Simulation and Fault Diagnosis in Post-Manufacturing Mixed Signal Circuits
In this paper, an effective method for predicting circuit failures in post-market circuit boards through simulation and deep learning is proposed and implemented.
Kyle Pawlowski, Sumit Chkravarty, Arjun Kumar Joginipelly
Kennesaw State University
A major problem in circuit board remanufacturing is the identification of parametric faults from age or stress to the individual passive components. We propose a deep machine learning system for simulating and identifying such faults. A simulated dataset is generated for the most common faults in a circuit. This dataset is used to train deep machine learning classification algorithms to identify and classify the faults. The accuracy of system is measured by comparing with real circuit boards in operation.
In this paper, an effective method for predicting circuit failures in post-market circuit boards through simulation and deep learning is proposed and implemented. Normally, such failures are hard to diagnosis and difficult to fix. This test can be performed quickly to provide results that can lead to circuit fixes. We have shown that with a small amount of development time, real tests can be created for circuits that incorporated this method. The accuracy of the networks shown here support the idea that the time saved repairing circuit boards using the proposed method can far exceed the costs of misclassifications. With verification from real circuit boards in operation, we know that our data provides an accurate ground truth for the deep network to learn.
Initially Published in the SMTA Proceedings
No comments have been submitted to date.