An Oak Ridge National Laboratory printer sped up when its deposited plastic became too cold. That correction sounds counterintuitive until the relevant moment is clear: the controller targets the temperature of a layer when the next layer arrives.
The feedback loop
A large-format robotic nozzle extrudes heated plastic composite layer by layer. The material must remain warm enough for one layer to bind to the next, yet cool enough to retain the printed shape. ORNL monitored nozzle position, print speed, and the temperature of dispensed plastic with sensors. Six small thermal cameras around the nozzle watched the deposited material as it cooled.
Computer vision located hot material in the live thermal images and estimated its temperature. When it differed from the target, the controller adjusted print speed so the layer could reach the intended temperature before another was added. This joined observation and correction within the printing process.
What the demonstration established
The team calibrated the controller and six cameras, then printed a hexagon larger than a truck tire. They began at a low speed to challenge the controller. Material was about 30 percent too cool by the next-layer step; the controller raised speed to keep the layers at a temperature suitable for fusion. The print bed lowered slightly as each plastic layer was added.
Earlier ORNL work combined thermal images with statistical models to recognize printing faults. In this demonstration the temperature reading led directly to a speed change during the print. ORNL also says the controller does not need retraining for each new design and was designed for different large-area printers, plastics, and shapes. The reported full-scale trial was the particular hexagon and low-speed start, so performance across those designs remains unmeasured in this article.
Sources:
Reader links:
Top comments (0)