Using both linear and non‐linear dimensional analysis, correction and prediction to improve PCB yields

Abstract

PurposeTo present an update and the latest results from work on a project aimed at enabling printed circuit board (PCB) manufacturing to reduce the effect upon production yields by material distortions during the manufacturing process.Design/methodology/approachMeasurement data was collated from inspection steps within the PCB manufacturing process and analysed to provide data describing the distortions of materials used. The data was used to generate a self‐learning model which can be used to predict material movement prior to manufacture.FindingsThe data collated enabled CAM engineers at the design introduction stage to model the PCB build and simulate the material movement that will take place during the manufacturing processes. With this knowledge, the correct compensations to counter the material movement can be applied to the production tooling, eliminating the need to run manufacturing test batches.Research limitations/implicationsThe technology has been used to ensure inner‐layer designs with nominal dimensions after the lamination stage. Further, development work should be undertaken to collate measured data from other parts of the PCB manufacturing process and model the material movement around all registration critical processes.Originality/valueThe paper details how data collated by production processes can be utilised to train mathematical models and provide modified‐tooling information to increase yields in the manufacturing process.