Neural Network based Indirect Estimation of Functional Parameters of Amplifier by extracting features from Wavelet Transform
Supriyo Srimani, Kasturi Ghosh, Hafizur Rahaman · 2021
In this work, a cost-effective methodology has been proposed for estimating functional parameters of analog circuits to ensure faster production testing. The new test algorithm predicts the functional parameters of the circuit under test from the output response, using wavelet transform as a preprocessor and artificial neural network (ANN) as a prediction algorithm. Daubechies wavelet transform is used for analyzing the output response obtained by exciting the test circuits with a pseudorandom analog signal. The dimension of extracted features is reduced by the principal component analysis method. Specification of the test circuits is then predicted with the help of an artificial neural network. The proposed methodology for performance prediction is verified for Two-Stage Operational Amplifier (OPAMP). The proposed algorithm is also validated with a practical IC, i.e., TLV2762 from Texas Instruments. The accuracy of the proposed prediction algorithm confirms that the proposed methodology can be easily implemented into a test board that can be treated as an inexpensive tester.