In this study, the power spectral density of simulated data which contain neuromuscular diseases and normal motor unit (i.e. control group) scenarios was calculated using Welch's method. Furthermore, the effect of Welch's method on differential diagnosis was investigated. Data were recorded both near innervation zone which is the area that motor unit action potential occurs and near tendon. Multi-layer perceptron was preferred as artificial neural network to investigate the effect of method on classification. When the data from innervation zone and from tendon were applied separately to the ANN, 68.67% performance was obtained. The accuracy of the network was increased up to 76% when data were applied together.