FDC2214: Remote Sensing for insulated metal container

Vince,

It is technically possible to reproduce FDC1004 applications with the FDC2x1x family of parts.

One major difference between the FDC1004 and the FDC2x1x family is the sensor waveform , and the other is the active shield drivers.
The two factors are related.

The FDC1004 uses a fixed 25kHz sensor waveform that consists of a DC voltage offset and two pulses, one positive and one negative (slightly delayed in time) about the voltage offset. The low frequency of the sensor waveform makes the active shields more effective since time delays between the sensors and shields due to he implementation and don't have as great an impact compared to a higher-frequency waveform. The reason: for the active shields to be effective, their waveforms need to match the sensor signal as closely as possible. If they don't, capacitive loading between the sensors and shields will degrade the shield's effectiveness and degrade system sensitivity.

The FDC2x1x uses a resonant LC circuit as the sensor, so the waveform is sinusoidal. The lack of on-board shield drivers means external drivers will be needed, and these will need to be relatively high-powered, high-bandwidth op-amps to be able to cope with highly capacitive shields and keep the sensor-to-shield time delays small. 
The higher frequencies of sensor waveform will add additional challenges to the system because of sensitivity to relative amplitude imbalances and time delays between the sensor and shields. 

The app note Capacitive Sensing: Out-of-Phase Liquid Level Technique describes a technique for implementing the sensors so the liquid being measured is electrically neutral, making it less sensitive to external loading from a nearby conductive object. The FDC1004 can implement this easily, and the app note gives guidance on how to do this.
It may be possible to implement tis approach with the FDC2x1x, but the challenges will be far greater, as mentioned above. 

If your container will be conductive, you may be able to use active shields to reduce the capacitive loading between the sensors and the container walls.
This would help your system sensitivity and up-front calibration. The app note Capacitive Sensing: Ins and Outs of Active Shielding gives background on this approach.

If you would like more info on the FC1004, the E2E FDC1004 Frequently Asked Questions gives a comprehensive list of collateral for this device.

Regards,
John.

Thank you for the detailed response - I understand a lot more now, but am still fuzzy on a few points.

I appreciate now that the FDC2x1x is a much more complex device to implement, which turns my attention back to the FDC1004. In another post you mentioned the 1004 is a bad fit for conductive containers, but as I learn more it sounds like that is a function of being more susceptible to loading by other things which could produce unpredictable values. If we take care to time the measurements and filter erroneous values due to temporary external loading, will the 1004 still be able to work through an insulated, conductive container in principle?

To make sure I'm interpreting your advice & the FAQ correctly, it sounds like I should first test the FDC1004 in a differential mode. We should definitely use the 3 main reference, level and environmental sensors - but it also sounds like there may be value to an additional proximity sensor facing the user so we can discard erroneous readings more easily. (FAQ #4, but this may be more focused on remote presence detection and not liquid sensing...)

This is great, thanks - I think I have enough to get started.