I am quantifying sensor noise and I need to remove the effects of temperature. The sensor is highly temperature sensitive and the variation in ambient from 72F to 76F causes the sensor signal to vary. I also have thermocouple data for temp.
So how to I remove the variation in temperature (thermocouple readings from ~72F to ~76F) from sensor data that is unfortunately varying with temperature (from -0.070V to -0.012V)?
Extra info: The sensor is not a temp sensor, but is highly temp sensitive and I just want to analyze the noise floor. The temp causes it to vary beyond its noise floor so I need to remove those effects. The thermocouple and sensor are being read from the same DAQ device and sampled simultaneously.
I tried simply subtracting a moving average from the data but depending on the window size I get different results. If that is the answer, what is the acceptable window size?
Is there a way to scale the temp to be able to calibrate away that effect on the data. It's effect is anticorrelated? to the data. Temp goes up, sensor signal goes down.
You need to make calibration measurements wish gives you the influence of the temperature. For a constant signal, you vary the temperature in small increments and note your readings. Doing so you get a curve. For your data linearly. Then you get a zero point offset, and a linear temperature-related component. If you want you can also or instead fit quadratic or exponential - and see if they fit better. Preferably repeat that for different signals strengths, and verify the findings