Characterizing digital light processing (DLP) 3D printed primitives

The resolution and repeatability of 3D printing processes depends on a number of factors including the software, hardware, and material used. When printing parts with features that are near or below the nominal printing resolution, it is important to understand how the printer works. For example, what is the smallest unit shape that can be produced? And what is the reproducibility of that process? This paper presents a method for automatically detecting and characterizing the height, width, and length of micro scale geometric primitives produced via a digital light processing (DLP) 3D printing process. An upper limit, lower limit, and best estimate for each dimension is reported for each primitive. Additionally, the roughness, rectangularity, and tilt of the top of each primitive is estimated. The uncertainty of the best estimate is indicated using standard deviations for a series of primitives. The method generalizes to unseen primitives, and the results illustrate that the dimension estimates converge as the size of the primitives increases. The primitives’ rectangularity also increases as the size increases. Finally, the primitives specified with 5 to 68μm varying heights have been estimated to group into five different heights with fairly low variance of the best estimates of the heights. This reflects how the requested geometry is parsed and produced by the printer.