FDC1004QEVM: with TIDA-00317

Chad,

Thanks for the post and your interest in TI products.
Sorry to hear you are having problems with your set-up. 

Some questions that will hopefully lead us to the root cause:
1. What is your container material? Is it at all conductive?
2. Is your sensor the same as the one in TIDA-00317?
3. If not, what are the nominal (designed) capacitances of your sensors with and without your container?
4. Depending on the answer to (3), would be possible to experiment with different values of the Offset Capacitance setting to see if it affects the Capacitance reading?
5. Can you confirm the connections between your sensors and the EVM are all low resistance?
    This sounds trivial, but in past work, the behavior you are seeing has sometimes been due to bad solder joints and/or poor wire harness connections. 
6. Can you experiment with your EVM and an alternate set of sensors to see if the EVM & GUI can read & report varying capacitances? 
    This step could help identify a problem with the EVM. 

Regards,
John

John,

My answers, to the best of my knowledge, are below:

1. We tried a couple of containers. The one in the photo is High Density Polyethylene. The other was just a generic, transparent plastic from a coffee maker water reservoir. Neither are conductive.
2. We took the gerbers straight from the TIDA-00317. Though we do have a black polyamide solder mask coating the entire cable. I'm very curious if you think this could have an effect?
3. Should match.
4. It is possible, though I could use some guidance one where to start and what to look for.
5. We bought a 10pin 50mil header to make the connections as clean as possible. Looking under a microscope, we could not find any poor connections or shorts. (Doesn't mean one doesn't exist, but want to eliminate other potential faults first)
6. Do you know of a source where we could purchase some known good cables? We could manufacture some new ones but the delay to receive is enough to cause issues with our timeline. If not, we can create new cables.

Thank you for your attention on this.

Chad

Chad,

Just to confirm, are the readings in your table in pF?

Also, are you driving an active shield, and is it possible for you to probe & compare the waveforms on the sensor and the EVM?

Would it be possible to confirm the capacitances of your sensor by other means (cap meter, etc.) just to confirm they are as expected? 

(4) I mis-spoke earlier w/respect to item 4.
I was wondering if adjusting the CAPDAC would cause the capacitance reading to change by the appropriate amount. CAPDAC normally serves as an offset for reading capacitances that are beyond ±15pF. You can change the CAPDAC in the EVM GUI by entering a number in the Code field. If the system is working properly, changing the CAPDAC capacitance slightly should cause the measured capacitance to change by about the same amount. Section 8.3.2 of the data sheet gives some background.

Unfortunately I don't have any special recommendations for cable vendors. The usual major distributors would be our go-tos for that.

Regards,
John

John,

Yes, I confirmed the tables are displayed in pF.

I'm not exactly sure what you mean regarding the active shield. We're using the TIDA-00317 design and the EVM, would that mean it's an active shield? In order to measure the waveforms on the cable sensor, we'd have to scrape through the solder mask, which I fear might effect our results. Do you recommend this?

We might be able to, but our meter may not be able to measure in the single digits of pFs.

Gotcha. I will look deeper into the CAPDAC and give that a shot.

Thanks,
Chad

I configured the CAPDAC on CIN1. From empty to full I'm still only getting a difference of about 1 pF. Below is a screen shot of the table in case you see any issues with my setup:

Chad

Thanks Chad. Today is a holiday for us and I will follow up with you tomorrow.

Chad,

The active shield connections I mentioned previously are described in section 4.3-Sensor Layout of the TIDU736A Reference Design document.
I have thought about your observations a lot, and it keeps coming back to confirming the sensors and shields are getting the right signals. 
Is there any way you can confirm your connections and measure the sensor and shield waveforms without harming your assembly?
If not, please let me know and we can try figuring out something else.
Regards,
John

I don't believe so. the solder mask covers the entire cable so we would have to remove in order to measure those waveforms. We've contacted our regional rep in the hopes that he has sample cables we can test with.

But we did order another devkit, and have a few cable assemblies to try with different kits. This should show us if something was damaged during our testing. Is there any difference between the FDC1004QEVM and the FDC1004EVM?

Thanks,
Chad

With a new dev kit, cable, and clean connections. We're still seeing a difference of less than 1 pF from 1-8 cm. We're going to try to measure the wave forms next week. Do you have any other thoughts on how we could source a cable that will eliminate any potential differences between the cable, or should that even matter?

Chad,

At this point, I'm not sure what role the cable is playing.
I believe the next step should be to probe the entire signal chain (e.g. cable, connectors and device pins) to confirm the waveforms are as expected.

Regards,
John

John,

We will work on getting that setup. Should we be measuring with reference to the shield cables or just GND? Demo guide says to configure differential measurements for CIN1-CIN4, should we measure waveforms in reference to CIN4?

Also, what should we expect the waveforms to look like? Do you have any reference material to compare to?

Thanks,
Chad

Chad,

I was sure we had a measured waveform example in the EVM User's Guide or at least one of our app notes.
Alas, I'm a little embarrassed to admit I couldn't find any..
The quickest thing I could dig up are the voltages listed in the data sheet tables for the excitation.
If you could confirm these in your system:

Of course the Sensor voltage measurements will be important as well.
If you could capture those time-domain waveforms with enough resolution to show DC and Vpp behavior, that would be great as well.

Regards,
John

John,

It's been a while, we've been focusing on other items. I recently came back to it and tested with an acrylic container and a new sensing cable without the black overlay. Using 2-sided tape, we had a MUCH better attachment to the container. This setup worked MUCH better. Results were consistent, up and down. 

I went back to the old container (an HDPE plastic container) and the results were not good. What I've been seeing for other tests: the capacitance would increase consistently while filling up, but would not decrease consistently when draining. We believe the container may be dirty, softer, more textured, causing a watery-residue to cling to the sides of the container leading to inconsistent results. 

Does that sound plausible? Are you aware of anyone else experiencing this?

Thanks,
Chad

John,

Thanks for the details. Do you have any information on what materials are most vulnerable to this when using this solution?

Thanks,
Chad

John,

Do you have ANY resources (engineers, clients, contacts) that you can introduce me to that have implemented this solution who we can talk to about pitfalls when it comes to material selection? Our concern is that granular liquid level detection may not be possible with this solution due to the possibility of build up over time, no matter the material.

Thanks,
Chad