« Back To Vision & Imaging Industry Insights
AIA Logo

Member Since 1984


AIA - Advancing Vision + Imaging has transformed into the Association for Advancing Automation, the leading global automation trade association of the vision + imaging, robotics, motion control, and industrial AI industries.

Content Filed Under:


Visual Inspection & Testing Visual Inspection & Testing

Injection Molding Inspection: They’re Everywhere; But it ain’t As Easy as it Looks

POSTED 02/18/2013  | By: Winn Hardin, Contributing Editor

Machine vision essentially has three jobs when it comes to inspecting injection molding production lines: Watch the mold to make sure it’s clear and that inserts, connectors and other parts are in their proper place; inspect the plastic part for quality; and direct robots, material-handling and deflashing equipment to transfer or finish the molded part.

And, much to the chagrin of both customers and machine vision integrators, that’s where commonality ends — and customization begins — when it comes to machine vision solutions for plastic injection molding.

The trouble with plastic mold inspection

Courtesy of Avalon Vision Solutions“Inspecting plastic injection molds and parts is a pretty common application for machine vision, but none of them are easy,” says David Dechow, President of Aptura Machine Vision Solutions (Williamston, Michigan). “It’s very hard to put a camera in a position where it can confidently see all the important areas of the mold. Sometimes, it’s simply impossible to get a sufficient field of view with the right illumination to do the job.”

Machine vision almost always involves differentiation. Injection molds typically have smooth, highly reflective surfaces that resemble mirrors. And often, the mold is the same color as the plastic polymer, or can appear that way due to a lack of contrast caused by difficulties lighting the inside of the mold. And like all manufacturing industries, plastic injection molding lines are more profitable the shorter the cycle time — so long as they produce quality parts. In reality, this means the molding machines are built with an eye toward using as little real estate as possible, opening just enough to eject the part, and with little consequence for machine vision systems requiring a clear line of sight and adequate illumination.

While there are a few machine vision companies that offer standard products for tending to injection mold machines, such as Avalon Vision Solutions’ MoldWatcher, Aptura’s Dechow says he’s not aware of any injection molding machine-makers that offer machine vision as an option. “The machine-maker may offer some type of sensors to make sure a mold is clear before moving on to the next cycle, but as far as machine vision solutions go, mold inspection is usually too difficult to make into a standard product,” Dechow says.

“But on the flip side, the molds can be extremely expensive. While plastic isn’t [terribly expensive], a plastic part can be large or impact downstream manufacturing operations that do have a significant financial cost,” Dechow adds. “If the part isn’t right or the process goes astray and produces hundreds of defective parts before the problem is identified, manufacturers in automotive and medical device industries, for example, have a problem.”

“Everybody talks about quality, but we typically see investments in automatic inspection when it saves money or when it is a necessity.” says Fred Turek, chief operations officer for FSI Technologies (Lombard, Illinois.).  Mold clearance verification is one area which saves money by protecting the mold.  .  It’s always a challenge to illuminate the part and mold in a way that clearly differentiates one from the other. “Sometimes you can use color and you’re always working in the tightest of space constraints.” More recently, we’ve moved to using non-scanning 3D vision.

Color and 3D

Both color and 3D machine vision add cost to the solution. However, both are well suited to mold tending and molded part inspection.

“Because you have tight quarters with mold inspection, you can’t get light exactly where you want it so when using color you have to depend a lot on software engines that have the ability to switch to HSI [hue saturation intensity] color space that are capable of ignoring artificial color variations in the image that were induced by the inevitable lighting issues”. explains FSI’s Turek.

“When color doesn’t work, you can use 3D vision and compare a profile of an empty mold against what your “camera” sees, but the vision sensor has to fit inside the machine and that usually means you can’t use scanning 3D vision,” Turek says. “Our GS sensor-based systems have been a big step forward for us because it uses a projected grid of light and doesn’t require you to scan the object.”

When it comes to inspecting the produced part, unless the application only requires checking the perimeter of the molded part for flashing, material-handling, robotics or both are often required. “We can inspect a part in 2D and find short shots in a part, but for complex geometries, you’ll need material-handling to align with the lighting solution and sometimes some sort of 3D vision to accomplish the task,” Turek says.

While a plastic toy may not need to be perfect, there are many applications where either the part is large enough, such as an automotive body part, or raises questions of manufacturer liability, such as medical devices, that exacting inspection is required.

“Sometimes when you think of injection-molded parts, you think of a little part … something you could pick up with your hand,” says Aptura’s Dechow. “In the automotive industry, for example, molded bumper covers are one example of a large molded part that could result in significant waste if a process gets out of control. Even small parts, such as electronic connectors, can have complex surfaces and cavities where tolerances are very critical because subsequent manufacturing processes can be negatively affected by a problem with the molding process. Small changes in temperature and pressure can have a major impact on injection-molded parts. We’re working on a job right now that involves a small medical device part, but we have to measure it down to the tens of microns and feed that information back into the molding process. Otherwise, they may produce hundreds of parts before they realize they have a problem. An individual part is cheap, but if you don’t catch a problem quickly, the costs can add up. And in the case of a medical product, there’s a liability issue, even if the patient’s health isn’t in danger. Defective products can put a key customer relationship at risk and impact a manufacturer’s business.”

Despite the fact injection molding processes require high temperatures and pressures, the designers of machine vision systems say the primary challenge isn’t ruggedizing the equipment against ambient conditions. The challenge, they assert, is to get the vision system into the process in the first place. “The hardest part can be getting the camera and correct illumination into the machine so that it doesn’t interfere with the operation or the operator but still gets a good view of the mold itself,” Dechow says.

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.