Abstract:
In the development of modern electronics based on printed circuit boards, cross-talk occurs due to the mutual influence of interconnections. If the interference signal exceeds the permissible threshold values, it can lead to false triggering of an element, and thus the whole system. The aim of this work is to implement a technique and predict the parameters of cross-talk in PCB interconnects based on an artificial neural network. The maximum amplitude and duration of interference at the near and far ends of the interconnect are studied. An artificial neural network with a fully connected architecture with a few layers and neurons was used for the research. An acceptable divergence of prediction results, not more than 18% for all cross-talk parameters on the test sample, was obtained. Examples of predicting the maximum amplitude and duration of cross-talk are given.
Fulltext:
PDF file (274 kB)