JST is one of the world’s leading manufacturers of electrical and electronic connectors. Achieving world class levels of quality is the company’s number one objective. Recently, JST was able to improve on its already extremely high levels of quality by implementing a vision system that measures parts to an accuracy of 0.1 mm, better than could be achieved in the past with a human inspector. The vision system also identifies variations much faster than human inspectors, making it possible to adjust machine parameters quickly enough to avoid defects. The result is that the number of nonconformances in the latest measurement period was reduced to 0 parts per million (ppm), well below the extremely tough six-sigma quality benchmark of 3.4 ppm.
JST, based in Waukegan, Illinois, makes over 30,000 different types of connectors including surface mount, flat flexible circuit, crimp wire-to-board, header, flat cable, wire-to-wire, board-to-wire, board-to-board, power-to-supply, wire-to-board, EMI shielded and IC memory card connectors. For most standard terminals, JST uses oxygen-free copper, which has high electroconductivity and inherent toughness and can be brazed with no chance of hydrogen embrittlement. JST’s standard products are all galvanized and treated with a special method that ensures maximum adhesion of electroplating, further reducing electrical resistance and enhancing corrosion resistance.
Tight dimensional tolerances
JST’s 2.54-mm pitch HCM solderless terminal is used in the automotive industry to attach wires to a printed circuit board using crimp-style connectors. The two primary components are the housing and retainer. The user inserts the wires into one side of the housing, then pushes the retainer into an opening on the other side. The user pushes a tab on the retainer to lock it to the housing. Both the housing and retainer are produced by injection molding to tight dimensional tolerances. But while the injection molding process is extremely accurate, there is a very small risk that a problem in the process conditions might cause the mold to be incompletely filled, a so-called short shot. A short shot that could cause the part not to work would be easily visible with the naked eye. However, JST’s tough quality specifications prohibit even much smaller short shots that cause only 0.1 mm of material to be missing… smaller than could be detected with the naked eye. Defects this small have no effect on the performance of the part, but they are still unacceptable to JST.
Manual inspection used in past
JST had maintained high levels of quality by devoting considerable time and resources to manual inspection. “We spent a lot of time continually checking these parts,” said Nate Hoselton, Facilities Manager for JST. “Often problems were not detected until a day or two after the parts had been molded. In that case, we had to go back and re-inspect a considerable number of parts. It was sometimes difficult after the fact to identify the problem with the molding machine. We maintained records for the process conditions, materials, machine parameters, etc., but the operator might have difficulty remembering anything unusual that had happened during the run. As a result, it typically took us several days and about $1,500 to investigate
Hoselton identified this application as a good one for machine vision. “We have long used machine vision in this plant because of our intense focus on quality,” Hoselton said. “In fact, we have systems from all three of the leading suppliers of machine vision systems. For this project, we selected the Cognex In-Sight® vision system because it is so easy and fast to program. I went to a Cognex training class and in only ten minutes developed a simple application that reads a business card. Based on that experience, I programmed this application in only about 45 minutes including interfacing the vision system to the assembly machine’s programmable logic controller (PLC) and human machine interface (HMI). There was no need to get internal engineering resources or an integrator involved.”
Hoselton selected an In-Sight 5600 vision system because it provides a complete solution in a modular package and does not require any additional hardware or other equipment. The 60mm x 110mm x 80mm package of the vision system easily fits within the tight confines of the manufacturing plant. The In-Sight 5600 model offers a resolution of 1600 x 1200 pixels and an image acquisition time of 15 frames per second. Cognex EasyBuilder® software walks the user step-by-step through the process of setting up a vision application. Users can drop in tools simply by clicking on features of interest.
Programming the vision system
Hoselton began the programming process by mounting the vision system on the assembly machine and connecting it to the PLC, HMI and a laptop computer running the EasyBuilder software. He pointed the vision system at a part and the resulting image appeared in the EasyBuilder interface on his laptop. He used a wizard provided by the software to set up the scale and nonlinear calibrations. He selected edge tools to identify the location of each side of the housing based on the transition from the black part to the white background. He used additional edge tools to identify each side of the opening in the part where the retainer is inserted, again based on the transition from black to white. Hoselton then used measurement tools to calculate the size of the opening by subtracting the lengths of the walls of the housing from its total length. He plugged the tolerances into the measurement tools from the blueprint. Based on these tolerances, the vision system measures the gap and sends an alert when it varies by an amount as small as 0.1 mm.
Hoselton used the point-and-click communications setup in EasyBuilder to use a signal from the PLC to alert the vision system to acquire an image when a new part is in position. He sent the results of the inspection back to the PLC and the HMI. Finally, he used additional tools to set up a user interface that shows the image of the latest part, a filmstrip that shows recent parts, and a table of results. The vision system inspects each part in only about 40 milliseconds, much faster than required strokes-per-minute production rate of the machine. The HMI used by the operator of the assembly machine is connected to the plant network so JST managers can now see the images from the machine vision system from their desks, either in real time or by searching archives.
Improving already high levels of quality
The In-Sight vision system has provided a significant improvement in the company’s already high quality levels. Manufacturing personnel receive immediate feedback from the vision system for even tiny variations, which are far too small to affect performance and too small to be seen with the naked eye. The assembly machine operator immediately gathers up the parts along with the inspection information and takes them to the quality control department. Quality control takes the parts to the molding technician where he immediately begins working to address the issue. As a result, the time required to determine the root cause of and address manufacturing variation has been reduced to one day. The cost of a nonconformance report has been reduced from $1,500 to $100.
Most important, the ability to detect and correct even the smallest variations has resulted in the company exceeding demanding six sigma quality objectives. In the most recent measurement period, there was not a single defect in several million parts produced by the company. JST managers attribute this to the implementation of the machine vision system. “Quality is by far the most important factor in our company’s success,” Hoselton concluded. “Machine vision has helped us make significant improvements by helping us detect variation faster so we can make corrections before defects occur.” Based on the success of this application, Hoselton set up another In-Sight vision system on a similar assembly machine that produces a different product. The main difference between the two products is that the gap in the housing is a different size. Hoselton started with the program that he developed for the first assembly machine. In only 15 minutes, he was able to tweak the original program for the new product.