The increased systems complexity and performance request for electro-hydraulic applications, ask for more performing electronic systems and control functions. The new powerful microcontrollers and efficient cross compilers, encourage the floating point mathematics usage in the software control routines, useful to directly reuse the routines generated by the simulation tools, despite the lack of control for precise resulting routine execution. The paper describes a practical experience of system performance optimization on a microcontroller installed on electro-hydraulic systems for mobile applications. A deeper analysis carried out on execution time occupied by floating point mathematic operations, working on the software side of the mechatronic component, led to a considerably better performance. Here it is demonstrated that, without lack of precision, fixed point mathematics are more performing, if executed by modern microcontrollers, even if more instructions are executed by the software routines due to the necessary rescaling of factors needed by the requested precision.
Fixed Point versus Floating Point Mathematics in Embedded System Programming for Fluid Power Mechatronic Components Control: a Real Case Study
RUGGERI Massimiliano;MARTELLI Massimo;DIAN Massimo
2011
Abstract
The increased systems complexity and performance request for electro-hydraulic applications, ask for more performing electronic systems and control functions. The new powerful microcontrollers and efficient cross compilers, encourage the floating point mathematics usage in the software control routines, useful to directly reuse the routines generated by the simulation tools, despite the lack of control for precise resulting routine execution. The paper describes a practical experience of system performance optimization on a microcontroller installed on electro-hydraulic systems for mobile applications. A deeper analysis carried out on execution time occupied by floating point mathematic operations, working on the software side of the mechatronic component, led to a considerably better performance. Here it is demonstrated that, without lack of precision, fixed point mathematics are more performing, if executed by modern microcontrollers, even if more instructions are executed by the software routines due to the necessary rescaling of factors needed by the requested precision.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.