The condition monitoring of an induction motor performs by contact or non-contact methods. The contact methods execute via vibration signal, temperature measurement, current signature analysis and so on, and the non-contact methods accomplish via thermal imaging, temperature measurement and acoustic emission measurement. The contact method requires human expertise to evaluate failures and it is laborious. The authors propose a novel, low-cost and non-contact method using software phase locked loop (SPLL) for electrical fault identification in the three-phase squirrel cage induction motor. The handheld Doppler ultra wideband radar transmitted signal is focused on the induction motor and the reflected signal is analysed with SPLL. The SPLL error signal correlates the faults. From the experimental results, the fault identification of motor in the earlier stage achieves better accuracy of the different motor fault condition. The pattern of error signal provides the information of various faults in squirrel cage induction motor.