Question

I got a Problem. I’ve got a MODBUS TCP answer, which is hex coded like this: 0 0 0 0 0 7 1 4 4 41 B8 66 64. For explanation: the first five zeroes are a specification of Modbus, the 7 are the numbers of bytes following after it. the 1 is the client address in the Modbus network and not relevant. the first 4 is the function code which is used. the second 4 is again the numbers of bytes following. The last four bytes is the hex coded answer, which should be converted to double. The array, in which this is stored is a unsigned char array.
Here a few examples how i tried it.
Here the first example:

value = (ibuf[9]<<24) + (ibuf[10]<<16) + (ibuf[11]<<8) + ibuf[12];

Value is the used double variable and ibuf the used char array.
Here the second one:

for(i = 0; i < k; i++)
{
    if (i==0)
    {
        sprintf(ergebnis, "%x%x", ibuf[9], ibuf[10]);
    }
    else
    {
        //sprintf(buffer,"%x%x",ibuf[9+i+i], ibuf[10+i+i]);
        //strcat( ergebnis, buffer );
        sprintf(ergebnis, "0x41b451e8");
        sscanf(ergebnis, "%l %lf ", &Value);
    }
    printf("Ergebnis %s\n", ergebnis);
}

Here I used a fixed value, but the problem is always the conversion from hex to double.
I’m glad about every help I could get.

Answer 1

You’re almost there. Unfortunately, conversion occurs automatically during promotion – the most “straightforward” way of circumventing this is through dereferencing a float pointer cast of an int, as so:

int ival = (ibuf[9]<<24)+(ibuf[10]<<16)+(ibuf[11]<<8)+ibuf[12];
float fval = *(float*)&ival;

For the data supplied, this gives 23.049995 instead of 1.1026039e+009, which you get from conversion.

EDIT:

As Mike Seymour pointed out in the comment below, a preferred way to write this is:

float fval = *reinterpret_cast<float*>(&ival);

Note that you cannot do this:

float fval = reinterpret_cast<float>(ival);

This will raise an error (as seen here in VS2005):

error C2440: 'reinterpret_cast' : cannot convert from 'int' to 'float'