FDC1004EVM: Differential measurements, saturation problems

Update: I cut the copper tape electrodes to approx. one fourth of the original width, and this removed the saturation problem. However, the sensitivity is now really, really bad. The change in measured capacitance from empty to full container could only barely be seen. Is this to be expected?

Hello Lars,

SHLD1 and SHLD2 are sometimes shorted and sometimes 180 degrees out of phase. The latter is the case for differential measurements. You can read more about the two shield pins in section 8.3.1 in the datasheet. 

In differential measurements, the CAPDAC is disabled, so you will not be able to add a capacitive offset to the measurement. Your sensors are very large, and it sounds like the differential capacitance between them is greater than 15pF. It's also possible that the input capacitance for each channel is larger than 115pF. 

As for the sensitivity issues, sensitivity increases with the width of the sensors. You may not be able to achieve the sensitivity you need with such large sensors using the FDC1004. You can also try the FDC2114, which has a much larger input capacitance range. If you would still like to experiment with the FDC1004, section 4.4 of this TI Design discusses system sensitivity.

Best Regards,

Hi Kristin,

Thank you for the answer. Just to confirm that my understanding is correct here, can you please confirm/deny/clarify the following statements? Is it correct that:

1. Wider electrodes have greater a capacitance between them and thus provides higher sensitivity to changes in liquid level, but also higher chances of measurement saturation.
2. Longer electrodes will also give greater capacitance readings and thus higher chances of measurement saturation.
3. Wider space between the electrodes give lower capacitance readings, less sensitivity and lower chances of measurement saturation.

If so, it makes sense that the measurements may saturate given our very long electrodes. However, we have absolutely no saturation problems using the same electrodes in a single-ended measurement with some CAPDAC offset enabled. Does that make sense to you?

Given this, we may move forward with just using traditional, single-ended measurements.

Hi Lars,

Your understanding of the electrode sensitivity tradeoffs is correct. 

It sounds like the capacitance of each electrode may be less than 115pF, but the differential capacitance between the two electrodes may be larger than 15pF, which would still cause the measurement to saturate. Singe I'm sure you can't change the size of your container, using the single-ended measurements is your best option.

Best Regards,

Hi Kristin,

Thank you for your answer. We're now on to designing our first flex-PCB sensor prototype, for use with the FDC1004 and single-ended measurements. I have some questions regarding the reference electrodes:

1. See the attached sketch, which is not to scale when it comes to lengths and widths, but only meant to show the sensor layout. You can see that I have placed the liquid and environmental reference electrodes beneath and above the level sensor itself, in an effort to minimize the total width of the flex-PCB. The liquid level would always be above the height of the Liquid reference sensor. Is this an OK layout from your perspective, or is there anything wrong with putting all the sensors "on one line" like this?
2. I understand that the liquid and environmental sensors should be equally sized. Is there any rule of thumb for their minimum height?
3. Shielding: Would it work well to use one large shield covering the entire bottom layer of the PCB, or should I use six separate shield patches, covering each of the six electrodes?

Any input on this, Kristin?

Hi Lars,

My apologies for the delayed reply; I was out for the Thanksgiving holiday. I've answered your questions below:

1. I don't see a problem with having all of the sensors in one line. The only thing I'd recommend changing is the trace to the reference liquid electrode. This runs quite a long distance next to the GND electrode, so your reference capacitance may change as the liquid level rises. You could either route this trace on the other side of the level electrode or you could add a shield trace in between the reference liquid trace and the GND electrode.

2. I would keep the height at least 1cm.

3. You can use a single copper pour, but I suggest only shielding the reference and level electrodes, not the GND electrodes. The GND electrodes do not need to be shielded and will needlessly load the shield driver, which can only drive up to 400pF.

Best Regards,

Hi Kristin,

Thank you for good answers. One more question: Are there any best-practices regarding the gap between the Level and reference sensors (marked in red in the drawing below)?

You got anything on this, Kristin?