Capacitive-based liquid level Sensing with FDC1004Q

Hi Lorenzo,

Thank you for the post. I have seen this before in a couple of the prototypes I set up. It's not expected behavior, and in our case, was not caused by condensation.

Would you mind describing what your current setup is like? Are you using copper tape for your sensors?

I saw this with my copper tape demos, to resolve it we redid all of our solder joints and used twisted the wires for the sensor and shield to reduce the parasitics. After this our signal settled well within a second. (each sensor and shield had it's own twisted pair).

Hope this helps! Let us know if this doesn't fix everything,

Nate

Hi Nathaniel,

the wires are about 4 or 5 cm long. The "tank" is small vial with a height of about 3 cm, and a capacity of 5 milliliters.In this figure you can see a section of the holder that houses the electrodes and the wires, that plug in the PCB just outside the wood.

Even if the system is very small, the sensitivity of the sensor is really good (more than 1pF/ml) and it would be more than enough for my purpose, if it weren't for the strange behaviours I described before.

Now I followed your advice, and made a simpler setup with aluminium tape electrodes, and still I am experiencing the same behaviour (plus, of course, more noise).

The errors are still more noticeable during the emptying phase, as they take several minutes to settle to the right value, but I have sometimes experienced drifts of some fF/sec also in tests without any liquid. 
I am still not sure if these two problems are of the same electrical source or if the water on the vial wall is really playing a role in this.

Thanks for kind attention,

Lorenzo

Hi Yibo,

thanks for your attention.

I haven't been able to work much on the device in the last few weeks, so yes, the problem is still ongoing. However, I have run several tests on different conditions, and I am now more inclined to think of an electrical problem. Now I have noticed that there are also minor rising drifts during the filling procedures. Plus, if I slower the sampling frequency of the FDC1004Q, the time constants seem to increase.

I know my system is very little if compared to the one showed in you application note (the height of the liquid in the full vial is about 1.5cm), but the sensitivity of the sensor is good enough(about 4pF/cm) and the noises are neglectable. If it weren't for those slow dynamics the measures would be more than suitable for a 5% or 10% level accuracy on full scale.

I still have to try to build a new holding setup, just to see if anything changes, but I won't be doing it shortly, because it would take a lot of time and I am alone in this.

Regards,

Lorenzo Ventura

Hi Lorenzo:

I have seen this similar behavior as I have worked with Nate on this, but I have not experienced this in any of the prototypes that I have done. I would suggest you to make sure you have a good ground connection and stable setup, and try taking measurements in a different environment. I would also suggest trying another EVM or part to try determine if it's the part itself that has some abnormalities with it.

The behavior is very similar to having some kind of large parasitic capacitance holding extra charge and discharging very slowly, since the FDC1004 utilizes a switched cap circuit, this extra charge is being picked up and transferred back to the converter. Just a thought, but have not been verified. Capacitive sensing is very hard to implement correctly because environmental factors can have a significant impact on reliability and repeatability.
 
Thanks
-David

Hi David,

I am quite sure it is not a problem of the part itself, as I have experienced the same behaviour both with you EVM and with the FDC1004 sensor mounted on our own PCB.

David Wang60 said:

The behavior is very similar to having some kind of large parasitic capacitance holding extra charge and discharging very slowly, since the FDC1004 utilizes a switched cap circuit, this extra charge is being picked up and transferred back to the converter.

Maybe I'm wrong, but this could explain why the problem worsen if I slow down the sapling frequency?

Anyways, I am now working on a new holding setup, and I will let you know the results as soon as I can.

Thank you for your kind reply,

Lorenzo Ventura

Hi Lorenzo,

Thank you for the update.

Based on your observation and the input from Nate and David, I think parasitic capacitance is most likely the cause of the slow settling. Since your "tank" is much smaller compared to many other systems, even a tiny parasitic capacitance may cause a very pronounced effect.

Regards,
Yibo