Sentinel uses a fast geometric search algorithm to locate a pre-programmed reference point and a high-accuracy adaptive comparison algorithm to verify that all regions are filled with rubber. Source: PPT Vision Inc.

Dickten Masch Plastics (Milwaukee) shuttles metal and plastic components from metal forming machines to multi-stage plastic and rubber injection molding machines, producing millions of precision electrical connectors and similar parts each year. The company certainly does not need a preventable manufacturing error bringing the multimillion-dollar operation to a halt.

“We realized we needed a fully automated machine vision system to look at the parts in the plastic injection molding process, catch short shots and protect our molds from parts that failed to eject and jammed ejector pins,” says John Berges, manufacturing engineer at Dickten Masch. “After interviewing three candidate suppliers, we choose PPT Vision’s (Bloomington, MN) Sentinel.”

Dickten Masch uses a two-camera Sentinel solution to inspect rubber injection molded electrical connectors. Source: PPT Vision

Smart Camera Solution

This was Berges’ second experience with machine vision. Dickten Masch previously had installed a PPT Vision Impact smart camera system to inspect out-of-mold parts for flashing, short shots and other defects. The new vision system, however, checks rubber insulated electrical connectors used in automobiles for short shots before the parts are ejected from the mold. The vision system also protects the mold by looking for parts that were not ejected, as well as ejector pins that failed to retract after ejecting.

“Real estate is really tight on our floor,” says Berges. “PPT Vision’s Sentinel system is so small you don’t even know it’s there. It’s mountable in multiple locations, and we only bring out the laptop to display the system’s operation during tours with customers or to check something with the mold system. Seeing each connector inspected in real time really gives the customer a good feeling that they can count on us to deliver the parts they need.”

Sentinel is designed specifically for molding operations, allowing operators to ensure quality by checking for short shots, and to prevent expensive tooling damage and downtime by checking for part ejection, pin retraction and insert position.

PPT Vision engineers developed the Impact line of general purpose smart camera machine vision systems into a software wrapper that offers molded part manufacturers the features they need in a small footprint. Gary Kocken, director of sales for PPT Vision Inc., suggested that Dickten Masch use a two- camera Sentinel solution to inspect the two rows of rubber injection molded electrical connectors-one camera for the top row of four connectors, and a second camera for the lower four connectors.

The Sentinel system offers up to four cameras without requiring any additional processing or I/O hardware, although multiple Sentinel solutions can be networked together. Sentinel’s smart camera approach to mold inspection means the computer processing elements are inside the system itself, eliminating the need for a separate computer to run the Sentinel mold inspection software. This eliminates the need for an industrial PC and frame grabbers.

During a training session, Berges learned the Sentinel system in a matter of hours, using its three-tab software setup. The first tab contains information on the number of cameras the system will use, camera settings for each inspection step and a diagnostic area to make sure the system and cameras are communicating. Like many molding companies, Berges uses a custom variety of the molding industry’s SPI data interface for his I/O needs.

The first tab allows operators to configure the eight-in/eight-out I/O to adapt to the molding equipment interface. The second tab allowed Berges to set up the inspection routine for the high-tension electrical connectors. By viewing a picture of the mold and using a laptop mouse temporarily connected to the system during setup, Berges acquired images of good product in the mold, and then designated regions of interest around critical mold areas. During operation, Sentinel uses a fast geometric search algorithm to locate a pre-programmed reference point and a high-accuracy adaptive comparison algorithm to verify that all regions are filled with rubber, and no short shots have occurred.

Berges continued programming the system to check that parts are ejected from the mold and that ejector pins retract.

The third tab shows the system summary during operation, including images from each camera as they inspect the molded parts and protect the mold from jams and damage. By clicking on a camera view, Berges can instantly see all production data from that particular camera.

“We designed the Sentinel software to automatically adapt to up to four cameras,” explains Kocken. “This makes it easy to switch from one camera view to another, or from one inspection stage to another. Sentinel can include up to four cameras, and five inspection stages per camera, which is a common need for injection molders who often mold one material around an insert or another material to create a finished part. Sentinel can store its recipes right on the camera. While many molding lines are dedicated with little product changeover, this feature will help those customers who run a variety of parts on the same molding machine.”

After the Sentinel is programmed, the system can be run in “output disabled” mode to assist in file optimization without causing press downtime, and after the operator is satisfied, the “outputs enabled” button is selected, which places the system online, and the laptop can be disconnected for headless operation. The system automatically does the rest, taking orders from the molding machine itself and sending checks, stops and alarms to the operator through the machine’s standard interface; there is no need to install additional displays or train operators on new system operation.

Up and Running

The most delicate stage of any machine vision installation is the programming and setup phase. Using simple Internet meeting software, customers do not have to worry that they will have to solve production floor challenges on their own.

“When we were setting up, a PPT Vision technician set up a Web-conference meeting while I was right on the shop floor,” says Berges. “I showed them my program; they looked at it and suggested improvements. Whatever they did, I could see and vice versa. It really helped being able to connect to PPT Vision right from the shop floor rather than having to run back and forth from my office, making a change to the system, and then calling them back. In 20 minutes, we had the system up and running.”

Sentinel’s small size and “headless” operation means that it can use small magnetic bases and arms to mount a variety of cameras and LED lights that complement any molding inspection operation. Sentinel builds on the Impact general-purpose machine vision line, which includes 17 different camera models, and almost any combination of optics and lighting.

“With Sentinel, PPT Vision has delivered a system that gives molding companies the quality assurance they need at an acquisition and operational cost they can afford to put on every piece of equipment,” says Kocken.

PPT Vision Inc.
(952) 996-9500


- Because the Sentinel is a smart camera system, Dickten Masch did not need to buy an industrial PC and frame grabbers.

- Dickten Masch’s manufacturing engineer was able to learn the Sentinel system in a matter of hours, using its three-tab software setup.

- Using Internet meeting software to assist Dickten Masch’s manufacturing engineer, the system setup was completed within 20 minutes.