C Arithmetic Different Output

Question

I'm getting different results when doing what I think is the same calculation. I imagine it's a type conversion issue.

I'm pulling a temperature value from a hardware sensor via I2C. The datasheet says the equation to do the conversion is as follows:

(((14_bit_ADC_value)/16382) * 165) - 40)

The two byte ADC value is stored in two chars and some shifting needs to occur to get the 14 bytes as such:

((((rawtemp[2] << 6 ) | (rawtemp[3] >> 2) / 16382) * 165) - 40);

The problem:

• If I break up the operations it works.
• If I combine them into one line it fails.

My Code is:

#include <stdio.h>
int main()
{
  char rawtemp[4];
  rawtemp[2] = 0x61; //example captured value
  rawtemp[3] = 0x40; //example captured value

  //method 1
  float test = 0;
  test = (rawtemp[2] << 6 ) | (rawtemp[3] >> 2);
  test = test / 16382;
  test = test * 165;
  test = test - 40;
  printf("test1: %f C\r\n", test);  //This works.

  //method 2
  test = (((rawtemp[2] << 6 ) | (rawtemp[3] >> 2) / 16382) * 165) - 40;
  printf("test2: %f C\r\n", test);  //This does not work.
  return 0;
}

Output:

$ ./k 
test1: 22.688316 C
test2: 1024280.000000 C

Why does the second method to derive the temperature value fail?

Answer 1

There are two bugs in your method 2:

1. (rawtemp[2] << 6 ) | (rawtemp[3] >> 2) / 16382 should be ((rawtemp[2] << 6 ) | (rawtemp[3] >> 2)) / 16382 , because the priority of bitwise OR is lower than divide.

2. In C, if the two operands of '/' are both int, the result is also int, or floor(a/b), you can use (float)((rawtemp[2] << 6 ) | (rawtemp[3] >> 2)) / 16382 (or write 16382 as 16382.0) to convert one of the operand into float, then the result will also be int.