Hall-Effect Sensor Fault Detection, Identification, and Compensation in Brushless DC Drives

In this paper, binary Hall-effect sensor faults are investigated in rectangular-current-fed brushless dc (BLDC) drives and a very effective methodology for their detection, identification, and compensation is explored. It is shown that these faults cause erroneous commutation, generally leading to unstable operation. Using a fault detection and identification technique proposed by the authors in a recent paper on low-cost fieldoriented drives, it is possible to pinpoint the faulty sensors. In this paper, it is demonstrated that the destabilizing effect of these faults on motion-state estimation can be compensated for in any position and speed estimation algorithm, as long as it is properly readapted. To this end, it is shown how to incorporate such fault-compensation in three state-of-the-art estimation algorithms: 1) the zeroth-order algorithm (ZOA); 2) the hybrid observer (HO); and 3) the vector-tracking observer (VTO). Comparative experimental tests are performed and it is verified that stable operation is achieved with three, two, or only a single Hall-effect sensor functioning correctly. These results show that the classical BLDC drive with three Hall-effect sensors has an inherent double redundancy to position-sensor faults. With the proposed method, this property can be exploited in systems that require very high reliability, such as in aerospace and automotive applications. Redundancy can be increased, using more than three Hall-effect sensors; reduced using two sensors; or eliminated using a single sensor, in ultra low-cost applications where redundancy is not a requirement.