BQ27742-g1 floating point converstion

Also wondering which byte of the resulting byte array goes where in the flash. Does rawData[0] go to address 0 (ccGain) and 4 (ccDelta), or does rawData[3] go to address 0 and 4.

Hello Bruce,

Would you be able to provide the code for us to look at? Also giving the input/output data would be beneficial. I believe sample code is in the works, but do not know when it will be ready to post to the web.

The flow chart is starting with the values you obtained for CC Gain and CC Delta. Section A.8 describes how CC Gain and CC Delta valuse are obtained. Then, Section A.11 shows how to take the floating point CC Gain and CC Delta values and convert them into raw hex data.

Once you have the raw hex data you would store the CC Gain (4 bytes) and CC Delta (4 bytes) in the corresponding memory location (please see Data Flash Summary section).

I hope this helps.

Below is the code I wrote that I tried to follow the diagram with. I actually found from the sluuax0b document 2 sections of flash where they store a fp number (the ccGain and ccDelta sections) and read out the default values and compared them to my function when using those same default values. So at least for those 2 data points I match. The data points are the following:

CC Gain = 0.9536 = 0x80741F21
CC Delta = 11.19e5 = 0x950898C0

Also, do you have an algorithm to go from the gas gauge's floating point representation back to a floating point number? Would be another way for me to check my algorithm.

gasGaugeConvFP(float input, unsigned char* output)
{
int exp = 0;

float tmpVal = 0;
if (input < 0)
{
tmpVal = -input * (1 + powf(2.0f, -25.0f));
}
else
{
tmpVal = input * (1 + powf(2.0f, -25.0f));
}

if (tmpVal < 0.5f)
{
while (tmpVal < 0.5)
{
tmpVal *= 2.0f;
exp -= 1;
}
}
else if (tmpVal > 1.0f)
{
while (tmpVal > 1.0)
{
tmpVal /= 2.0f;
exp += 1;
}
}

if (exp > 127)
{
exp = 127;
}
else if (exp < -128)
{
exp = -128;
}

tmpVal = powf(2.0, 8.0f - (float)exp) * input - 128.0;

int byte2 = (int)tmpVal;

tmpVal = powf(2.0, 8.0) * (tmpVal - (float)byte2);

int byte1 = (int)tmpVal;

tmpVal = powf(2.0, 8.0) * (tmpVal - (float)byte1);

int byte0 = (int)tmpVal;

if (input < 0)
{
byte2 |= 0x80;
}

output[0] = (unsigned char)(exp + 128);
output[1] = (unsigned char)byte2;
output[2] = (unsigned char)byte1;
output[3] = (unsigned char)byte0;
}

Hello Bruce,

Your values look correct to me.  The only difference I see in the code is that the if-statement checking if tmpVal is > 1.0f should have greater than or equal as the test condition and the same goes for the while condition in that if statement.

if(tmpVal >= 1.0f)

{

while(tmpVal >= 1.0f)

{

...

I am not aware of an algorithm to go from the raw data to the floating point information.  But, you can use the "Advanced Comm" plug-in in bqStudio in order to read back the raw data.  That lets you check what you come up with against what our tools produce.

I hope this helps.