Evaluation of the Applicability of Neural Networks to Suppress Harmonic Distortion of HF Power Amplifiers

Author(s):  
A. P. Pavlov ◽  
I. E. Kashchenko ◽  
S. V. Krivaltsevich
Algorithms ◽  
2021 ◽  
Vol 14 (11) ◽  
pp. 329
Author(s):  
Venkataramana Veeramsetty ◽  
Bhavana Reddy Edudodla ◽  
Surender Reddy Salkuti

Zero-crossing point detection is necessary to establish a consistent performance in various power system applications, such as grid synchronization, power conversion and switch-gear protection. In this paper, zero-crossing points of a sinusoidal signal are detected using deep neural networks. In order to train and evaluate the deep neural network model, new datasets for sinusoidal signals having noise levels from 5% to 50% and harmonic distortion from 10% to 50% are developed. This complete study is implemented in Google Colab using deep learning framework Keras. Results shows that the proposed deep learning model is able to detect zero-crossing points in a distorted sinusoidal signal with good accuracy.


1999 ◽  
Vol 35 (12) ◽  
pp. 950 ◽  
Author(s):  
I.D. Mosely ◽  
P.H. Mellor ◽  
C.M. Bingham

2021 ◽  
pp. 86-94
Author(s):  
Yu.A. Bryukhanov ◽  
K.S. Krasavin

The output power amplifier is operated in nonlinear mode to increase its efficiency, at which the signal is distorted more. The goal is to analyze effects of nonlinearity on the modulating signal of angle-modulated signal distortions for single-ended and push-pull output power amplifiers with memory and without it. The amplifier characteristic ic(vBE) is preset by the cubic polynomial. The output voltage of memoryless amplifier uout(t) is found by Ohm's law. And in case with amplifier with memory, the output voltage uout(t) is found by Complex amplitude method with a Fourier series of function ic(t). The output voltage uout(t) is represented as the narrowband process with envelope of the waveform U1(t) and initial phase Ф1(t) to determine the angular modulation law Ф1(t) at the amplifier output. The function U1(t) is defined as combination of in-phase А(t) and quadrature В(t) components. The distortion estimate based on Total Harmonic Distortion (THD). The measurement of THD is defined as the ratio of the RMS amplitude of a set of higher harmonic frequencies to the RMS amplitude of the first harmonic. The spectrum of function Ф1(t) is represented a Fourier series. In single-ended memoryless amplifier and non-cutoff mode, the THD decreases as m increases and THD of modulating signal is {37.6; 18.8; 6.3%} at m {0.5; 1; 3}. In cutoff mode, the smallest THD is {15.18; 5.13%} at m {1; 3} as θ = 140o. In push-pull memoryless amplifier and θ = 90o, the THD is {22.2; 11.1; 3.7%} at m {0.5; 1; 3} which is 1.7 times less than in a single-ended amplifier without cutoff. This is due to compensation even-order harmonic of amplified signal at the output of a pushpull amplifier. In this case of memoryless amplifier, the transistor load is a parallel RLC circuit. The distortion decreases under m increases at first (up m is 1.5), but then rapidly increases (from m is 1.7). This is due to increase of the parasitic harmonics level caused by the nonlinearity of the phase-frequency characteristic of RLC circuit. The calculations show that THD of modulating signal is 3.35, 9.02 and 22.88 % at Q = 10 and m {0.5; 1; 3}, respectively. In cutoff mode, the smallest THD is at θ = 90o and it does not depend on θ at large values of m. In push-pull amplifier with memory, the THD is 2.8, 2.5 and 3.0 times less than for a singleended amplifier without cutoff at Q {2.5; 5; 10}, respectively. The results of this paper can be used for the design of signal processing systems.


Sign in / Sign up

Export Citation Format

Share Document