosure Size and Weight Issues
For actual inline use, consideration must be given to the size and weight of the system’s enclosure, because the X-ray inspection system is relatively bulky and weighty with respect to the other SMT equipment. This means that the transportability in a standard-size shipping container or the availability of a carry-in route and a sufficiently durable floor must be taken into consideration with regards to delivery and installation onto the customer’s production line. If the weight and size of a unit increase as a result of the increased rigidity of mechanical parts, the system will become even heavier and bulkier, leading to an increase in transportation and installation workloads and costs. This poses an obstacle to the introduction of the capital investment that must be addressed when intending to introduce any system as inline equipment onto the production lines in the SMT market.
3. Dose Reduction Issues for Workpiece Inspection
An increase in the number of large circuit boards or parts to be inspected will lead to a longer irradiation time, thereby resulting in a higher dosage to the circuit board as a whole. In addition, there is also a tendency for more circuit boards to be mounted with semiconductor devices and other parts that are vulnerable to radiation exposure. Continuous efforts must be made to explore technologies for reducing the dosage on circuit boards.
4. Maintainability Issues
The conventional model had several problems with maintainability. Both the inspection space and the device housing unit were in the same shielded space, and the individual devices were vertically arranged in a way that was poorly accessible by hand, invisible from a standing position, and unable to easily allow tools in. Inline operation of a system in a customer’s production line meant an increase in the availability rate of the system. In addition, because the system was between other pieces of equipment, the downtime required for maintenance or the like had to be reduced to the very minimum. Moreover, the system had to be improved so that maintenance of all its components and parts could be performed from both the front and back.
Solving the Takt Time Problem with Continuous Imaging
The first characteristic to note is that the mechanism keeps running without stopping during rotational imaging. To reduce the image acquisition time per field of view (FOV), Omron developed a technology that allows the mechanism to rotate without stopping and obtain one FOV’s worth of all images rather than repeat the cycle of moving, stopping, and acquiring each image as in the conventional model’s stop-and-go method ( Figure 7). The mechanism repeats rotary motion at a constant speed. The key is to synchronize the axes’ motion and make the rotation trajectory as close to a perfect circle as possible. This is supported by a synchronous complete circular trajectory control technology based on the PLC NJ controller and the 1S servo system. This technology supports not only simple high-speed rotation but also dynamic and high-precision synchronous control.
The second characteristic to note is that this system performs image acquisition and data processing at high speed. This system is built in such a manner as to obtain one image after another — even in the middle of mechanical action — on the basis of image acquisition triggers issued by a proprietary circuit network. It is known that high-speed image acquisition during movement of the mechanism tends to result in blurred images 2. Therefore, the system records the image acquisition start time on a 1PJ-by-1PJ basis and then reduces blur using its image reconstruction capability that takes into consideration the relative positions of the FPD and the X-ray source at that time. Moreover, with the addition of a graphics processing unit (GPU) for processing speed improvement, the system has successfully achieved a speed that is fast enough to process a series of images transmitted at high speed.
The advantage of this technology is that axis actions proceed at a constant speed and thereby help to reduce vibrations and impacts on the rotating hardware units. This led Omron engineers to adopt a geometry structure in which the X-ray source rotates with the stage unit — a heavy object — fixed in place. As a result, the rotation radius of the FPD and that of the X-ray source were reduced to approximately 60 percent of their respective corresponding conventional values and contributed to a significant reduction in the speed required for a full rotation ( Figure 8). These reduced rotation radii also helped to minimize image blur.
A sophisticated architecture can maximize the advantage of continuous imaging technology. Moreover, the availability of faster inline CT inspections makes it possible to inspect far greater numbers of circuit boards than in conventional offline sampling inspections. This helps improve product quality in a wider range of customers’ circuit board production lines in preparation for the day when automated driving will become prevalent.
This article was contributed by Omron Americas, Hoffman Estates, IL. For more information, visit here .
- Sugita, S. High-speed CT inspection technology for wider coverage of mounting quality assurance (in Japanese). Proceedings of the 52nd Soldering Breakout Session, Japan Welding Society, 2011, p. 4.
- Japanese Society of Radiological Technology (Supervising Ed.). Ichikawa, K.; Muramatsu, Y. eds., Standard X-Ray CT Image Measurement (in Japanese). Ohmsha, 2009, pp. 27- 28.