Signal processing techniques for improved digital tachometry

The application of digital tachometers to velocity estimation in open-loop measurement and closed-loop motion control systems is examined. The very significant impact on performance of the quantization and nonideality of the discretely sampled position sensor signals that typically provide the tachometer input is highlighted. A number of methods by which the influence of sensor nonideality can be reduced are described. The performances of the following are compared: a simple pulse-count tachometer, an oversampled digital differentiator, a constant sample-time digital tachometer (equivalent to an M/T-type tachometer), with or without parallel edge measurement, and a tachometer based upon a sinusoidal encoder, with or without probabilistic-compensation. The advantages and limitations of each method are demonstrated. The paper concentrates on systems that rely solely upon the information available from the position sensor, rather than on model-based observer structures (within which the tachometers can be incorporated, if desired). Practical implementation using field-programmable gate-array (FPGA) circuitry and DSP-based algorithms is described. The paper illustrates the improved performance that can be achieved in digital tachometers through enhanced circuitry and algorithms.