S12 / MagniV MicrocontrollersのトピックRe: Fraction data convert to Float

Fraction data convert to Float

ahnjoseph — Tue, 05 Apr 2016 10:36:00 GMT

Fraction data convert to Float

I have result fraction data format with below Math_Lib

But that is the data type difficult show to me

So, i want to show "0x10000000(FRAC32)" --> "0.125"

How can i do?

#include "mlib.h"

tFrac32 f32In1;

tFrac32 f32In2;

tFrac32 f32Out;

Function MLIB_Mul_F32

Automotive Math and Motor Control Library Set for Carcassonne MC9S12ZVM devices, Rev. 8

450 Freescale Semiconductor, Inc.

void main(void)

{

// first input = 0.5

f32In1 = FRAC32 (0.5);

// second input = 0.25

f32In2 = FRAC32 (0.25);

// output should be 0x10000000 = FRAC32(0.125)

f32Out = MLIB_Mul_F32(f32In1,f32In2);

// output should be 0x10000000 = FRAC32(0.125)

f32Out = MLIB_Mul (f32In1,f32In2,F32);

// ##############################################################

// Available only if 32-bit fractional implementation selected

// as default

// ##############################################################

// output should be 0x10000000 = FRAC32(0.125)

f32Out = MLIB_Mul (f32In1,f32In2);

Re: Fraction data convert to Float

kef2 — Tue, 05 Apr 2016 13:34:37 GMT

0x10000000 / 0x80000000 = 0.125, so for unsigned fractions conversion could look like this

float f = (float)x / 0x80000000 ;

Signed:

if( x & 0x8000000) // is it negative

f = - ( (float)(0-x) / 0x80000000 );

else

f = (float)x / 0x80000000 ;

Re: Fraction data convert to Float

RadekS — Thu, 07 Apr 2016 09:23:55 GMT

Hi Ahn,

As correctly mentioned by Edward:

We should divide 32-bit fixed point number in format Q1.31 by 0x80000000 for getting signed fractional [-1,1) number.

FRAC32() is defined in SWLIBS_Defines.h file as:

/*! Macro converting a signed fractional [-1,1) number into a 32-bit fixed point number in format Q1.31.*/
#define FRAC32(x) ((tFrac32) ((x < FRACT_MAX) ? ((x >= FRACT_MIN) ? (x*2147483648.0) : INT32_MIN) : INT32_MAX))

Note: 2147483648=0x80000000

Since tFrac32 type = tS32 type = signed long type (according SWLIBS_Typedefs.h), we do not need to test sign in this case and command

float f = (float)x / 0x80000000;

should work correctly also for negative numbers.

Code

if( x & 0x8000000) // is it negative
    f = - ( (float)(0-x) / 0x80000000 );
else
    f = (float)x / 0x80000000 ;

is necessary when input data x are unsigned type.

I hope it helps you.

Have a great day,
RadekS

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------