I have a heating system with 10kohm floor temperature sensor cable connected to controller unit. The controller displays floor temperature around 7.0 degC higher than reality, the sensor is installed in a channel inside the floor above insulation and below floor boards (directly between heating elements). I do not believe (although cannot confirm) that the sensor head location is compromised in any way, I can withdraw the sensor from conduit and leave it exposed in the room above floor and the display temperature remains the same. The sensor cable is installed in its complete calibrated length. It is the right sensor resistance for the controller, yet the 7.0 degC discrepancy persists. I've tried swapping controller and replacing sensor. I have adjusted the heating set point upwards to account for the discrepancy; the heating system modulates effectively and the relative temperature tracks correctly, but it annoys me that the controller displays heating at 28 degC when it's really around 21degC (confirmed by thermostat in lounge room). Any ideas what I can test or try next?
Electrical – How to correct/adjust floor temperature sensor resistance to account for calibration error
calibrationsensortemperature
Related Solutions
Physicist's answer ahead:
Prepare a bowl of (melting) ice water on a day where the barometer is near 1013 mbar of pressure. Stick your sensor in it. The reading you get at equilibrium is your 0C reading.
Prepare a pan of slowly boiling water, also on a day where the barometer is near 1013 mbar of pressure. Stick your sensor in it. The reading you get at equilibrium is your 100C reading.
Divide the intervening range into 100 equal parts.
The beauty of this method is that you are not tied to errors made while calibrating the sensor you are using to calibrate with, or errors made when calibrating the sensor they used to calibrate the sensor you are using to calibrate with, or ... (etc).
Diodes are not as good as diode-connected transistors in this kind of multi-current \$\Delta V_{BE}\$ temperature measurement circuit. I note that the part number you specify is actually a BJT, not a diode.
Recall the ideality factor that is close to 1 for a diode-connected transistor and more like 2 for a diode.
The manufacturer also has detailed recommendations for the type of transistor to be used, and recommends that large power transistors not be used.
You could try a small power transistor such as a BD159STU (That one is in the convenient TO-126 through-hole package), or simply make a small PCB with the SMT part on one side and no tracks or vias on the other, and attach that to the panel (that's probably what I'd do- it could be connectorized, and made as a snap-off bit attached to some other board).
Best Answer
7°C is a huge error- typically sensors like that are good to a fraction of a degree C (better than most semiconductor temperature sensors).
The 10K resistance is typically specified at 25°C. If you have a "Beta" spec you can look up the temperature-resistance chart and find the expected resistance at 21°C. For example a beta 25/50 of 3380K (a common type) would have a resistance of 10K at 25°C and about 11.67K at 21°C.
Once you know the expected temperature-resistance curve you can determine where the error lies. My bet would be with the controller, however such a huge error (reading resistance of about 8.9K in this case rather than 11.7K) would imply either something is seriously wrong with the controller or there is some configuration setting that you need to make. Maybe look for a service manual for the controller or contact the supplier.
Just to be absolutely clear on this- with such a large error (assuming no hidden configuration setting you've possibly missed), there's a good chance it's about to fail completely and without further notice.