SBIR Phase I: Yield Improvement for Miniature Cameras

这个小型企业创新研究(SBIR)第一阶段项目将解决制造错误识别的技术障碍，该障碍从根本上限制了微型相机生产中可实现的产量。微型相机制造中使用的方法严重限制了对部件级计量数据的访问。只有在模块组装后的出厂质量控制中才能获得足够的表征数据，向制造商展示最终的产品性能，但几乎无法洞察限制性能的关键因素。当前的故障模式分析涉及在审计的基础上进行破坏性测试，以尝试识别错误。在这个项目中，将开发和测试一种算法，该算法基于典型的出厂质量控制数据来识别关键的组装和制造错误。该项目将开发系统模型，以评估没有模糊预测的可靠算法所需的测试数据输入或修改的最小集合，在硬件中确定有效数据分析所需的测试条件，并确定此类算法可实现的精度。预计具有设计意识的算法可以根据真实世界的测试条件准确识别制造错误，并对当前外出的质量控制测量进行最小程度的更改。该项目的更广泛影响/商业潜力是开发依赖自动化和精密工程而不是人工的美国制造基础设施，使美国公司能够在微型相机市场获得吸引力。此外，该算法提供的统计制造数据能够预测新设计的性能，从而允许更积极地探索创新的相机解决方案。在过去的十年里，微型相机市场经历了爆炸性的增长，现在超过70%的手机都有相机，每年售出超过10亿台相机。在这个大批量的行业中，产量和制造时间的改进可以对成本节约产生重大影响。对进一步降低成本的追求催生了大规模制造，数千个透镜元件同时制造并固定到传感器上，从而消除了昂贵的光学镜筒组装的需要。由于大量制造微型相机所涉及的众多工艺还不成熟，制造错误使高产量无法实现，从而抵消了任何成本节约。该项目的商业潜力很大，将使大规模生产的快速采用和可行性以及提高所有微型相机模块的产量成为可能。

This Small Business Innovation Research (SBIR) Phase I project will address the technical barrier of manufacturing error identification which fundamentally limits the achievable yields in the production of miniature cameras. The methods employed in miniature camera manufacturing severely limit access to metrology data at the component-level. Adequate characterization data is only obtained in out-going quality control after the modules are assembled, showing manufacturers the final product performance, but giving little insight into the key contributors limiting performance. Current failure mode analysis involves destructive testing on an audit basis to attempt to identify errors. In this project an algorithm will be developed and tested which identifies key assembly and fabrication errors based on the typical outgoing quality control data. The project will develop system models to assess the minimum set of test data inputs or modifications necessary to seed a reliable algorithm free from ambiguous prediction, identify - in hardware - the test conditions necessary for valid data analysis, and determine the accuracy achievable by such an algorithm. It is anticipated that a design-aware algorithm can accurately identify manufacturing errors with real world testing conditions and minimal changes to the current out-going quality control measurement.The broader impact/commercial potential of this project is the development of a U.S. manufacturing infrastructure that relies on automation and precision engineering instead of manual labor, enabling U.S. companies to gain traction in the miniature camera market. Furthermore, the statistical manufacturing data supplied by the algorithm enables predicted performance of new designs, allowing more aggressive exploration of innovative camera solutions. The miniature camera market has seen explosive growth in the last decade as now over 70% of cell phones have cameras and more than one billion cameras are sold each year. In this high volume industry, improvements in yield and manufacturing time can have a significant impact on cost savings. The pursuit of additional cost reduction has given rise to mass manufacturing where thousands of lens elements are simultaneously fabricated and affixed to sensors, eliminating the need for costly optical barrel assembly. Due to the immaturity of the numerous processes involved in mass manufacturing of miniature cameras, fabrication errors make high yields unattainable, negating any cost savings. The commercial potential of this project is large and will enable the rapid adoption and viability of mass manufacturing as well as improving the yields of all miniature camera modules.