Machine vision: Eliminating mold damage
When an injection molding operation is completed, pins on the movable side of the mold push out the plastic part. Occasionally, however, a part sticks in the mold, and often an operator is responsible for checking the mold after each cycle to ensure the part was removed. Sometimes, an operator doesn’t notice that the part had not been ejected and, at TNT Plastic Molding in Anaheim, CA, a typical result was $14,000 in damage to the mold. TNT Plastic Molding has overcome this problem by using a vision-based mold monitoring system developed by A.S.K. Technologies of Yorba Linda, CA.
TNT Plastic Molding’s 55,000 square foot facility is equipped with 28 molding machines ranging from 55-ton to 610-ton capacity. Its mold-making capabilities range from single-cavity prototypes, multi-cavity, and hot runner, to three-plate and automatic unscrewing production molds. (TNT has ISO 9001 and AS 9100 certification, as well as a mold-making facility in China.)
Injection molding machines can run in either semi-automatic or automatic mode. In the past it was necessary to run a 300-ton JSW press in semi-automatic mode because of the danger of damage to the mold if a part was not ejected. The operator had to open the safety guarding door around the mold after every cycle to check that the part had been ejected into a drop chute leading to a conveyor. During a typical eight-hour shift, the operator inspected the mold about 640 times.
Not only was there great potential for human error due to fatigue, variable inspection times caused dimensional variations in the parts. Worse yet, it was not uncommon for the operator to accidentally cycle the press even though the part had not been ejected, which would result in damage. It would take up to three weeks to repair the costly mold, resulting in substantial lost revenues for the company.
“We were interested in taking human error out of the equation in order to improve quality and reduce costs,” said Murray Anderson, director of sales and marketing for TNT Molding. “We had noticed that the performance of vision systems designed for injection molding had improved at the same time their cost was being reduced.
“We selected A.S.K. Technologies’ mold monitoring system because their system is very easy to program for a wide range of mold monitoring applications including insert molding, short shots, part presence or absence, runner presence or absence and before-and-after shot inspections. The flexibility of the system also enables many other possible applications. We were also very impressed with the cost of the mold monitoring system which, at about $10,000, is less than the cost of a single accident caused by failure of the part to eject.”
The system uses the Cognex In-Sight vision system. “Its small size and flexible mounting capability make it easy to fit into nearly any press application,” says Askin. “The system is very easy to program with a library of 22 vision tools.”
The vision system was trained in minutes by taking images of the mold in the proper condition without a part. In operation, it checks the mold for the presence of the part on every cycle, eliminating the risk of damage.
A.S.K. Technologies used the In-Sight Software Development Kit to develop a user interface for the new mold monitoring system with a simple point and click setup. The user can specify a tolerance that determines the vision system’s rejection standards. Up to 16 configurable windows can be activated or deactivated by the user.
“We had worked with another vision company and made suggestions as to how their product could be improved, but no changes were ever made,” Askin says. “When we started working with Cognex we discovered they were much more responsive. Cognex made changes to their In-Sight system that made it better for molding applications. In particular, they increased the flexibility of their interface, making it easier for us to develop our mold monitoring system.”
It took just 10 minutes to develop the program for the JSW press, says Askin. After the setup was complete, the job was saved onto the camera and PC for backup and future reference. The operator simply turns the camera on and waits for it to boot, then starts the PC to power up the mold monitoring system. The camera automatically loads the program specified by the user and puts it online. The pattern-find function is able to locate the area that needs inspection from anywhere in the field of view, so the program can consistently detect the presence or absence of the part despite the fact that the two sides of the mold might be in different positions.
“The minute a tool closes on a part or insert, there’s a strong probability that damage is going to occur,” Askin says. “The mold monitoring system inspects the mold before it closes for part presence, part absence, insert loading, etc. It stops the press from even beginning to close, and prevents a cycle from taking place until the mold is cleared of any problems. This saves a considerable amount of repair cost and downtime.”
TNT Plastics ran a series of tests that confirmed that the mold monitoring system is able to consistently identify the absence of a part in the mold and shut down the press. Anderson says, “This system has eliminated damage to the mold and associated lost production. It also made it possible to reconfigure the machine for full automatic operation. Quality has also been improved because the cycle time is consistent which means the hold time, cool time and shrink time are now the same for every cycle. By eliminating variation from the operation—such as temperature fluctuations—it is now possible to hold the part to tighter tolerances.”
Eliminating manual checks frees up time for other tasks. Previously, the press operator could only handle a single machine, but now he or she has time to handle two, substantially reducing labor costs. Even running two machines, the operator has more time than before for inspecting parts, which has helped to improve quality.
Edited by Renee Robbins, senior editor, Control Engineering. Reach her at firstname.lastname@example.org .