The transmission characteristics of the printed circuit board (PCB) ensure signal integrity and support the entire circuit system, with impedance matching being critical in the design of high-speed PCB circuits. Because the factors affecting impedance are closely related to the PCB production process, circuit designers and manufacturers must work together to adjust the target impedance to maintain signal integrity. Five machine learning models, including decision tree (DT), random forest (RF), extreme gradient boosting (XGBoost), categorical boosting (CatBoost), and light gradient boosting machine (LightGBM), were used to forecast target impedance values. Furthermore, the Optuna algorithm is used to determine forecasting model hyperparameters. This study applied tree-based machine learning techniques with Optuna to predict impedance. The results revealed that five tree-based machine learning models with Optuna can generate satisfying forecasting accuracy in terms of three measurements, including mean absolute percentage error (MAPE), root mean square error (RMSE), and coefficient of determination (R2). Meanwhile, the LightGBM model with Optuna outperformed the other models. In addition, by using Optuna to tune the parameters of machine learning models, the accuracy of impedance matching can be increased. Thus, the results of this study suggest that the tree-based machine learning techniques with Optuna are a viable and promising alternative for predicting impedance values for circuit analysis.

/pdf/tree-based-machine-learning-models-with-optuna-in-predicting-2z3qqk7b.pdf

Tree-Based Machine Learning Models with Optuna in Predicting Impedance Values for Circuit Analysis

An optimization routine is applied for the decoupling capacitor placement on Power Distribution Networks to identify the limit beyond which the placement of additional decaps is no longer effective, thus leading to wasting layout area and components, and to a cost increase. A specific test example from a real design is used together with the required target impedance and frequency band of interest for the PDN design. The effectiveness of the decap placement while selecting different layers of the stack-up, and while moving the upper limit of the PDN design band is analyzed. Such analysis leads to helpful insights based on the progression of the input impedance during the optimization process, and to develop useful guidelines for avoiding over-design of the PDN. 

A Methodical Approach for PCB PDN Decoupling Minimizing Overdesign with Genetic Algorithm Optimization

Decoupling capacitors are used to provide adequate and stable power for integrated circuits in printed circuit boards (PCB). For complicated and large designs, it is difficult to select capacitors to meet voltage ripple limits while also minimizing cost because the search space is too large. In this work, a new genetic algorithm (GA) is proposed for the selection and placement of capacitors to meet a target impedance using as few capacitors as possible. The GA is centered around controlling the number of unused port locations in the GA population solutions, with the result of smoothing out the GA convergence and speeding up the convergence rate. A result comparison is made of the proposed GA against other algorithms and found the GA competitive if not better for the select test cases.

Bo Pu

Papers

A Modified Genetic Algorithm for the Selection of Decoupling Capacitors in PDN Design

Fast impedance prediction for power distribution network using deep learning