Abstract. The AC series fault arc tests are conducted by a self-design platform according to several common test standards for fault arc circuit breakers in this paper. Compared with the no-arc current waveform, the current of fault circuit has shoulders when the current passes zero, and the current of fault circuit is lower than the no-arc current. A united method of the wavelet transform (WT) and the information entropy (IE) is proposed to distinguish the fault and normal circuit. By the WT of arc current and the analysis of multiple-sampling, the results reveal that the arc fault frequency bands (FB) range is (625, 1250) Hz. The wavelet component D3 of arc current and its wavelet energy (WE) have large value during the arc period. The wavelet energy of no-arc circuit is lower than 0.006. The IE is bigger than 1 in the fault circuit. Therefore, the FB, WE, IE can be chosen as the fault characteristics to detect the fault circuits.