temperature compensated capacitance measurement

[hemnath]

Hi,

I need to measure a tank level of cold stored liquefied gases like nitrogen using capacitance measurement. I checked that the capacitance measurement is sensitive to temperature. Is it possible to compensation network or any other suggestions will be appreciated.

Earlier, I built a circuit for capacitance to frequency converter using relaxation oscillator circuit.

Thanks.

 

[stenzer]

Hi,

usually capacitive measurements are sensitive to moisture, where the relative humidity and thus the moisture in the air depends on the temperature. But this is mostly an issue for the sensor itself and depends on your sensor structure e.g. if interdigital electrodes are used.

Regarding the influence on your measurement electronics, this depends on the used circuitry. By using a differential measurement approach including a refernce measurment, the temperature dependency might be compensated for. You might also use a CDC with on-chip calibration e.g. from ADI [1].

Earlier, I built a circuit for capacitance to frequency converter using relaxation oscillator circuit.

Please share a schematic of your circuitry, so we are able to estimate how it might behave at different temperatures.

[1] https://www.analog.com/en/analog-di...-digital-converter-technology-healthcare.html

BR

 

[hemnath]

Stainless steel rods are used as sensor. Tank is made of SS.
One rod will go through the tank as a signal. Another rod is connected to tank wall as a reference.

Existing circuit:
https://pasteboard.co/K0KyEBQ.png

 

[FvM]

It's not clear where the temperature dependency is located, in the oscillator circuit or in the sensor capacitance. First step would be to measure the oscillator circuit temperature dependency in a thermostat chamber. These days, compensation is mostly done numerically based on frequency and temperature readings.

Temperature dependency of the sensor itself can be compensated in a similar way, an additional temperature measurement at the sensor is required.

 

[danadakk]

In addition to compensation via least squares or power curve fit, off a reference setup, eg. an
algorithm to interpret the T dependency. One could consider using a separate sensor, known
geometry, as a reference. That way take geometry out of the issues, etc,,, might be
also compensated ? Not an expert here, just a thought.

Regards, Dana.