Abstract: It is vital for reducing the spatter with surface tension transfer methods in CO₂ arc welding to extract the information of necking down. Once the necking down came into being, the arc voltage curve will create a push-up inflexion, and then, the arc voltage curve sharply rise. If this arc voltage saltation information can be extracted, we will get the characteristic information of necking down formation. Because wavelet methods have merits of local property in time domain and frequency domain, this paper studied the signal locality and singularity with wavelet methods. The study proved that wavelet methods could check the position and intensity of signal saltation. This method can accurately extract the characteristic information of necking down formation in CO₂ arc welding short circuiting transfer.