FDC1004: Application Help

Hello Renan, 

For the out of phase technique, the GND should not be connected to the shields. The GND can be used with a shielded cable around the Shield connections, but the SHLDx connections should be used to protect the CINx connections as shown in the following image: 

For the saturated measurement, if you use the CAPDAC instead of the CIN4, are you able to get a measurement value that isn't saturated? You may have to adjust the CAPDAC to see this. 

For the water shake issue, the measurement can be impacted by water droplets or residue on the tank that is left behind after shaking the water. If you let the container settle for a few minutes after shaking, does the measurement decrease or is it still the same value as immediately after the shake? 

Best Regards, 

Justin Beigel

Hello Justin,

Please see my customer update below:

Thank you for your reply Justin and sorry for the late response.
We now have not connected the ground to anything, that should be okay for the OoP technique to work?

With CAPDAC enabled, we have indeed values that are not saturated. Then the behavior is similar to a single-ended approach (so CHB disabled).

For the shaking, I waited 1,5 minute and the capacitance value dropped maybe 0.05 pF, but still was 0.5 pF higher (20%) than the value before shaking. It could be water droplets, but not that many for such a difference in capacitance I would suspect.

Regards,

Renan

Hello Renan, 

Thank you for the additional information. Not having GND connected to the sensor is fine for OoP technique. I have a couple more clarification questions for you. 

What is the value of the sensor capacitance/CAPDAC when they no longer have the saturated issue? 

Can you test shaking the water bottle without any water in it? Does the shaking motion itself cause any offset in the measurements? 

Thank you, 

Justin Beigel

Hello Justin,

Please see response from my customer:

In the attached image, the values can be seen with the sensor layout of the flex-PCB as in the design guide. We are switching with connections and cables in between, so this has some influence on the absolute value. Right now the wiring to the sensor is quite long. in the FDC1004 datasheet, it is stated that CAPDAC has no effect for a differential measurement, so why would we want that since our objective is to get the differential technique to work?

I also tried to set an offset to get the value in the right range, as in the second image. However, then the value does not change in time, stays exactly the same.

I did a test with shaking without any water. The values are constant (not fixed) so this is not of influence. By the way, we shake the bottle from a distance via a table, because otherwise the influence of a human hand is significant.

TI response 0704.zip

Regards,

Renan

Hello Renan, 

Thank you for the additional information. The CAPDAC measurement is just to help me understand the sensor that is connected in this testing but with the two screenshots you provided, I don't think we still need to do the CAPDAC measurement. 

When taking CIN1 - CIN4, it should just be a direct subtraction of the two measurements and based on the first image you shared, I would expect around 0.5pF as the result once you switch the input channels. If you setup the CIN1-CIN4 measurement without any sensor connected, do you still get a saturated result? If so, do you have another FDC1004EVM you can test with? 

In regards to the shaking, do you have a reference sensor on the bottle? If so, does the reference also increase by the same amount? 

Best Regards, 

Justin Beigel

Hello Justin,
Eventually I am able to post a reply with my own account. Thank you for the answers, they are helpful in our effort to make it work. We have managed to get the differential technique to work now, first with only a level sensor but now also with a reference. This reference (liquid) also does not take so long to get back to the value it was before shaking. Then the conclusion could be that the droplets on the inside of our bottle explain the slow decrease of the level measurement? For our application, it would be important to have not such long waiting times before the original value is obtained. Do you have any tips for changes in software or layout to compensate for that?

Best regards, 
Maarten Boschloo

Maarten,

Justin is out today.

Droplets sticking to the side of the bottle might be causing the slow settling of the readings.
After shaking, do the droplets' movement correlate to the slow change in readings?

I will consult with Justin about possible software or layout changes to compensate for the slow settling.
One of us will update this thread by Wednesday of next week.

Regards,

John

Hello John,

We do not see droplets moving, we can imagine there is some water as a film that needs to settle. But since it takes so long (order of 100 seconds), we wonder if it is something else than the water coming down. However, we just found out that with a parallel (so the sensor on a plate fixed), the value returns quite immediately after shaking. Our earlier tests were with the sensor sticked on a round bottle. So we think either the round form or the distance to the water is of interest there. 

Kind regards, Maarten

Hello Maarten, 

It may be that something in the connection of the sensor to the plastic water bottle is changing when you shake it. The effects of placing the sensor remotely to measure the liquid level is covered in the Capacitive Sensing: Direct vs Remote Liquid Level Sensing Performance Analysis app note. 

One concern can be a lack of sensitivity for the remote placement but it may not be enough to matter for your application: 

If you test with different distances from the water bottle, do you still get results that don't have the offset after shaking? 

Best Regards, 

Justin Beigel