Detection of starting motor position in a brushless dc motor

Abstract

A microprocessor-controlled-circuit applies pulses to a standing motor whose rotor position is to be determined. The time required for the injected current to reach a predetermined threshold in a comparator is measured. The pulses are long enough to effectuate an accurate measurement, but not so long that the rotor moves. The pulse is applied twice to each phase, once with a positive polarity and once with a negative polarity, for a total of six measurements. The positive and negative rise times are subtracted from each other to determine whether the positive or negative current rise time was greater. Based upon the differences in the readings for all of the phases, the position of the rotor is known. There are differences in which the difference is zero or negligible. There are at least two methods to determine rotor position when there are ambiguities in the sign of the difference. In one method, the combination of the sign of the rise time difference or its absence plus the magnitude of the rise time can be used along with look-up tables to determine the rotor position. Another method is to resolve a position uncertainty by changing the effective rotor position. This is done by pulsing all three phases actively. Instead of leaving a phase open during pulsing, the third phase is tied high or is tied low. Once start-up ambiguities are resolved, control of acceleration from stand-still through medium speed to constant, nominal speed is provided.