Product Details Supplier Info More products

Cognex has revealed how its In-Sight vision sensors have helped New Jersey-based Compex Corporation manage its component counting and packaging processes more efficiently.

In today’s electronics industry, the reduced geometries of high-volume components, coupled with ever-increasing demand for faster customer turnaround times, has led to a growing need for automated packaging solutions.

While the accurate placement of, and accounting for, thousands of tiny components into packages may seem trivial, the time it takes component manufacturers to perform these packaging tasks without automation can be a bottleneck to business growth.

This is particularly true with companies that build components based on customer specifications.

New Jersey-based Compex Corporation is a case in point.

The company manufactures single-layer components that are used in high-frequency RF, microwave, telecommunications and fibre optics products.

To increase output and accelerate delivery time, Compex recently implemented new automated inspection systems based on compact machine vision sensor technology.

David Gordon, general manager of Compex, said: ‘Historically, packaging has been one of the major delays for us in getting product out the door.

‘A lot of our time was spent manually inspecting components to ensure they were oriented properly before being placed into packages, and later counting components in packages to make sure customers were getting exactly what they had ordered.

‘For a rush order of 25,000 pieces, for example, we would need multiple operators in the packaging line for two or three shifts just to ship within a few days’ notice.’ Gordon added: ‘Our goal was to be able to start shipping within one week for any order we received.’ The new automated inspection systems have enabled Compex to not only reach its one-week turnaround goal, but to also increase production yield and redeploy nearly 80 per cent of the resources previously spent on packaging to other areas of the operation.

Compex is no stranger to machine-vision technology.

In fact, the company had implemented a proprietary vision inspection system in 2000 to ensure components were oriented the correct way before being robotically loaded into packaging.

The system, however, had a number of limitations that prevented it from speeding the packaging process up.

For example, the vision software tools were not sophisticated enough to handle wide variations in part appearance.

Different types of substrate materials, including quartz and ceramic, would confuse the vision system and cause failures.

High reflectivity of the components – most of which are gold-plated – was also problematic.

Additionally, various marks and scratches on the surfaces of production feed track, caused by normal wear and tear, would inevitably show up in inspection scenes.

Unfortunately, the vision system had trouble ignoring this type of ‘background noise’ and additional failures resulted.

Gordon said: ‘The vision system we had been using was basically just a pixel-counting system, so even if a component was mis-oriented or off-centre, the system would accept the part if the pixels added up.

‘We had to constantly tweak the light sources for this system, and ultimately the systems limitations demanded an upgrade.’ Acknowledging the limited reliability and throughput of the current vision equipment, Compex shifted to a semi-automated process to be used with new pick-and-place equipment.

The vision system would capture and display an image on a monitor, and an operator would manually make a go/no-go decision based on what he saw on the screen.

If a component was not oriented properly, the operator would physically remove and recycle the component back into a vibratory feed bowl.

If orientation was correct, the operator would hit a button on a controller to activate the robot – a multi-axis Cartesian model – which would pick the component off the feed line and place it into the packaging.

Gordon said: ‘Ironically, bringing humans into the inspection process with new pick-and-place equipment was a high-speed alternative to relying on the vision system itself.

‘But we believed full automation could take us to the next level.’ The company was also spending considerable resources on another stage of the packaging process: counting.

Thousands of components placed onto film carriers needed to be manually counted before the packages were shipped.

The process, according to Gordon, was slow, tedious, and error-prone.

He said: ‘One operator could spend 20 minutes or more counting a single package, and there were often inaccuracies.

‘Errors were unacceptable, since some of our customers would report back to us that two components were missing out of a shipment of 80,000.

‘The process was a nightmare for us, and just one more thing slowing us down.’ With far too much labour tied up in manual inspection tasks, Compex began its search for new automation systems that would streamline the component-packaging process and reduce the need for operator assistance.

In early 2005, the company learned about a local machine builder/system integration firm that specialised in vision-based inspection solutions.

The company, Serview, appeared to have the right combination of technological know-how and industry experience required to improve Compex’s packaging process.

According to Brian LeBlanc, engineering manager of Serview, the objectives of the project were clear at the outset.

He said: ‘Compex wanted to improve output and reduce turnaround time without changing the number of personnel.

‘Instead of having multiple operators working in the packaging area, they wanted to have a single person oversee a number of automated systems and use the remaining manpower elsewhere.’ Gordon and a few associates flew to Boston in 2004 to attend the Vision Show East (sponsored by the Automated Imaging Association).

During the show, they connected with Cognex, which was showcasing its line of In-Sight vision sensors.

Although Gordon and his team had been impressed by the Cognex products, they ultimately deferred to Serview on vision-system selection based on the firm’s experience.

As it turned out, Serview was a Cognex Certified Vision Integrator and designed a majority of its systems around Cognex technology.

While many of the systems Serview developed over the years incorporate PC-based vision systems, the primary inspections Compex needed to accomplish – component orientation verification and counting – were both single-camera applications.

For this reason, Serview integrators selected In-Sight vision sensors as the vision platform of choice.

Specifically, the team chose an In-Sight 1000 for the orientation system, which is a low-cost, general-purpose vision sensor, and an In-Sight 5403 for the counting system, which is a high-resolution vision sensor featuring 1600 x 1200 pixel CCD sensing.

Both sensors incorporate all image-processing hardware, software, and communications in a compact, self-contained package.

The integrated design reduces the time and expense associated with integrating vision into the production line.

Serview began work on developing the new systems.

The main components of the orientation system designed by Serview include the In-Sight 1000 vision sensor, Compex’s existing Cartesian robot, a vibratory feed bowl and track, an air blow-off for part rejection, and an Allen-Bradley Micrologix PLC.

A Cognex I/O module was used to link the In-Sight sensor with the PLC, and a Dell PC is used by the operators to monitor the inspection during runs and perform product changeover, and run the In-Sight Explorer vision software application, as well as a custom graphical user interface.

One of the main challenges in designing the system was configuring lighting and optics so the full range of Compex components could be inspected without making any set-up changes.

LeBlanc said: ‘We had to get the lighting and lensing repeatable across all components, which took a fair amount of development time to tune the inspection software to match the full range of part types and materials, along with normal part variations that were deemed acceptable.’ Since changeovers are frequent, it was desired that a single lighting setup would work across all part types.

This goal was accomplished and product changeovers are done by loading product configurations with no optics or lighting adjustments.

The Serview team experimented with various configurations on the range of part types and materials and ultimately settled on a combination of Xenon strobe and fibre optic lighting.

This combination provided white light at a short pulse width, along with tightly synchronised image acquisition and strobing via the Cognex I/O module, which provided proper image formation for the various components types.

The vision application, which involves a two-step inspection routine, was developed using In-Sight Explorer software.

The software includes a suite of development, management, and monitoring tools to make the entire vision application development cycle easy and fast.

The vision sensor first verifies the orientation of the component, and then measures critical dimensions.

The dimensional checks include measuring the outer dimension of the components and the dimensions of the metal coating.

Steve LeBlanc, director of operations at Serview, said: ‘Even though Compex uses accurate dicing saws, tolerances are still an issue.

‘Components as small as 0.015in2 are viewed under high magnification using microscope lenses, and we make dimensional measurements as part of the inspection process.

‘So, by measuring the parts during the orientation stage, we were able to give Compex another safeguard.

‘Any defective parts that are chipped or are out of dimensional tolerance that were missed in previous inspection steps are rejected by the system.

‘This gives an extra level of quality inspection that finds more defects than we originally expected to see.’ Both the orientation and dimensional inspections were set up using various edge-detection and histogram vision software tools, which are part of the standard In-Sight 1000 tool library.

A customer graphical user interface makes it simple for operators to select different inspection ‘jobs’ during part changeover.

An operator can take the system offline, select the new job, and put the system back into run mode within 10 seconds.

Serview engineers also designed a built-in calibration program into the In-Sight application.

This allows operators to easily reset the camera lens magnification within five seconds if the lens goes out of adjustment.

While the system is completely automated, Compex felt it was still important to have the option of using operators to inspect the components if, for example, computer systems shut down, or new component parameters had not yet been programmed into the vision application.

For that reason, Serview designed a manual-mode feature into the system.

Manual mode works in a similar fashion to its previous operator-assisted vision system, whereby an inspection image appears on the screen and the operator makes the pass/fail decision.

The orientation system is located at the end of a linear feed track, onto which components are fed from a vibratory feed bowl.

Once a component moves under the In-Sight 1000 sensor, which is mounted approximately 6in above the track, an image of the component is captured and sent to the In-Sight’s on-board processor where the vision tools analyse the image.

Operators can view the inspection in real-time as it occurs.

If the orientation is correct, the PLC signals the robot to grip and place the component onto a film ring or waffle pack.

The entire inspection and placement time, from image acquisition of a good chip to robot placement, is approximately 300 milliseconds.

The Insight vision system is faster than the material handling, so the process is limited by material handling speeds.

If there is a problem, such as the component being upside-down or stacked on top of another, the vision sensor fails the part, a red LED appears on the inspection monitor, and the part is automatically recycled back into the feed bowl.

Only good parts are accepted and placed by the system.

Further downstream, packages containing thousands of components travel through Compex’s automated counting system just before they are shipped.

This system, according to Gordon, was a necessary tool for a number of reasons.

He said: ‘Our guys were trying to do a manual count of thousands of pieces, which could take one person 20 minutes.

‘In addition to being time-consuming, it was very difficult to get 100 per cent accuracy, which our customers demand.’ Unlike the orientation system, the counting system is a tabletop, standalone system that requires no PLC.

A single-camera application as well, Serview again chose to incorporate an In-Sight vision sensor for the job.

However, this time the engineers selected the In-Sight 5403, a high-resolution (1600 x 1200) model.

LeBlanc said: ‘One of the key design considerations with this system was building in the flexibility to automatically count smaller and smaller components in the future.

‘The high-resolution In-Sight was a good solution in that it would give the company this leeway.’ In addition to the In-Sight sensor, the system consists of a combination of LED backlighting, which works well to image the components, and a Dell PC, which runs a custom-designed Visual Basic program.

The user interface for this application was developed using Visual Basic.

LeBlanc said: ‘The application we wrote gives the user just a couple of selections on the user interface to simplify the set up and operation of the system.

‘For example, the operator can just pick a given job from a drop-down list and all various set-up parameters for that job, such as brightness settings, change dynamically.

‘The application user interface is simple enough that most operators feel comfortable using the system right away.

‘The system is flexible, and we have already added new products as Compex adds them to the mix.’ The In-Sight vision tools used to perform the counting operation include a combination of blob and greyscale morphology pre-processing tools.

According to LeBlanc, getting the blob tool to run effectively was a challenge due to the fact that the components are so small and tightly laid-out on the package with very little separation between components.

For that reason, images are first pre-processed with the morphology tool, which provides the necessary feature separation for blob to work reliably.

Once a part type has been selected from the Visual Basic user interface, operators select ‘live mode’, which allows them to manually align the package with a template overlay.

The operation is viewed on the PC monitor.

Once the manual alignment is performed, the operator then selects ‘Count’, which triggers the In-Sight sensor to acquire the image and perform the count.

As the sensor counts, it also verifies that the components are the correct size and that components are not touching.

It typically takes less than 20 seconds to perform the whole sequence of alignment and counting.

The system reports the number of components in the package and indicates if there are any problems.

The image of the package is saved to disk and count data is saved in a SQL database, giving Compex a permanent record of every package it ships.

LeBlanc said: ‘So far, all reports are that the system is 100 per cent accurate, and we found up-front testing really paid off as we have not had to make any changes on this system after it was delivered.

‘We’re in the process of working to integrate the vision-count data so it is printed onto a label with barcode and serial number of the package.’ Through careful system design and implementation, Serview has helped Compex make improvements to its component packaging process without changing the number of personnel at the West Berlin facility.

The systems have eliminated a bottleneck in Compex’s ability to ship quickly, and labour savings in the packaging area alone are approximately 80 per cent.

Compex and Serview are currently working together to identify additional areas of the manufacturing process that could benefit from automated inspection, including component surface defect inspection.

View full profile