Investigating Machine Learning Techniques for Predicting the Process Characteristics of Stencil Printing.

Stencil printing is the most crucial process in reflow soldering for the mass assembly of electronic circuits. This paper investigates different machine learning-based methods to predict the essential process characteristics of stencil printing: the area, thickness, and volume of deposited solder paste. The training dataset was obtained experimentally by varying the printing speed (from 20 to 120 mm/s), the size (area ratio from 0.35 to 1.7) of stencil apertures, and the particle size (characterized by a log-normal distribution) in the solder paste. Various machine learning-based methods were assessed; ANFIS-adaptive neuro-fuzzy inference systems; ANN artificial neural networks (with different learning methods); boosted trees, regression trees, SVM-support vector machines. Each method was optimized and fine-tuned with hyperparameter optimization, and the overfitting phenomenon was also prevented with cross-validation. The regression tree was the best performing approach for modelling the stencil printing, while ANN with the Bayesian regularization learning method was only slightly worse. The presented methodology for fine-tuning, parameter optimization, and the comparison of different machine learning-based methods can easily be adapted to any application field in electronics manufacturing.

Modified Split Ring Resonators Sensor for Accurate Complex Permittivity Measurements of Solid Dielectrics.

In this paper, a sensor using modified Split Ring Resonators (SRRs) is designed, simulated, fabricated, and used for advanced investigation and precise measurements of the real part and imaginary part solid dielectrics' permittivity. Adding vertical strips tightly coupled to the outer ring of the SRR leads to the appearance of two resonant frequencies at 1.24 GHz and 2.08 GHz. This modified geometry also assures an improved sensitivity. Using the full wave electromagnetic solver, both the unloaded and loaded sensors are investigated. The numerical simulations are used to develop a mathematical model based on a curve fitting tool for both resonant frequencies, allowing to obtain analytical relations for real and imaginary parts of permittivity as a function of the sample's thickness and quality factor. The sensor is designed and fabricated on 1.6 mm thick FR-4 substrate. The measurements of different samples, such as transparent glass, acrylic glass, plexiglass, and Teflon, confirm that the modified SRR sensor is easy to implement and gives accurate results for all cases, with measurement errors smaller than 4.5%. In addition, the measurements highlight the importance of the second resonant frequency in the cases in which numerical limitations do not allow the usage of the first resonant frequency (1 mm thick sample).