I have to check if float number is negative or not. If negative assign it zero otherwise print it So what I did was:

float my_num;
/* do some calculation */

if( (int32_t)my_num < 0 )
{
    my_num = 0.0f;
}
else
{
   print_float_on_lcd(my_num);
}

The above code fails for float values less in range less than -0.9.
for example if number is -0.001, -0.0000006 then it fails . How to implement in C.

 

I would have used a union to get access to the individual bytes in the floating point number http://bytes.com/topic/c/answers/830745-converting-floating-point-number-byte-array-c Then just checked the floating point sign bit http://en.wikipedia.org/wiki/Floating_point

 

why not use

if (my_num < 0.0)

 

floats create further issues when compared....

 

You could cast the referenced location to a byte pointer and dereference it once you add the offset to the most-significant byte, and AND with 0x80 to determine if the sign bit is set...

This is probably what MrChips' suggestion does..

 

The problem with the first bit of code

if( (int32_t)my_num < 0 )

Is the cast is performed first... So the comparison is done against the 32bit long and not the float... MrChips's logic is the correct one...

 

If you refuse to use MrChip's excellent idea there is another simple fix:

float my_num;
/* do some calculation */

if( (int32_t)my_num <= 0 )
{
    my_num = 0.0f;
}
else
{
   print_float_on_lcd(my_num);
}