Analog filter diagnosis using the oscillation based method

Abstract

Oscillation Based Testing (OBT) is an effective and simple solution to the testing problem of continuous time analogue electronic filters. In this paper, diagnosis based on OBT is described for the first time. It will be referred to as OBD. A fault dictionary is created and used to perform diagnosis with artificial neural networks (ANNs) implemented as classifiers. The robustness of the ANN diagnostic concept is also demonstrated by the addition of white noise to the “measured” signals. The implementation of the new concept is demonstrated by testing and diagnosis of a second order notch cell realized with one operational amplifier. Single soft and catastrophic faults are considered in detail and an example of the diagnosis of double soft faults is also given.

Parametric Circuit Fault Diagnosis Through Oscillation-Based Testing in Analogue Circuits: Statistical and Deep Learning Approaches

“…Firstly, to verify the efficacy of ANN-based OBD under more realistic circumstances, all non-faulty passive component tolerances are included in the simulation for the first time in this work, by means of a generalized Monte Carlo sampling. This stands in contrast to current techniques that use ANN's to diagnose faults [2], [11], where the assumption is that all non-faulty passive components are fixed at their nominal value. This is a far more accurate representation of real-world test conditions [14], [15] than has been considered previously.…”

“…In [10], a feed-forward artificial neural network, with the basic Multi-Layer Perceptron (MLP) structure, is used to diagnose faults. A supervised learning procedure for capturing the fault dictionary is first applied to OBD in [11], with the feature extraction extended to the first four harmonics in [2]. White noise, with an amplitude of 1% of the output signals, was added to the simulated signals to verify robustness, though the link between this introduction and probability component value variation is not presented.…”

“…This time series diagnosis is performed both with a statistical learning approach [16], and a deep learning approach [17]. Both approaches are compared to the harmonics analysis currently used in OBD [2], [11], [12] and class activation maps (CAMs) used to highlight which data are more crucial in correct classification. This is a significant finding in selecting of suitable data, potentially reducing the resources required in collecting data that does not contribute to classification accuracy.…”

Design of an Oscillation-Based BIST System for Active Analog Integrated Filters in 0.18 µm CMOS

“…The oscillation-based BIST method is based on splitting complex mixed signal or analog circuit into smaller functional blocks-operational amplifier, comparator, phase detector, Schmitt trigger, first or second-order filter [5,[15][16][17]. Then, the functional block ( Figure 1a) is converted into a robust oscillator by connecting additional feedback circuit, as shown in Figure 1b [18].…”

Testing of low voltage two stage operational amplifier using oscillation test methodology

“…Natural oscillation frequency is determined by the Fast Fourier Transform (FFT) analysis of oscillatory circuit. The tolerance band of nominal frequency of the CUT is determined by Monte-Carlo simulation [14]. The tolerance of 10% and 15% has been considered for resistance R 1 , R2 and C 1 , Cc respectively for the oscillator circuit.…”