Electric Power Systems Research 77 (2007) 385–392 Rotor bar fault diagnosis in three phase induction motors by monitoring fluctuations of motor current zero crossing instants Hakan C ¸ alıs ¸ ∗ , Abd ¨ ulkadir C ¸ akır Department of Electronics-Computer Education, The Faculty of Technical Education, Suleyman Demirel University, Isparta, Turkey Received 8 December 2005; received in revised form 30 March 2006; accepted 31 March 2006 Available online 8 May 2006 Abstract This study describes broken bar detection in induction motors without using additional sensor. It is based on observation of the fluctuations of stator current zero crossing times (ZCT). Instead of sampling motor current with a high resolution A/D converter, zero crossing instants are recorded as waveforms cross zero. Fluctuations in the intervals between successive zero crossings of the three phase current waveforms are analysed using Fast Fourier Transforms (FFT). Diagnostic information is found in the spectrum of the ZCT signal through the presence of specific fault related frequencies. A rotor bar fault is manifested as an increase in the amplitude of the 2sf and other spectral components. This paper analyses the effect of an electrically unbalanced rotor on the ZCT spectrum of stator current, and discusses the various frequency components associated with rotor bar faults seen in the ZCT spectrum. It is important to eliminate the dependence of the index on the parameters of the induction motor. Particularly, the effect of motor inertia, supply harmonics, and variable load are discussed to increase the reliability of the rotor fault index, and simulation results are presented. It is found that the 2sf frequency component is independent of inertia, load, and harmonics, and thus it is suitable as an index for broken rotor bar. © 2006 Elsevier B.V. All rights reserved. Keywords: ZCT signal; Fault diagnosis; Rotor fault; Broken rotor bar; Induction motors 1. Introduction Induction motors are most commonly used electrical rotat- ing machine in industry. However, it is becoming increasingly important to use condition monitoring techniques to give early warning of imminent failure. Many of motor faults have an elec- trical reason. Rotor bar faults are usually associated with high temperatures, high mechanical loading particularly during the starting time, or any defective casting or poor jointing during the manufacturing process. Initially, they started as high resis- tance causing high temperature and then progress as cracking or small holes in the rotor bars. They are more likely to take place near the cage end rings. Although rotor bar faults covers approximately 10% of overall fault conditions in squirrel-cage induction motors, many research works have been implemented for detection of induction motor rotor faults during the past years. Therefore, many monitoring algorithms have been proposed in ∗ Corresponding author. E-mail addresses: hcalis@tef.sdu.edu.tr (H. C ¸ alıs ¸), cakir@tef.sdu.edu.tr (A. C ¸ akır). the literature. Now, there is a good understanding for the rotor bar fault mechanism. Rotor bar fault detection has been imple- mented by monitoring pulsations in speed, airgap flux, axial flux, vibration, and current [1–7]. The main disadvantage of these methods is that they all require using additional sensor for the monitoring. Additionally, the sensitivity of fault detec- tion mainly depends on the inertia of the load, and it is difficult to determine the degree of fault level. The preferred and widely used approach to rotor bar fault detection is the analysis of stator current in the frequency domain, due to its non-invasive acces- sibility [7]. A sensorless zero crossing times method, observing the changes on the motor current zero crossing instants, has also been shown to be able to use for speed measurement and fault detection [8]. The conventional indicator of the broken rotor bars in the single phase of motor current spectrum is well recognized by sidebands displaced by 2sf Hz around the mains peak at full load working conditions with high slip, where f is the supply frequency and s is the motor slip. In practice, motor load may not be steady and if current is sampled when the motor load changes, sidebands may not be 0378-7796/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2006.03.017