HDC2010: Temperature measurement with Single digit after decimal point

The source code for HDC2010METER-EVM might be useful. It's available here: http://www.ti.com/tool/HDC2010METER-EVM

Specifically, tmp-decode.c is an example of converting the two byte response from HDC2010 to humidity in string data type with one digit of precision. 

char* humstring(unsigned char HiByte, unsigned char LoByte){
    long ih;
    float fh;

    //combine 2 bytes
    ih = (long) (HiByte & 0xFF) << 8 | (LoByte & 0xFF);
    fh = (float) ih / 65536;
    fh *= 100;
    sprintf(sh, "%6.1f", fh);
    return sh;
}

I had originally written a version that avoided the use of float data type as well as printf library. It was not pretty. It was scrapped because it actually used more CPU cycles than the above code with float & printf.

char* humstring(unsigned char HiByte, unsigned char LoByte){
    char i;
    unsigned long it;

    //combine 2 bytes
    it = (unsigned long) HiByte << 8 | (LoByte & 0xFF);
    //LSB
        it *= 15258; //2^-16*100*1e7
        sh[st_char_len-1] = '%';
    #ifdef DEBUG
    printf("it: %lli i: %i\n",it,i);
    #endif
    //sign
//    if (it<0) {
//        sh[0] = 0x2D;
//        it *= -1;
//    }
//    else
    sh[0] = 0x20;
    //decimal
    sh[4] = 0x2E;
    for(i=st_char_len-2;i > 0;i--){
        #ifdef DEBUG
        printf("it: %lli i: %i\n",it,i);
        #endif
        if (i!=4){
            sh[i] = it % 10 + 0x30;
            it /= 10;
        }
    }
    //fix leading zeroes
    for (i=1; i < 3; i++){
        //if equals zero
        if (sh[i]==0x30) {
            //if previous equals minus
            if (sh[i-1]==0x2D) {
                sh[i] = 0x2D;
                sh[i-1] = 0x20;
            }
            //else if previous equals space
            else if (sh[i-1]==0x20)
            sh[i] = 0x20;
        }
        else break;
    }
    return sh;
}

Ren

Hello,

Thank you for your inputs. I have gone through the example code. Let me give more details about my requirement.

1) We want to store Temperature & Humidity data into Non-Volatile memory (Ex: Flash). We have a very stringent requirement for Flash space remaining.

2) So we don't want to store converted values in float format into Flash which will take 4 bytes for Temperature and 4 bytes for Humidity.

3) With respect to Temperature we want to store only 11-bits (Take MSB 11-bits and then right *** by 5) of register data into Flash. For Humidity only 8-bits (Take MSB 8 bits and then right shift by 8). With this will be using only 19 bits (11-bits for Temperature and 8-bits for Humidity) into Flash rather than 32-bit (16-bit register values of Temperature and Humidity).

4) While communicating the Sensor values, read those values from Flash and then apply below formula and transmit over communication port.

Formula:

Temperature :     (((MSB 11-bit temperature value + 5 LSBit as 0) / 2 ^ 11) *165) - 40;

                           Where 2 ^ 11 is 2048

Humidity: ((MSB 8-bit Humidity value + 8 LSBit as 0) / 2 ^ 8) *100;

                           Where 2 ^ 8 is 256

Is above approach is fine or is there any alternate and better approach.            

Thanks & Regards

Vishnu Beema