Researcharrow
Sensorless Short Shot Detection Technology

Sensorless Short Shot Detection Technology

This is some text inside of a div block.

Background

One of the types of defects in injection molding is short shot. This phenomenon occurs when the resin solidifies before reaching the end of the mold, leading to an inability to accurately transfer the product's shape. Short shots are particularly prone to occur in thin-walled and small-sized molded products.

 

 

Figure 1. Short shot

The common method for detecting short shots involves installing internal pressure sensors in the mold and analyzing data from these sensors to monitor how far the resin has reached within the mold. However, this method requires modifications to the mold and the purchase of internal pressure sensors, leading to increased product costs. Consequently, there is a growing need for a more cost-effective and efficient method to detect short shots.

 

Figure 2. Defect Detection Using Mold Internal Pressure Sensor

In light of this background, MAZIN conducted an experiment to detect short shots without using internal pressure sensors in the mold. 

Experimental Method

In this study, we adopted a new approach that does not require mold modifications, using an externally attachable clamp-type current sensor. 

Figure 3. Defect Detection Using Current Sensor

As shown in Figure 3, this sensor was attached to the servo amplifier of the injection molding machine, and electrical current data was collected for each molding shot.

In the experiment, data for 33 shots was first collected under normal molding conditions for good quality products. Then, the molding conditions were deliberately altered to induce short shots, and data for 10 shots under each of the six different conditions were collected, totaling 60 shots. 

Consequently, this method enabled the collection of electrical current data for a total of 93 shots, encompassing both the conditions for good quality products and those leading to short shots.

Figure 4. Experiment Overview

Results

The time-series data obtained from the experiment is shown in Figure 5. In this figure, the orange line represents the current of the Plasticization Motor, which is the motor current used to rotate the screw.

Figure 5. Servo Motor Current Data

On the other hand, the blue line represents the current of the Injection Motor, which is the motor current used for advancing the screw and for the piston ejection movements.

From the collected current data, features were extracted and dimensionality reduction was performed to determine the centroid of the data group for 33 good quality shots. Figure 6 shows the plot of the distance from the centroid of the reduced-dimensionality feature set and the good quality data group for each shot.

Blue dots represent good quality shots, while others represent short shots under various molding conditions. From this figure, it is observed that when a short shot occurs, the distance from the centroid of the good quality data group tends to be relatively larger. By setting a threshold at the dotted line shown in the figure, it becomes possible to identify any shot exceeding this threshold as a short shot.

Figure 6. Anomaly Level for Each Shot

 Conclusion

The method developed in this study, which analyzes the current data from servo motors, is limited to detecting molding defects related to the filling and metering processes, as it cannot utilize information from the holding pressure stage. Compared to internal pressure sensors, the information that can be extracted is limited; however, it has been demonstrated that the detection of short shots is possible by extracting unique feature sets. This study marks the first successful detection of short shots using a configuration that installs a clamp-type current sensor on the servo amplifier of the injection molding machine.

 The greatest advantage of this method is that it allows for the easy initiation of short shot detection without the need for modifying the mold. This enables manufacturers to reduce costs while conducting efficient and rapid quality control.

 Moving forward, we will endeavor to develop new technologies that cater to a broader range of defect detection in molding processes.