Calculation for temperature false

[Inspectron]

In the data sheet of SHT21 is this the formula to calculate the temperature:

T = -46.85 + 175.72 * temp_data / 2^16

temp_data has the value 26'464.

The result should be: T = 24.1°C

But in my code I get this result = 18052.3105

Code:

float temperature;

temperature = (-46.85 + (175.72*(temperature_data / 65536)));

What is wrong with my calculation for the temperature?

[MartinBL]

Hi. I guess you have declared the variable temperature_data as an Int? You will need to cast it to a float like this:

temperature = (-46.85 + (175.72*( (float) temperature_data / 65536)));

This is because if you don't do the casting the '/' operator (correctly) assumes that both values are integers and that you want integer as result as well. If (at least) one of the variables is of type float (or double) the compiler will produce a float as a result instead. A bit more detailed answer here.

[Daniel Tralamazza]

We use basically the same sensor here, take a look at the code github.com/.../htu21.c

ps: Try to avoid floats.

[RK]

That doesn't make sense however. If temperature_data is an int, or integral type, and really has the value '26,464', then the division would be done as integer division, the result would be zero (as 26464 < 65536) and the entire expression would then evaluate to -46.85, which it doesn't, according to the original poster.

If you work backwards, the only way you get 18052.3105 is if the initial value is 26464 x 256. At that point yes the integer division rounds off the result to 103 (instead of 103.375) and the value would then work out to 18052.3105.

So it seems to me the OP has temp_data wrong by a factor of 256. When they correct that then I will agree that they will have integer division issues with their code as-written.

edit: and I'm also totally in agreement with the 'avoid floats' comment. You can do without them here with a little bit of judicious scaling.

[MartinBL]

Oh, yes you are right. I was a little too fast to my conclusions.

[Inspectron]

@RK yes, you are right. The problem was that I did wrong shift the Data. I changed it from <<16 to <<8 and now I get results which make sense. Thank you