Case Studies
Cognex Corporation designs, develops, manufactures, and markets a range of products that incorporate sophisticated machine vision technology that gives them the ability to “see.” Cognex products include barcode readers, machine vision sensors, and machine vision systems that are used in factories, warehouses, and distribution centers around the world to guide, gauge, inspect, identify, and assure the quality of items during the manufacturing and distribution process. Cognex is the world’s leader in the machine vision industry, having shipped more than 1 million vision-based products, representing over $4 billion in cumulative revenue, since the company’s founding in 1981. Headquartered in Natick, Massachusetts, USA, Cognex has regional offices and distributors located throughout the Americas, Europe, and Asia.
Content Filed Under:
Industry:
N/A
Application:
N/A
Vision System Performs 100% Inspection in High Volume Valve Cover Production
POSTED 12/19/2014
Dim
sional integrity on automotive engine valve covers has become much more critical than in the past. Today, engineers increasingly design additional functionality into valve covers, including mounting points for engine timing components. That’s why an automobile original equipment manufacturer (OEM) recently required 100% inspection from its valve cover supplier. Formerly, the component manufacturer performed precision inspection on a sample of parts with a coordinate measuring machine (CMM). The challenge, however, is that CMMs are generally unsuitable for performing 100% inspection in high-volume production environments. The machines are typically too slow and must take place off-line in a quality control lab because CMM equipment is generally not built to survive in a production environment. To deliver 100% CMM-like inspection for high-volume automotive component manufacturing the component supplier tapped HAHN Automation to design and build an inspection system that relies on machine vision. The system identifies the true position and determines the size of seven holes, in addition to checking the presence and position of inserts, and inspecting the valve covers for flatness, all within the 35-second cycle time of the machine. Compared to the only conceivable 100% inspection alternative, automated touch probe inspection, the vision system costs half as much, takes up one-fourth the space and can be easily reprogrammed to handle design changes, new models or even entirely different parts.
Need for 100% inspection
The increasing need for 100% inspection of automotive components and other parts produced in large quantities at high production rates presents considerable difficulties for component manufacturers. CMMs are typically used to perform precision dimensional inspections in the quality control lab by touching a probe to points on the part under the control of a program or the machine operator. These machines are designed for low volume inspections so they are much too slow to keep up with a production line and also not designed to survive in a factory environment. The relatively small number of applications to date that provide 100% dimensional inspection in a production environment have typically used dedicated touch probe technology incorporated into an index machine. The limitation of this approach is that there is usually only room for a few inspection operations in a typical machine station. Multiple stations are required for components that need many dimensions inspected, which drives up machine cost and footprint. And because this approach is based on hard automation, any changes to the part require modifications to the machine.
The manufacturer of these V6 valve covers needed to find a better approach and worked with HAHN Automation to address this challenge with machine vision. Machine vision is commonly used to perform inspection operations that are less complex and have lower accuracy requirements than are involved in this application. But in recent years, machine vision systems have been driven by the advance of Moore’s Law to provide continually greater levels of resolution along with improvements in speed and
programmability. HAHN took advantage of these improvements and also utilized laser sensors and linear variable displacement transducers (LVDTs) to develop a flexible vision system that provides 100% high-precision dimensional inspection of the valve covers in a single machine station. Since the inspection is performed by vision rather than hard automation it can easily be changed to accommodate design changes or entirely different parts merely by writing new software.
HAHN Automation worked with vision integrator Crescent Electric Supply Company to select Cognex In-Sight Micro vision systems because they are completely contained within a 30 mm x 30 mm x 60 mm size sealed enclosure, making them ideal for mounting in tight spaces on robots and in hard-to-reach machinery. These vision systems include the Cognex vision tool library, which handles nearly any inspection operation without writing code, and the EasyBuilder graphical programming environment. The In-Sight 1403 model was selected because of its 1600 x 1200 pixel resolution, which provides the accuracy needed for precision inspection, and its high acquisition rate. In this case, the vision system needs to inspect features at varying distances so it was outfitted with a telecentric lens. The telecentric lens removes the perspective or parallax error that makes closer objects appear to be larger than objects farther from the lens so it enables consistent measurements of objects at varying distances from the lens.
Inspecting seven holes in one station
Performing all of the required inspections in a single station with machine vision required a considerable amount of expertise. A vision system and a laser range finder were mounted on a six-axis Kuka KR 5 sixx R650 robot and another vision system is mounted to a slide that travels on the Y axis. During the 35-second cycle time of the machine, the robot travels to and inspects three different holes and the linear slide travels to and inspects four different holes. The inspection station has been programmed to run a family of 10 valve cover models and the program can easily be changed to accommodate design changes or new models. Each part has a slightly different configuration. Depending on the part number, the holes on each part accommodate one or more positive crankcase ventilation (PCV) valves, oil fill caps, linear servos, spark plugs and camshaft position sensors.
The inspections are based on the geometric design and tolerancing of the print. The top surface of the fixture serves as the “a” datum while two features on the part are the “b” and “c” datums. All of the holes are inspected with the vision system and some are also inspected with the laser range finder to determine the z axis position of the hole. When a part is indexed into the inspection station, the robot moves into position above the “a” datum on the fixture and the laser sensor takes a reference measurement.
Then the robot moves the camera to the first hole to be inspected. A backlight below the fixture turns on and is shaded with cross-hairs that provides a known 0,0 reference of where the hole should be. The vision system with a telecentric lens acquires an image of the hole and a series of FindLine, PointToPoint, PointToLine and LineToLine vision tools are used to construct two lines through the center of the cross-hairs and detect the intersection of these two lines. This establishes the X,Y center of the cross-hairs. Then eight FindLine tools are used to detect the edges of the backlight around the circumference of the holes. The CircleFromNPoints tool is used to establish the circle representing the ID of the hole that is being measured. The diameter of the hole is calculated using the radius from the circle tool. A PointToPoint tool is used to find the X and Y offset of the center of the circle compared to the center of the cross-hairs.
Some of the holes have brass threaded inserts spaced at 120 degrees around the hole that act as the mounting points for the actuators. The vision system inspects for the presence of these inserts and also measures their profile to be sure they are all in the same plane. An additional inspection operation is performed by linear LVDTs on the fixture to check the flatness of the part. The inspection system also includes a CO2 laser marking device that engraves the part with a serial number.
100% inspection of 60,000 parts per year
HAHN Automation has built six of these assembly stations, three for left hand parts and three for right hand parts, which provide 100% inspection of a total of 60,000 parts per year. All inspections are performed in a three-station cell. The first station installs the inserts, the second station installs the datum pins and the third station is for machine vision inspection. The robots, vision systems and sensors are controlled by a programmable logic controller. The measurements captured by the inspection system are stored against the part’s serial number in a database residing on a local PC. The database is backed up every day to a plant server.
The only other apparent way to perform 100% inspection of these seven holes would have been to use an inspection system with dedicated touch probes for each of the holes. The same inspection operations would have required a four station rotary index machine that would have cost about twice as much and had four times the footprint of the single station that was used in this application. The machine vision inspection method used here is also much more flexible than a touch probe system. The current system runs 10 different part numbers without requiring any hardware changes while the touch probe system would have required hardware changes to run all of these parts. The machine vision approach also has the advantage that in case the company ever stops making valve covers the vision systems, robots and other hardware can be redeployed to other inspection operations.
sional integrity on automotive engine valve covers has become much more critical than in the past. Today, engineers increasingly design additional functionality into valve covers, including mounting points for engine timing components. That’s why an automobile original equipment manufacturer (OEM) recently required 100% inspection from its valve cover supplier. Formerly, the component manufacturer performed precision inspection on a sample of parts with a coordinate measuring machine (CMM). The challenge, however, is that CMMs are generally unsuitable for performing 100% inspection in high-volume production environments. The machines are typically too slow and must take place off-line in a quality control lab because CMM equipment is generally not built to survive in a production environment. To deliver 100% CMM-like inspection for high-volume automotive component manufacturing the component supplier tapped HAHN Automation to design and build an inspection system that relies on machine vision. The system identifies the true position and determines the size of seven holes, in addition to checking the presence and position of inserts, and inspecting the valve covers for flatness, all within the 35-second cycle time of the machine. Compared to the only conceivable 100% inspection alternative, automated touch probe inspection, the vision system costs half as much, takes up one-fourth the space and can be easily reprogrammed to handle design changes, new models or even entirely different parts.Need for 100% inspectionThe increasing need for 100% inspection of automotive components and other parts produced in large quantities at high production rates presents considerable difficulties for component manufacturers. CMMs are typically used to perform precision dimensional inspections in the quality control lab by touching a probe to points on the part under the control of a program or the machine operator. These machines are designed for low volume inspections so they are much too slow to keep up with a production line and also not designed to survive in a factory environment. The relatively small number of applications to date that provide 100% dimensional inspection in a production environment have typically used dedicated touch probe technology incorporated into an index machine. The limitation of this approach is that there is usually only room for a few inspection operations in a typical machine station. Multiple stations are required for components that need many dimensions inspected, which drives up machine cost and footprint. And because this approach is based on hard automation, any changes to the part require modifications to the machine.
The manufacturer of these V6 valve covers needed to find a better approach and worked with HAHN Automation to address this challenge with machine vision. Machine vision is commonly used to perform inspection operations that are less complex and have lower accuracy requirements than are involved in this application. But in recent years, machine vision systems have been driven by the advance of Moore’s Law to provide continually greater levels of resolution along with improvements in speed and
programmability. HAHN took advantage of these improvements and also utilized laser sensors and linear variable displacement transducers (LVDTs) to develop a flexible vision system that provides 100% high-precision dimensional inspection of the valve covers in a single machine station. Since the inspection is performed by vision rather than hard automation it can easily be changed to accommodate design changes or entirely different parts merely by writing new software.HAHN Automation worked with vision integrator Crescent Electric Supply Company to select Cognex In-Sight Micro vision systems because they are completely contained within a 30 mm x 30 mm x 60 mm size sealed enclosure, making them ideal for mounting in tight spaces on robots and in hard-to-reach machinery. These vision systems include the Cognex vision tool library, which handles nearly any inspection operation without writing code, and the EasyBuilder graphical programming environment. The In-Sight 1403 model was selected because of its 1600 x 1200 pixel resolution, which provides the accuracy needed for precision inspection, and its high acquisition rate. In this case, the vision system needs to inspect features at varying distances so it was outfitted with a telecentric lens. The telecentric lens removes the perspective or parallax error that makes closer objects appear to be larger than objects farther from the lens so it enables consistent measurements of objects at varying distances from the lens.
Inspecting seven holes in one stationPerforming all of the required inspections in a single station with machine vision required a considerable amount of expertise. A vision system and a laser range finder were mounted on a six-axis Kuka KR 5 sixx R650 robot and another vision system is mounted to a slide that travels on the Y axis. During the 35-second cycle time of the machine, the robot travels to and inspects three different holes and the linear slide travels to and inspects four different holes. The inspection station has been programmed to run a family of 10 valve cover models and the program can easily be changed to accommodate design changes or new models. Each part has a slightly different configuration. Depending on the part number, the holes on each part accommodate one or more positive crankcase ventilation (PCV) valves, oil fill caps, linear servos, spark plugs and camshaft position sensors.
The inspections are based on the geometric design and tolerancing of the print. The top surface of the fixture serves as the “a” datum while two features on the part are the “b” and “c” datums. All of the holes are inspected with the vision system and some are also inspected with the laser range finder to determine the z axis position of the hole. When a part is indexed into the inspection station, the robot moves into position above the “a” datum on the fixture and the laser sensor takes a reference measurement.
Then the robot moves the camera to the first hole to be inspected. A backlight below the fixture turns on and is shaded with cross-hairs that provides a known 0,0 reference of where the hole should be. The vision system with a telecentric lens acquires an image of the hole and a series of FindLine, PointToPoint, PointToLine and LineToLine vision tools are used to construct two lines through the center of the cross-hairs and detect the intersection of these two lines. This establishes the X,Y center of the cross-hairs. Then eight FindLine tools are used to detect the edges of the backlight around the circumference of the holes. The CircleFromNPoints tool is used to establish the circle representing the ID of the hole that is being measured. The diameter of the hole is calculated using the radius from the circle tool. A PointToPoint tool is used to find the X and Y offset of the center of the circle compared to the center of the cross-hairs.Some of the holes have brass threaded inserts spaced at 120 degrees around the hole that act as the mounting points for the actuators. The vision system inspects for the presence of these inserts and also measures their profile to be sure they are all in the same plane. An additional inspection operation is performed by linear LVDTs on the fixture to check the flatness of the part. The inspection system also includes a CO2 laser marking device that engraves the part with a serial number.
100% inspection of 60,000 parts per yearHAHN Automation has built six of these assembly stations, three for left hand parts and three for right hand parts, which provide 100% inspection of a total of 60,000 parts per year. All inspections are performed in a three-station cell. The first station installs the inserts, the second station installs the datum pins and the third station is for machine vision inspection. The robots, vision systems and sensors are controlled by a programmable logic controller. The measurements captured by the inspection system are stored against the part’s serial number in a database residing on a local PC. The database is backed up every day to a plant server.
The only other apparent way to perform 100% inspection of these seven holes would have been to use an inspection system with dedicated touch probes for each of the holes. The same inspection operations would have required a four station rotary index machine that would have cost about twice as much and had four times the footprint of the single station that was used in this application. The machine vision inspection method used here is also much more flexible than a touch probe system. The current system runs 10 different part numbers without requiring any hardware changes while the touch probe system would have required hardware changes to run all of these parts. The machine vision approach also has the advantage that in case the company ever stops making valve covers the vision systems, robots and other hardware can be redeployed to other inspection operations.
Vision in Life Sciences
This content is part of the Vision in Life Sciences curated collection. To learn more about Vision in Life Sciences, click here.