Numerical considerations to implement a control system on FPGA

In this work the numerical considerations to arise when making a control system on FPGA are presented. It takes as study case the implementation of the position controller of a manipulator robot. The system is made using two numerical representations: integer and fixed-point. It shows the procedure to define the width in bits of the integer and fractional part of the fixed-point number. In addition, the impact of different types of rounding is analyzed. It concludes that, if there is not frequency requirement, to opt by HDL-Coder is a good option as a result of its usability and low area consumption, finding the best performance for the rounding types: floor and simplest. Further, the optimization shows that a synthesis strategy that prioritizes the area gives the maximum frequency, even it be unintuitive.