Syringe Needle Inspection
Detect defects in complex beveled needle tips
Hypodermic needles, or cannulas, are made from metal rolled into tubes that are extruded through dies to thin and harden them. To penetrate the skin, the tip must be beveled, typically multiple times at various angles, to create a tip appropriate for the intended purpose. The process of grinding the bevels can be uneven, or leave burrs, hooks, and other irregularities that impair function, or can actually be dangerous to the patient.
Because it takes multiple die pulls the create the proper gauge of needle, there can be variations in the internal or external diameter. Inspecting hypodermic needle for defects is important to protect patient safety and ensure the proper administration of vaccine.
Cognex Deep Learning detects the variety of subtle defects of the beveled tip. The defect detection tool is trained using a small sample set of images at high magnification representing the range of acceptable bevels. Any variations in lighting will reveal the structure of the needle surface, with high reflectance indicating smoothness and dullness indicating potential flaws. This same procedure can also allow a needle dimensional inspection to reveal the inside and outside diameters of the needle’s tip.
Additionally, High Dynamic Range Plus (HDR+) technology adjusts to localized contrast changes automatically, and minimizes problems caused by changing reflections, providing a more stabilized illumination of the bevels. HDR+ differs from standard HDR as it can be done with a single acquisition at high-speed on moving parts, whereas standard HDR would need to be stationary and capture multiple images to obtain the same results. The use of collimated backlight and a telecentric lens can also be used to keep magnification constant making precise measurement easier.
As new needle tip models are developed or manufacturing lines are changed over, Cognex Deep Learning is easily and quickly trained on the different models, minimizing downtime and optimizing throughput.