16-bits fixed-point PID controller

This simulation illustrates the effects of the fixed-point
computations on the closed-loop precision for a 16-bits PID controller.
The rounding mode, the point where the truncations take place and the
format width for the internal variable have a significant action on the
closed-loop permanent error.

Three versions of the same PI(D=0) controller are
implemented : i) double (floating point), ii) 16 bits single
precision (fixed point) and iii) 16-bits double precision (fixed point).