I am referencing the datasheet for the FDC2214 and in figure 56 it shows an example liquid level measurement.
It is drawn single ended,the sensors are surrounded by ground. I was planning on implementing a differential solution as I have roughly sketched and attached to this post.
From some simple experiments with the EVM, the differential input does not seem to offer any significant rejection. If you place two capacitors to gnd, one off each side of the differential pair, the measured value is the same as a single capacitor across the pair. A single capacitor placed on one side of the pair is about 10 dB down from the differential result.
My question is what is the benefit of the differential input if the rejection is so small? Perhaps I have missed the point or intent of the differential output? In order to take advantage of the differential sensor, I would minimize the capacitance between the traces on my layout, then only allow the differential sensor to provide the capacitance between the pair through the intended measured dielectric. Unfortunately, the influence of gnd to each trace will be hard to manage. My plan was to place the components directly on the sensing flex PCB with gnd as shielding. Running the traces to the top and bottom of the sensor for the enviro and reference pads would develop an offset capacitance due to the coupling to ground. This offset may not be an issue, but I would like to minimize it as environmental effects could change the offset value.
Is this a property of the internal design of the FDC that limits this? Is there anything I can do to improve it?
