Analysing a mutual capacitance sensor's signal disparity

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
jcsb1994
Level 4
Level 4
First solution authored 50 replies posted 50 sign-ins

*There is a lot of text (hopefully you can learn something) but the question is at the end!

Hi all,

as explained in this previous topic Minimum and maximum overlay thickness for CSX Capsense widgets , a CSX Capsense widget will undergo a capacitance increase (or a rawcount drop) as the finger gets too close to the sensors, hence the importance of picking an overlay with a certain thickness. This effect that we can call signal disparity is explained (even though not mentioned) in the section 7.5 of the capsense design guide for psoc 4 and 6: Effect of Grounding in CSX Method.

I want to try analyzing the exact behaviour of signal disparity on my device. For this, I will test my capsense PCB without any overlay on top of my 3x5 CSX touchpad widget. I am using the CSX tuning code example.

1. Testing without an overlay, with a finger.

pastedImage_7.png

Figure 1: Bare PCB, the signal disparity is most important when using capsense this way.

The signal seems to behave the expected way. as I approach my finger, the raw count increases, until I directly touch the PCB. As I start pressing with more force and my finger is squeezed, the raw count quickly drops to 0.

Once I release, there is a weird behaviour I cannot quite explain. It is probably due to the baseline. When I release, the raw counts increase back to a fairly high number and remain high (detecting a touch) even though the touchpad isn't touched anywhere.

Can this be corrected by changing the baseline (I don't exactly understand how to do it?)

pastedImage_2.png

Figure 2: Readings from Capsense Tuner UI with a bare PCB

2. Testing with a very thin overlay, with a finger

pastedImage_8.png

Figure 3: PCB with a 2mil (0.05mm) thick overlay

I tried adding a 2mil thick kapton tape as an overlay (2mil ~ 0.05mm so 1/10th of min recommended thickness). The behaviour is similar to described above, but the widget is almost usable -- if I do not press my finger too much, the touch reporting in the capsense tuner UI is pretty accurate.

The baseline problem still occurs after removing my finger, but the false touch is much less perceptive. This makes sense, I am keeping the finger at a larger distance than before, so the signal disparity isn't as important.

3. testing with a very thin overlay, with a grounded conductive layer on top

pastedImage_9.png

Figure 4: PCB with a 2miloverlay and a conductive copper tape layered on top and connected to ground

I want to have a conductive layer on top of variable overlay thicknesses so I can have a methodical approach to how signal disparity evolves with the overlay thickness.
I am layering a copper tape, connected to ground, over the kapton overlay. The copper tape acts as a grounded conductor (like a finger) and affects capsense signals.

So if I start the MCU with the copper tape grounded, nothing happens as the baseline is set accordingly, and the rawcounts are set to 0.

But if I start the MCU with the copper tape floating, and connect it afterwards, then the capacitance drastically drops as the sensors pair with the grounded copper tape, and the raw counts drastically increase accordingly. This lets me see how much the sensors pair with the conductive layer. I can then repeat with various overlay sizes and determine how distance affects pairing.

pastedImage_6.pngpastedImage_11.png

Figure 5: Readings from Capsense Tuner UI, left has a 2 mil overlay, right has

I can effectively tell that the pairing is weaker when the overlay is thicker, which seems to make sense as the ground is further away for the sensors to sink into. But somehow I was expecting something different and I don't understand: I am not able to reproduce the 0 rawcount I get when pressing a finger hard on the sensor. I thought the capacitance would be high at low distances, due to signal disparity. Even though I place the conductive layer at 0 distance from the sensors, the raw counts are the highest, but they should be lower with an increase in capacitance as I get too close to the sensors... Why am I not seeing the signal disparity phenomenon with my conductive layer? It works well when testing with a finger as seen in paragraph 1.

Thanks a lot for reading!

0 Likes
1 Solution
Hari
Moderator
Moderator
Moderator
750 replies posted 500 replies posted 250 solutions authored

Hi @jcsb1994 

 

I see a couple of questions. 

 

First, regarding the sensor being always on. That is an expected behavior and it is because the baseline is resetting to the lower value. You can tweak the firmware to make sure that the baseline does not drop down but since this is not a usual use case, it might not be worth adding the extra firmware overhead.

 

Observation regarding  grounded copper vs finger without the overlay - 

This again is expected since the Cm variation is different in both the cases. 

Case 1 - Finger directly touching the sensor - In this case, the current drawn when a finger touches the sensor increases effectively causing an increase in the Cm as seen by PSoC.

Case 2 - Grounded copper on the sensor - In this case, the copper plane will take the field lines and will not let the field lines go to the Rx electrode. This means that the effective Cm as seen by PSoC decreases and therefore the raw counts increases.

An additional point to keep in mind is that when the actual Cm decreases, the raw counts increases in the PSoC. This is a firmware inversion kept in the component to make sure that users aren't confused.

 

Case 3 - we can have another experiment where we keep the copper plane floating. Start the PSoC and initialize CapSense and then place the floating copper on the sensors. This will simulate finger press without overlay condition as the Cm will increase when the copper layer is kept on top. This is due to the fact that the copper layer will ease the flow of field lines from the Tx to the Rx electrode.

Harigovind_0-1609910539301.png

 

 

This is a very useful thread and I want to thank you for the experiments you are conducting and sharing with the community. 🙂 

 

Best regards, 
Hari

View solution in original post

4 Replies
Hari
Moderator
Moderator
Moderator
750 replies posted 500 replies posted 250 solutions authored

Hi @jcsb1994 

 

I see a couple of questions. 

 

First, regarding the sensor being always on. That is an expected behavior and it is because the baseline is resetting to the lower value. You can tweak the firmware to make sure that the baseline does not drop down but since this is not a usual use case, it might not be worth adding the extra firmware overhead.

 

Observation regarding  grounded copper vs finger without the overlay - 

This again is expected since the Cm variation is different in both the cases. 

Case 1 - Finger directly touching the sensor - In this case, the current drawn when a finger touches the sensor increases effectively causing an increase in the Cm as seen by PSoC.

Case 2 - Grounded copper on the sensor - In this case, the copper plane will take the field lines and will not let the field lines go to the Rx electrode. This means that the effective Cm as seen by PSoC decreases and therefore the raw counts increases.

An additional point to keep in mind is that when the actual Cm decreases, the raw counts increases in the PSoC. This is a firmware inversion kept in the component to make sure that users aren't confused.

 

Case 3 - we can have another experiment where we keep the copper plane floating. Start the PSoC and initialize CapSense and then place the floating copper on the sensors. This will simulate finger press without overlay condition as the Cm will increase when the copper layer is kept on top. This is due to the fact that the copper layer will ease the flow of field lines from the Tx to the Rx electrode.

Harigovind_0-1609910539301.png

 

 

This is a very useful thread and I want to thank you for the experiments you are conducting and sharing with the community. 🙂 

 

Best regards, 
Hari

jcsb1994
Level 4
Level 4
First solution authored 50 replies posted 50 sign-ins

@Hari,

Thank you for your kind words and your useful reply!

Testing with a floating copper plane

I managed to test the floating copper plane experiment you suggested. I used a piece of plastic card with copper tape, mounted on a plastic handle to keep the hand away from the sensor

 

JeSi_4326976_2-1610046515327.png

Figure 1: plastic tool used to lay a copper plane on the CSX capsense touchpad widget

It is interesting to see the different impact of a grounded finger compared to the floating plate. The former increases the rawcounts (lowers capacitance, normal CSX behaviour as seen in PSoC 4-6 Capsense guide rev X section 2.1.2 Mutual-Capacitance Sensing), while the plate lowers the rawcounts quite a lot (increases the capacitance a lot).

JeSi_4326976_4-1610046643628.png

Figure 2: Graph view of capsense tuner UI, seeing the effect of applying the copper plane plastic tool on the touchpad, and then applying a finger on the touchpad

Above we can see the copper plane  lowering the raw counts (increasing the capacitance). As you said, the copper plane is easing the flow of rx-tx field lines, but I have created a post on stackexchange to seek help to better understand the theory behind this phenomenon https://electronics.stackexchange.com/questions/541355/how-to-redirect-electric-field-lines-with-a-c...

The answer given by user Voltage Spike explains how the floating copper plane is actually forming a capacitor with Tx and with Rx, since AC is driving CSX anyway. This results in a bigger capacitance since Cm between Rx and Tx is unchanged, but the copper plane comes in as an additional capacitor plate pairing with Rx, reducing the raw counts. This seems to make sense, I was expecting more of a modified flow of the electric field, as if the copper plane was acting like a mirror and compressing the electric field, reducing the distance between the plates. But Voltage Spike's explanation made me understand this totally differently. Do you (or anyone reading this post) have any input on this?

 

 

0 Likes
jcsb1994
Level 4
Level 4
First solution authored 50 replies posted 50 sign-ins

@Hari,

Thank you for your kind words and your useful reply!

Testing with a floating copper plane

I managed to test the floating copper plane experiment you suggested. I used a piece of plastic card with copper tape, mounted on a plastic handle to keep the hand away from the sensor

 

JeSi_4326976_2-1610046515327.png

Figure 1: plastic tool used to lay a copper plane on the CSX capsense touchpad widget

It is interesting to see the different impact of a grounded finger compared to the floating plate. The former increases the rawcounts (lowers capacitance, normal CSX behaviour as seen in PSoC 4-6 Capsense guide rev X section 2.1.2 Mutual-Capacitance Sensing), while the plate lowers the rawcounts quite a lot (increases the capacitance a lot).

JeSi_4326976_4-1610046643628.png

Figure 2: Graph view of capsense tuner UI, seeing the effect of applying the copper plane plastic tool on the touchpad, and then applying a finger on the touchpad

Above we can see the copper plane  lowering the raw counts (increasing the capacitance). As you said, the copper plane is easing the flow of rx-tx field lines, but I have created a post on stackexchange to seek help to better understand the theory behind this phenomenon https://electronics.stackexchange.com/questions/541355/how-to-redirect-electric-field-lines-with-a-c...

The answer given by user Voltage Spike explains how the floating copper plane is actually forming a capacitor with Tx and with Rx, since AC is driving CSX anyway. This results in a bigger capacitance since Cm between Rx and Tx is unchanged, but the copper plane comes in as an additional capacitor plate pairing with Rx, reducing the raw counts. This seems to make sense, I was expecting more of a modified flow of the electric field, as if the copper plane was acting like a mirror and compressing the electric field, reducing the distance between the plates. But Voltage Spike's explanation made me understand this totally differently. Do you (or anyone reading this post) have any input on this?

 

0 Likes
Hari
Moderator
Moderator
Moderator
750 replies posted 500 replies posted 250 solutions authored

Hi @jcsb1994

 

If you take a step back, you can see that your understanding as well as VoltageSpike's explanation are identical. In some sense, we can think that the electric field lines are "reflected" by the copper plane and the reason for this is as shown in the picture that was shared.

Harigovind_2-1610364630415.png

(sharing the image here for context)

 

The 3 capacitors, Ctx Crx and Ctx help the field lines reach the copper plane, beyond which the conductor helps in transmitting them to the node.

This effectively increases the capacitance between Tx and Rx. 

I like to visualize the capacitance based on the electric field line's path (which I see you doing here as well) since I can apply it to more scenarios...

 

Best regards, 
Hari

0 Likes