White Paper on Automatic Inspection of Electronic Connectors

In the electronic connector manufacturing industry, the speed of punching connector pins is extremely high. Until recently, machine vision technology has been unable to match the high-speed requirements of punching quality inspection. Achieving 100% online inspection of connector (pin) punching quality has always been a dream but difficult problem for manufacturers.

The difficulty in mastering and using machine vision systems is the second reason that restricts the application of machine vision technology in the electronics industry. Fortunately, thanks to the development of camera and image processing technology, today's machine vision systems can capture images at a speed of 10,000 pieces per minute; at the same time, innovative human-computer interaction technology has made machine vision systems easier to master and use. All of these have greatly contributed to the successful application of machine vision inspection technology in the electronic connector manufacturing industry.

This article details the entire process of using the PPT vision system to perform 100% online inspection of the stamping quality of electronic connector pins, introduces the visual inspection tools and inspection methods used to detect typical quality defects, and the graphical programming language used to build the system.

1. Introduction

Machine vision technology has experienced nearly two decades of development in industrial practical applications, but it was not until the last five years that the technology really began to meet the quality inspection requirements of the electronic connector manufacturing industry. Among them, the most important obstacle is the extremely high manufacturing speed of electronic connectors, especially the stamping process of connector pins. Generally speaking, the stamping speed of pins is close to 2,000 pieces/minute-this rate is difficult for early machine vision systems to match. New machine vision systems continue to adopt the camera and image processing technologies that have developed rapidly in recent years, including various DSPs (digital signal processors) - chips designed specifically for high-speed analysis and calculation of various digital signals. The application of these hardware and software technologies enables today's machine vision systems to reach and exceed image detection rates of 10,000 pieces/minute.

2. Manufacturing process of electronic connectors

There are many types of electronic connectors, but the manufacturing process is basically the same, which can generally be divided into the following four stages:

· Stamping
· Plating
· Molding
· Assembly

2.1 Stamping

The manufacturing process of electronic connectors generally starts with stamping pins. Electronic connectors (pins) are stamped from thin metal strips using large high-speed stamping machines. One end of the large roll of metal strip is fed into the front end of the stamping machine, and the other end passes through the hydraulic workbench of the stamping machine and is wound into the take-up wheel, which pulls out the metal strip and rolls it up to stamp out the finished product.

2.2 Plating

After the connector pins are stamped, they should be sent to the plating section. During this phase, the connector's electrical contact surfaces are plated with various metal coatings. Similar problems as in the stamping phase, such as twisting, chipping, or deformation of the pins, can also occur during the process of feeding the stamped pins into the electroplating equipment. With the techniques described in this article, these quality defects are easily detected.

However, for most machine vision system suppliers, many quality defects that occur during the electroplating process are still "off limits" for inspection systems. Electronic connector manufacturers want inspection systems to be able to detect various inconsistent defects such as small scratches and pinholes on the electroplated surface of connector pins. Although these defects are easy to identify on other products (such as aluminum can bottoms or other relatively flat surfaces); due to the irregular and angled surface design of most electronic connectors, it is difficult for vision inspection systems to obtain images sufficient to identify these subtle defects.

Because some types of pins are plated with multiple layers of metal, manufacturers also want the inspection system to be able to distinguish between the various metal coatings to verify that they are in place and in the correct proportions. This is a very difficult task for vision systems using black and white cameras because the grayscale images of different metal coatings are virtually the same. Although the color vision system camera can successfully distinguish these different metal coatings, the problem of difficult lighting still exists due to the irregular angles and reflections of the coating surface.

2.3

The plastic box seat of the injection molded electronic connector is made during the injection molding stage. The usual process is to inject the molten plastic into the metal tire membrane and then quickly cool it to form. When the molten plastic fails to completely fill the tire membrane, the so-called "short shots" appear, which is a typical defect that needs to be detected during the injection molding stage. Other defects include filling or partial blockage of the sockets (these sockets must be kept clean and unobstructed so that they can be properly connected to the pins during the final assembly). Since the use of backlight can easily identify the box seat missing and the plug-in hole blockage, the machine vision system used for quality inspection after the injection molding is completed is relatively simple and easy.

2.4 The final stage of assembly

electronic connector manufacturing is the assembly of the finished product. There are two ways to connect the plated pins to the injection molded box seat: single plug-in or combined plug-in. Single plug-in means plugging one pin at a time; combined plug-in means plugging multiple pins into the box seat at the same time. Regardless of the connection method adopted, manufacturers require that all pins be inspected for missing and correct positioning during the assembly stage; another type of routine inspection task is related to the measurement of the spacing on the mating surface of the connector.

Like the stamping stage, the assembly of connectors also poses challenges to the automatic inspection system in terms of inspection speed. Although most assembly lines have a beat rate of one to two pieces per second, the vision system usually needs to complete multiple different inspection items for each connector that passes the camera. Therefore, the inspection speed is again It has become an important system performance indicator.

After assembly, the connector's overall dimensions are orders of magnitude larger than the dimensional tolerance allowed for a single pin. This also poses another problem for the visual inspection system. For example, some connector box seats are larger than one foot in size and have hundreds of pins. The detection accuracy of each pin position must be within a few thousandths of an inch. Obviously, it is impossible to complete the inspection of a one-foot-long connector on one image. The visual inspection system can only detect the quality of a limited number of pins in a smaller field of view at a time. There are two ways to complete the inspection of the entire connector: use multiple cameras (increasing system costs); or continuously trigger the camera when the connector passes in front of a lens, and the visual system will "stitch" the continuously captured single-frame images. The latter method is the detection method usually adopted by the PPT visual inspection system after the connector is assembled.

The detection of "true position" is another requirement of the connector assembly for the inspection system. This "true position" refers to the distance between the top of each pin and a specified design reference line. The visual inspection system must make this imaginary reference line on the inspection image to measure the "true position" of each pin vertex and determine whether it meets the quality standard. However, the reference point used to define this reference line is often invisible on the actual connector, or sometimes appears on another plane and cannot be seen at the same time in the same lens. In some cases, the plastic on the connector box must be ground off to determine the position of this reference line. There is indeed a related topic here - detectability design. Inspectability design

(Inspectablity)

Due to the continuous requirements of manufacturers to improve production efficiency and product quality and reduce production costs, new machine vision systems are becoming more and more widely used. As various visual systems become more and more popular, people are becoming more and more familiar with the characteristics of such inspection systems and have learned to consider the detectability of product quality when designing new products. For example, if a reference line is desired for detecting "actual position", the connector design should take into account the visibility of this reference line.

3. A Deeper Look at Vision Technology - Stamping Quality Inspection

As mentioned earlier, the production speed of electronic connector pins is usually calculated in thousands of pieces per minute. Until recently, inspection speed has been an insurmountable obstacle for automatic quality inspection systems. Early machine vision inspection systems could not provide such a high 100% quality inspection speed.

3.1 Typical Application Example of Stamping Quality Visual Inspection

This article will detail a typical application example of a PPT vision system in stamping quality inspection. Although similar application examples vary due to the changes in the pins of their respective inspection objects, most stamping quality visual inspection systems follow a basically consistent overall structure. As mentioned earlier, pins are stamped from thin metal strips. Based on different stamping processes, some pins are stamped and formed in one go by a single stamping machine, while others are stamped and formed in several times by multiple stamping machines. When the metal strip passes through the stamping machine, some pins are blocked and positioned incorrectly during the stamping process, resulting in stamping defects such as distortion or fragmentation.

Because stamping production is done at such high speeds, it is important to detect stamping defects in time to avoid scrapping the entire roll of metal strip. Random quality defects, such as metal chips (generated during the stamping process and attached to the metal strip), must also be detected. If carried all the way to the assembly line, these metal chips may cause equipment jams or unqualified assembly products, or even worse - customer complaints! In order to detect product defects as early as possible, quality inspection should be carried out immediately after the stamping process is completed.

PPT vision systems can be integrated into existing production lines in various ways. If there is an existing PLC (Programmable Logic Controller) on the production line, it can control the operation of the PPT vision system like any other sensor or mechanical device. These external PLCs are not necessary - the PPT vision system itself is fully capable of communicating with various external devices (trigger sensors, scrap rejection mechanisms, etc.). Some PPT customers have enabled their vision systems to stop the stamping machine operation as soon as a problem is detected so that the operator can troubleshoot the problem. Other similar solutions, such as controlling the spray gun to mark the defective products, can also be easily realized through the PPT vision system. [page]

Backlight is an extremely effective lighting method for detecting typical quality defects in the stamping process. Backlight can be achieved by arranging a beam of scattered light behind the target to be inspected - it can create a strong contrast between the target to be inspected and its bright background to obtain a clear inspection image with a silhouette effect. Such an ideal image is usually used to measure the dimensional characteristics of the target, such as length or width, and is also used to detect the contour of the target (shape detection).

3.2 Material conveying and component composition of the visual inspection system for stamped parts

This additional visual inspection system can be easily installed on the material take-up wheel system of the stamping machine. The stamped pins can be inspected immediately after they are sent out of the stamping machine and then rolled onto the take-up wheel with the metal tape. The whole system includes:
an adjustable open slide mechanism to guide the metal tape carrying the pins through the bottom of the camera,

a PPT-DS2 camera installed above the metal tape;
a strobe light and a diffuse reflection resin glass installed under the metal tape;
a set of photosensitive detectors to provide the visual inspection system with a trigger signal that the workpiece to be tested is in place.
A PPT Passport visual processing system

Figure 2 below shows the basic structure of the system.

(Figure 2: Equipment layout of a typical visual inspection system for stamping quality)
(Take-up reel, Camera, sensor,
Carrier strip with contacts, Stamping press,
Guide block and strobe light)

When the stamped pins are sent out of the stamping machine with the metal strip, they pass through the visual inspection system via the guide block and pass in front of the diffuse reflective resin glass. For the arrival of each pin, the photosensitive detector provides a trigger signal to the detection system, turns on the strobe light, captures the image, and starts the inspection. The purpose of using the strobe light is to avoid the metal strip from stopping in front of the camera every time. When integrating the above-mentioned visual inspection system into an existing stamping production line, only a few changes to the stamping section are required.

3.2.1 Partial Scanning

In order to achieve the maximum image resolution and simplify the trigger execution scheme, the image of each pin is usually captured and inspected separately, that is, the visual system only detects one pin at a time instead of multiple pins. When the system runs at an inspection speed of thousands of pins per minute, it becomes very important to use partial scanning technology and resettable cameras. The PPT-DS2 camera used by the PPT Vision System has a "Double-scan" mode, in which the camera can acquire a full-resolution (640x480) progressive scan image in 16.7 milliseconds - twice the speed of the standard RS-170 camera.

However, in many stamping quality inspection applications, this speed is still too slow, so the PPT Vision System uses a technique called "partial scanning" to further reduce the total image acquisition time. Cameras using partial scanning technology only scan a limited number of scan lines - the fewer scan lines, the faster the image acquisition speed. The number of scan lines is not fixed and can be changed according to the user's intention. For example, it takes 16.7 milliseconds to scan a 480-line image, but if the camera is set to scan only the first 120 lines, it only takes 4.18 milliseconds. Because the full horizontal resolution is still used, partial scanning is very suitable for targets whose dimensions are much larger in one direction than in another, such as pins.

Figure 3 below shows the principle of partial scanning technology.

Figure 3: PPT-DS2 camera scan modes: full-frame double-speed scan and partial scan. Unit: Pixels
(left - full-resolution double-speed scan mode: full (640x480) image acquired in 16.7 milliseconds);
(right - example of partial scan mode: partial (640x120) image acquired in 4.8 milliseconds, unscanned image areas remain blank)

Generally speaking, quality defects of stamped parts include deformation or distortion of pins, excess metal adhesion (metal debris) and dimensional errors. Some typical quality defects of stamped pins that can be identified by the vision system are shown in Figure 4.

(Twisted contact - twisted pin, Metal flash - adhered debris,
Crushed contact - contact extrusion deformation, Thin contact - pin is too thin)

3.2.2 Vision tools

Due to various mechanical and electronic interferences or hysteresis, the position of the part (pin) to be tested will change every time the image is captured by the vision system. Therefore, before performing part inspection, the system must first find the position of the target part within the field of view. The PPT vision system uses a vision inspection tool called "Origin Tool" to obtain the position and angle of the target part on the screen. This tool consists of three mutually perpendicular ROI (Region of Interest) "region of interest" coordinates on the image. These ROIs will tell the vision system where to find the target and the information found. The first and second ROIs use the contrast detection method (Contrast Sensing) to obtain the edge information of the target part and calculate the starting angle; together with the edge information obtained by the third ROI (vertical direction), the position and angle of the target part on the image plane are determined together.

Figure 5: Three ROI coordinates - the location of the detection pin in its backlit image
(Primary ROI, Secondary ROI, Perpendicular ROI)

The pin edge coordinates and starting angles are passed to some subsequent visual inspection tools and used to calibrate their ROI relative to the pin tip.

The "Template" tool provides a simple means to detect whether the pin is deformed or distorted and the consistency between the pin shapes. This tool is very suitable for detecting the basic shape of the part. Since the visual system mainly uses the "gold rule" (standard part) image as the self-teaching initialization method, the template tool is also very easy to establish. The user simply defines a feature area covering the entire pin image and tells the system to ignore all pixels outside the given gray level range. The system can "learn" and generate a dot matrix image template based on this. After that, only those pixels within this feature area and "activated" by the template are detected. If all "activated" pixels are within the user-defined gray range, the template tool detection passes. As shown in Figure 6, all pixels covered by the template (shown in single gray in the figure) should be black; if there is pin deformation or distortion, some of these pixels will become bright and fail to pass the template detection.

Figure 6: Template tool - detection of pin deformation defects and shape verification
(Template of "turned ON" pixels - Grid of ROIs - ROI grid: the vision system only detects those pixels that are confined to the feature area and "activated" by the template) [page]

"Complementary Template" This tool can be used to detect the area around the target object (pin). An example is shown in Figure 7.

Figure 7: Auxiliary template - inspecting the area around the pins and whether there is any adhering debris

If there is redundant metal or metal debris on the pin, black pixels will appear on the image and it will not pass the auxiliary template tool detection.

The external dimensions and errors are detected by tools such as "Gauges". Gauge tools are available in two types: pixel accuracy and sub-pixel accuracy. For common typical stamping defects, pixel-accurate gauges can achieve satisfactory detection results. For example, an image with a field of view of 13mm x 10mm can enable the system to achieve a detection resolution of +/- 0.0203mm (13mm/640 pixels = 0.0203 mm/pixel). Obviously, the smaller the image field of view, the higher the system resolution.

The Sub-pixel Line Gauge Tool can achieve a detection accuracy of 1/4 pixel (standard deviation). It can complete multiple detection items for the same target object, and can also perform the same detection item multiple times. That is, define an inspection item and repeat it in a selected image area. This function is particularly effective for tasks that detect multiple targets along the center of a series. The Sub-pixel Line Gauge Tool uses a connectivity analysis algorithm to detect image edges, which can achieve sub-pixel detection accuracy. When the system is initialized, the user can specify two edges on the "golden rule" (i.e., standard part) image and tell the system to measure the distance between the two edge lines. In the actual production inspection afterwards, the system will measure the distance between the corresponding edge lines of each part. If this distance is within a given range, the inspection passes; otherwise, it fails. So what dimensional errors can the visual system detect? This question will depend entirely on the actual condition of the image being inspected. Figure 8 shows the location of a set of ROIs of the Sub-pixel Line Gauge Tool, as well as several typical sizes that manufacturers hope to be detected.

Figure 8: Sub-pixel Line Gauge Tool - Detecting dimensional errors
(a set of ROI positions of the Sub-pixel Line Gauge Tool to measure
the width and longitudinal length of multiple cross-sections of the pin)

The Contrast Tool can be used to detect many of the quality defects discussed above. The tool uses the ROI to determine whether there is a contrast in the gray level within the area - contrast can confirm the presence of the target feature, and no contrast indicates that the target feature (such as metal debris) is not present.

Once the various visual inspection tools mentioned above are running, the system will send the inspection results (pass/fail signals) to other equipment on the production line through the built-in Opto22 output module. For example, in a typical stamping quality inspection example, if the vision system detects a defective part, the inspection failure signal will be immediately sent to the stamping machine controller to stop the stamping machine operation. At the same time, the operator will receive an alarm signal. Only after the cause of the quality defect is identified and resolved can the operator start the stamping machine to resume production.

The inspection results of the various visual inspection tools mentioned above will also be collected and displayed on the operator screen; or output to other software running on the host computer for further statistical analysis, such as spreadsheets or statistical process control (SPC) programs.

3.2.3 Visual Inspection System Programming

PPT All machine vision systems are equipped with PPT's visual program management software - VPM (Vision Program Manager). VPM uses a graphical programming language, allowing users to create various visual inspection programs extremely freely without having to master computer programming languages. VPM has two distinct operating modes: Edit Mode and Run Mode.

In Edit Mode, users first edit the inspection program to instruct the visual system what to do. In this process, users do not have to type in lines of instruction code, but only need to grab various icons and drag and drop them together to form a visual flow chart. Each icon here represents a basic machine function. Users can also connect these icons with lines of various colors to indicate the execution order and data flow route of the entire inspection process. Finally, set specific functional parameters for each icon.

After editing the inspection program, users can create their own personalized "Control Panels". These panels are for workshop operators to use on the production line. Using the click combination method used in the above-mentioned creation of inspection programs, users can create these panels in a very short time. The shape, position, logo and color of all components in a panel can be modified according to user preferences; multiple panels can also be created and linked together at the same time. The user can design the operation panel completely independently and add password protection to limit the ability of workshop operators to interfere with the inspection program.

VPM usually runs in execution mode. The above control panel will be displayed and activated at this time, and the workshop operator uses the control panel through the integrated touch screen. Through various buttons on the touch screen, the operator can start the visual inspection system, run the system offline, load various inspection files and all other functions set by the system designer.

The process of the stamping parts quality inspection system described above can be given in the graphical programming language of VPM, as shown in Figure 9. Camera Tool - The entire inspection process starts with the camera tool,
Origin Tool - The origin tool determines the position of the part in the field of view,
Template tools - Template tools,
Sub-pixel Line Gauge Tool
Parallel Output Tool - If any inspection tool fails to detect, the parallel output tool sends a "stop" signal to the stamping machine.

The camera tool specifies that the entire inspection process starts with the camera. The Origin Tool determines the position and angle of each pin in the camera field of view and provides this information to each subsequent inspection tool. Two sets of template tools detect whether the pins are deformed, twisted or have metal debris and the consistency of the contact shape. The sub-pixel line gauge detects whether the shape and size errors related to the pins are qualified. One gauge tool can measure several dimensional data. The

flow line in the figure specifies the execution order of all inspection tools, that is, only after the current tool passes the inspection, the next tool can be executed. If any tool fails the test, the Parallel Output tool will send a failure signal to the press controller according to the flow line regulations, and the press will stop running immediately. Although all inspection tools have their own counters for failure/success/execution, the inspection of each part will first trigger the origin tool, so the total number of inspected parts will be given by the execution counter of the origin tool.

3.2.4 Problems that may be encountered when installing a visual inspection system

Most of the time, there will be some problems when installing a visual inspection system. In the past, PPT encountered the same obstacle many times when installing a stamping part quality inspection system: when the material transport mechanism was cleaned, the camera and strobe light could be installed immediately in front of the take-up wheel. However, during the trial run, the system's inspection speed obviously could not keep up with the production rhythm - even though the speed of continuous stamping production at this time did not exceed the system's inspection speed limit. After analysis and research, it was found that this "overrun" problem was caused by the irregular speed of the take-up wheel. When the conveying metal belt is slack, the take-up wheel speeds up to tighten the conveying belt; once the slack is eliminated, the speed of the take-up wheel slows down again. Since the speed of the take-up wheel is always behind the conveyor belt, each time the take-up wheel slows down after tightening the conveyor belt, a large section of the conveyor belt is loosened again. As a result, the conveyor belt accelerates irregularly and suddenly when passing through the guide block of the visual inspection system, sometimes even exceeding the detection speed limit of the visual system. The solution to the "overspeed" problem is to carefully adjust the speed of the take-up wheel. As long as the irregular mutations in the movement of the conveyor metal belt can be eliminated, the visual inspection system can keep pace with the stamping production speed.

The lubricant remaining on the pin brings another problem to the visual inspection system. The metal belt needs sufficient lubrication when passing through the stamping process, but these lubricant drops will often adhere to the stamped pins. If they are mistaken for metal adhesion debris by the visual inspection system, it will cause misjudgment of the inspection. To solve this problem, a pair of air flow spray devices can be installed before the pin contacts reach the inspection system to remove the residual lubricant on the parts.

4. Summary

At present, machine vision systems provide excellent solutions for quality inspection of the four stages (stamping, electroplating, injection molding and assembly) in the electronic connector manufacturing industry. High-speed manufacturing processes, such as stamping, require extremely high detection speeds in the detection system, which has always been the main obstacle to the realization of fully automated quality inspection in the electronic connector manufacturing industry. Fortunately, thanks to the advancement of camera and image processing technology, vision systems today finally have enough capacity to achieve an online detection speed of 10,000 pieces/minute.

At the same time, the advancement of machine vision system application technology has made it extremely convenient to use. It is worth noting that when evaluating the feasibility of installing machine vision systems on production lines, major manufacturers of electronic connectors increasingly consider the ease of installation and maintenance of machine vision systems as a major factor.

In the past five years, many successfully installed machine vision systems have begun to appear on the market. These vision systems that have been installed and used in the electronic connector manufacturing industry have greatly reduced the scrap rate, prevented substandard products from reaching customers, and brought manufacturers a rich return on investment. In most cases, the investment in installing a vision system can be recovered within a few months.