Research On Technology For High Precision Defect Inspection Of FPC Connectors

With the rapid development of the electronics industry, PCB and FPCBelectronic products continue to increase, resulting in the gradual expansion ofdemand for components such as electronic connectors, so the quality of the productsis also a growing concern. At the same time, in order to meet the demand forelectronic products’ development, connectors are gradually to small size, narrowpitch, high-precision trend development, which put forward higher requirements forproduct quality testing. The machine vision algorithm is adopted to inspect badconnector functionality and portion of bad appearance in this paper, and to set up aset of hardware and software models of the inspection system. This system has a highdegree of automation, high accuracy, strong anti-interference ability, less manualintervention and simple operation. There are several innovative points in thefollowing.Firstly, the edge redetection algorithm based on threshold segmentation isproposed in this paper to inspect the pin terminal contacts using sub pixel accuracyalgorithm combined with run-length tree algorithm. It is verified that this improvedmethod is more accurate and the duplicate detection accuracy can reach to0.01mmcompared with the traditional geometric measurement method.Secondly, a kind of line fitting method based on the second iteration of theweighting function is proposed in this paper, using weighting function to calculatethe weight of every point to fit a straight line which can circumvent larger outliers.The experiment compares the results between this method and the traditional linefitting algorithm when coming up with the outliers. It is verified that the precisionand accuracy of the algorithm are better than the traditional one.Thirdly, traditional optical inspection system is mainly for functional defect testing,therefore, judgment on the part of the appearance defect of the connectors usingmathematical morphology is put forward, and it mainly includes fuzziness adhered tothe welding pin detection. The algorithm uses the mathematical morphology methods, and the processing steps are elaborated in detail. This method can accurately detectthe fuzziness adhered to the welding pin, and the detection accuracy can reach90%or above.Industrial field test results show that, it can be effective in FPC electronicconnectors’ all functional adverse and portion of appearance defects such asfuzziness adhered to the welding pin inspecting, locating and identifying.