Other Parts Discussed in Thread: FDC1004
We are using the FDC2214 in level Sensing application(because of the high capacitance of our Level Sensor). Our conductive Tank is grounded, the level Sensor must be mounted close to the wall of that Tank. The tank there are 2 or 3 Sensors driven by the FDC2214. For testing the Sensor we have a small ungrounded Test stand, where we can decide where the conductive parts are connected to. Our Level Sensor is using a 18uH Inductor with a 33pF Tank capacitor.
If we connect the conductive parts of our test stand to the ground of the FDC2214 we get a capacitance over time diagram like shown in CapacitanceDropWithoutActiveShield.pdf . Here we are filling our Test stand with a constant flow rate which should correspond to a monotonous increasing capacitance. But what can be seen in the plot is a capacitance which decreases in a small time interval during increasing fluid level. We think that this results from common mode oscillations of the FDC2214. Is that assumption correct?
By adding a active shield for a single sensor which is exactly centered between the two active poles of the FDC2214 and which is connected to all conductive parts of our Test stand we don't have this effect anymore. But what we get now are Jumps in the capacitance these are quite low if the grounding wire is short enough. For demonstrating this effect we choose a long grounding wire which leads to very heavy jumps in capacitance shown in FillWithLongWireJumps.pdf . Even small jumps are unacceptable to us. We think that this results from the fact that the real oscillation circuit is not a ideal single pole one. The real oscillation circuit is having more than one eigenfrequencys, which is the result of parasitic wire inductance's. This fact leads to jumps in measured capacitance if the oscillator starts to oscillate on another frequency.
For us it is important to know if there are any solutions to the following problems.
- We are forced to connect the tank and all conductive parts electrically to our supply ground. Because of this we can't drive our shield active, like done in standard applications and our test stand. We considered isolating the I2C bus and Supply of the FDC2214 from our supply and driving the ground of the FDC2214 active. This solution would not be ideal to us because we would need a active multiplexer to switch the shielding to the active sensor on one FDC2214. Another solution for this would be to use output transformers for each sensor and drive one pole of each sensor actively, but we don't have any experience with that.
- Capacitance Jumps are unacceptable to us. We know that we can reduce that by improving our grounding shortening wires etc... , but we are forced to get this effect total away or below a error of 0.05%. Are we right that we can do this by using a lower excitation frequency by increasing the main inductance?
FillWithLongWireJumps.pdfCapacitanceDropWithoutActiveShield.pdf
