CODEWITHSUNDEEP
cfixedlong-integer
Matt
Matt

Reputation: 11

converting fixed point number stored as unsigned long to float

I read signed fractional value from sensor in q16.4 format and receive this value through i2c bus as unsigned long. I want to convert this value to float to perform algorithm. I was trying to convert it to float like this:

float vOut = (float)read_data * 65536;

with no result. Data look like is still unsigned long with point. Can you give my advice what will be the best way to convert it correctly?

Upvotes: 0

Views: 607

Answers (2)

chux
chux

Reputation: 153457

Sorry, not more detail GTG.

uint8_t tab[3];
Read_sensor(tab);
unsigned long ul = tab[0];
ul <<= 8; 
ul |= tab[1];
ul <<= 4; 
ul |= tab[2] & 0xF;

// I am fuzzy on q16.4  format, so some assumptions.
// Assuming Bit 19 is the sign bit and data is sign-magnitude
float x = ul & 0x7FFFFL;
x /= 16.0f;
if (ul & 0x80000L) {
  x = -x;
}

Upvotes: 1

ooga
ooga

Reputation: 15501

Try:

float vOut = read_data;
vOut /= 16.0f;

Dividing by 16 pushes the four fractional bits of the fixed point number into the fractional bits of the float.

Upvotes: 3

Related Questions