In the aim of inventing a new method of myopathy's EMG signals classification, diagnostic and treatment the first step is the creation of an EMG database. The raw EMG signals are rich in information but they have to be amplified, converted to a digital form and filtered to be useful in the diagnostic of myopathy. So, to create an EMG database a low cost and high performance acquisition board is designed. An instrumentation amplifier AD524, an Arduino board(to have a real time analogical to digital conversion) and a Wavelet based filter processed on Matlab software were used to obtain an exploitable EMG database. The correctness of stored signals is evaluated by comparing them with themselves when they were stored on computer using a digital scope(metrix 0X6062-II). An SNR metric of 74 dB is calculated, this SNR value is not sufficient to decide of the high performance of the acquisition board. Thus a turn back to compare the representation of a square signal obtained either with the designed board and with the digital scope was done. A calculation of some signal's parameters as period and amplitude shows that the new acquisition board reproduce the input signal shown directly on the screen of the digital scope more closely than that reproduced with the scope's generated data file.