My DS3231 (chronodot) consistently reports time, but after running for 10-12 hours, it starts reporting a bogus temperature.
On cycling the power, it returns to a normal state.
I know the temperature is bogus because I have 3 other temperature sensors in the cabinet, plus cycling the power resets the temperature.
Has anyone seen this behavior, or have any hints on troubleshooting?
Here is a sample data set:
date time clocktemp LM35_temp
12/27/10 02:39:53 32.00 32.00
12/27/10 02:41:59 32.00 32.00
12/27/10 02:46:45 32.00 32.00
12/27/10 02:48:52 261.95 32.00
12/27/10 02:50:58 261.95 32.00
And here is the temperature reading function I'm using (off the internet)
float get3231Temp()
{
//temp registers (11h-12h) get updated automatically every 64s
Wire.beginTransmission(DS3231_I2C_ADDRESS);
Wire.send(0x11);
Wire.endTransmission();
Wire.requestFrom(DS3231_I2C_ADDRESS, 2);
if(Wire.available()) {
tMSB = Wire.receive(); //2's complement int portion
tLSB = Wire.receive(); //fraction portion
temp3231 = (tMSB & B01111111); //do 2's math on Tmsb
temp3231 += ( (tLSB >> 6) * 0.25 ); //only care about bits 7 & 8
}
else {
//oh noes, no data!
}
return temp3231;
}
temp3231 = (tMSB & B01111111); //do 2's math on Tmsb
temp3231 += ( (tLSB >> 6) * 0.25 ); //only care about bits 7 & 8
What is the type of tMSB and tLSB? - assume they are byte, why not declared local?
The math seems correct to me - although you drop the sign bit, but that is a choice - the value of the var temp3231 cannot be above 127.75. BUT you should give tmp3231 a value - e.g. 0.0 - in the else path. If the I2C communication drops somehow this is not noted and tmp3231 will get a random value depending on what's on the stack that moment.
SO two possible causes:
tampering of the value outside the function
I2C not returning a value.
Please share the whole code, as the problem might be outside the function.
I followed it, returning '999' when there is a problem reading the clock temperature. I also decreased the scope of 2 variables used in reading the clock temperature so that they are no longer global.
I cannot see why, but these 2 changes which you suggested have, prevented my bug from re-occuring.
The bug used to occur every 24 hours. Now it has been running 96 hours without any problem.
For those who may be following this thread for a solution to strange DS3231 temperature output below freezing, see this forum link for the solution (thanks to Coding Badly).
