Converting to scaled fixed-point control

The underlying principle in converting an existing floating-point subsystem into an equivalent scaled fixed-point system is that every input, operation, and output be configured to reflect the realities of the target. Fundamental details — such as whether the target is an 8-, 12-, 16-, 24- or 32-bit processor — become important. As you traverse the controller implementation, from left to right, you encounter several operations, including summation, sample-and-hold, absolute value, sign, multiplication, constant, division, gain, and unit delay. Using Fixed Point blocks, each of these operations is replaced with the equivalent fixed-point operator, assuming a 16-bit target.

Blocks such as unitDelay and sampleHold are fixed-point-aware; that is, they automatically adjust to the incoming data type. Consequently, they can be used as is.