Optimized mask image projection for large-area based additive manufacturing process

Large-area additive manufacturing based on mask image projection such as digital micro-mirror devices (DMD) has the potential to be a fast and inexpensive. More and more research and commercial systems have been developed based on these processes. For this kind of additive manufacturing process, the mask image planning is an important process planning step. In this research work, we present an optimization based method for mask image planning. It is based on a light intensity blending technique called pixel blending. By intelligently controlling pixels’ gray scale values, we can achieve a much higher XY resolution and accordingly better part quality. We mathematically define the pixel blending problem and discuss its properties. Based on the formulation, we present several optimization models for solving the problem including a mixed integer programming model, a linear programming model, and a two-stage optimization model. Both simulated and physical experiments for various CAD models are presented to demonstrate the effectiveness and efficiency of our method. ❧ In this large-area additive manufacturing system, we use a low cost off-the-shelf digital light processing (DLP) projector based on DMD. Due to the limited resolution and certain image blurring, the projection pixels are not uniform and consistent, thus we further present a calibration method for capturing this non-uniformity of the projection image. Our method is based on two calibration systems, a geometric calibration system that can calibrate the position, shape, size, and orientation of a pixel and an energy calibration system that can calibrate the light intensity of a pixel. Based on both results, the light intensity at various grayscale levels can be approximated for each pixel. Developing a library of such approximation functions is critical for the optimized pixel blending to generate a better mask image plan. Experimental results verified our calibration results. ❧ To further improve the accuracy and resolution of built components, we present a process by using multiple DMDs. Besides the original pixel blending, the mask image planning in this process needs to compensate the calibrated light intensity in a projection image that is non-uniform and non-linear. We present a general optimized pixel blending method based on direct discrete search (DDS). Its mathematic model and computing method for the mask image planning are presented. Various test cases have been performed to verify its effectiveness and efficiency. Based on the multiple DMD idea, we step further and proposed a novel method based on the mask video projection. For each layer, as set of mask images, instead of a single image, are planned based on the optimized pixel blending. The planned images are then projected in synchronization with the small movement of the building platform. Experimental results have verified that this method can effectively improve the XY resolution of the large area projection based process. ❧ Even though the proposed optimization methods such as LP, DDS, etc can optimize the mask image planning process, they still have difficulties in real word applications. The commercial systems usually use projectors with much higher resolution and built part with much larger scale. It is very challenging to solve the problems in customer acceptable time. To address this issue, a new method called boundary erosion is proposed. The experimental results verified its effectiveness in dimensional accuracy. However, boundary erosion method is only based on geometrical information and does not consider the energy information. Thus, it can not preserve the thin feature and sharp features. To solve this problem, another heuristic method called error diffusion is proposed. Error diffusion method combines boundary erosion and pixel blending together, uses the error as feedback and works in a so called closed loop system. A pixel blending simulator was developed for the physical parameters calibration. An adaptive sampling strategy was used to enhance the efficiency of the error diffusion method. Experimental results showed that the proposed method can achieve acceptable accuracy (50μm), it can also recover the thin feature and sharp features. ❧ In addition to improve the surface quality, we are also aiming at enhancing the physical property such as the mechanical property, electrical and electronic property, multifunctional property, etc. We investigated the feasibility of using digital mask projection based additive manufacturing process in fabricating multiple materials. Several challenges such as shallow vat, part separation, transition between multiple tanks including washing and drying, material bonding etc are presented. A novel method based on two-channel bottom-projection was proposed and a prototype based on rotation table was also fabricated. Experimental results show that we can build parts with different mechanical properties, different electrical conductivities properties and even flexible digital material can be effectively fabricated by using the proposed method. ❧ Finally, we investigated high frequency ultrasound transducer fabrication using DMD projection based additive process. Several challenges such as the high viscosity and low penetration depth are presented. The curing characteristics of the PZT slurry was analyzed and corresponding manufacturing process based on two-channel system was proposed. Both green-part fabrication and post-processing of green-parts including pyrolysis and sintering have been studied. A testbed has been developed to fabricate photo cured piezoelectric materials with micro-scale features. Different piezoelectric characteristics of the materials have been tested and compared with the bulk case.