Files
Abstract
This project final report contains the project background, specifications,results, and analysis for a Master of Science in Engineering capstone design project. The purpose of the project was to design and implement sensorless speed detection in cordless power tools which are powered by a brushed DC motor [i.e., a permanent magnet DC motor, or PMDC]. Project requirements included a speed detection operating range of 500 to 18,000RPM, and a steadystate error of 5%. The implementation needed to operate on 8-bit microcontrollers in high dynamic load conditions. The cost of the components utilized in the sensorless design needed to be less than 50% of the cost of the hall encoder components used in the existing design.
Traditional methods of motor speed detection in cordless power tool designs involve either a hall sensor and a ring magnet or a rotary encoder. Implementing a sensorless design would not only reduce the total size and complexity of the electronics hardware design, but also significantly reduce the overall cost of the electronics. This project report features a review of the relevant literature, design of various speed measurement methods, test results, and analysis of the performance of each method.
A new implementation method was developed for detecting motor speed by modifying and combining components of the three established methods for detecting motor speed: Back Electromotive Force [BEMF], Current pulse counting, and inductive BEMF spike measurement. Real world loading data were collected using a modified Milwaukee Tool M18 Multitool design. The original hall sensor and ring magnet encoder provided a reference for the actual speed that the new implementation was checked against. Performance of the new sensorless speed detection method was compared to the original sensor-based implementation in controlled loading scenarios and real-world applications.
The BEMF method met all the project goals and requirements. The performance in the open-loop, closed-loop, and application testing all fell well below the targets. The cost associated with this method was only 0.8% of the original component cost. The BEMF inductive spike method was found to be technically feasible, but impractical because of the large amount of motor specific variable mapping required. The current ripple method was the most accurate algorithm and does not require the use of individual motor parameters, but the method is limited by the sampling rate and clock frequency of the microcontroller.
A significant result is that autocalibration was used to successfully combine the BEMF method and the current ripple method, such that the BEMF method could be automatically calibrated by the current ripple method. The current ripple method could be used exclusively in microcontrollers with high clock speeds. As the performance of low-cost microcontrollers continues to increase, this method will likely become the best option for sensorless speed detection for power tool embedded systems.