Suggestions for Shielding for touch IC (AT42QT1070)

Hi all,

I hope all is well. I’ve included both a short version and a long version (with background) about my question.

Short version: I need to somehow shield out a lot noise from the capacitive touch sensor that I’ve got from Adafruit (AT42QT1070, link: Standalone 5-Pad Capacitive Touch Sensor Breakout - AT42QT1070 : ID 1362 : $7.50 : Adafruit Industries, Unique & fun DIY electronics and kits) so it only “registers” a touch when a conductor actually “touches” one of the breakouts. There is backend processing done at an Arduino Mega afterwards.

Long version: In my setup, the goal is to register each individual licks of laboratory rat (an intermediately step in the project). We have a water dispenser that terminates in a metallic tube from which the rat can drink, or lick off, water. I’ve included a picture that shows how I’ve hooked up one of the sensor breakouts to this metallic tube (please let me know if anything’s unclear in the picture). Ideally, the sensor would register a touch only if the rat licks the metallic rod. The issue is that when the rat is in the general proximity of the sensor (with sometimes inches separating the two) the sensor tends to also register a touch.

Is there a way to mitigate this issue? If not, would getting analog or otherwise adjustable touch sensors a be a better idea?

Thank you again, and please let me know if anything is unclear.


So you only want one sensor?

Have you tried a simpler sensor, such as the TTP223?

Thank you for your response!

No I haven't. If you don't mind explaining it, could you tell me why this board would behave differently?

The problem you describe is excess sensitivity of the sensor. The datasheet suggests that it has a means to adjust that sensitivity but that the particular board and manner of connection does not provide for that adjustment.

The TTP223 may be intrinsically less sensitive, or may not be but its sensitivity can be reduced if necessary by adding parallel capacitance between the sensor pin and ground.

Thank you!

Now, I'd imagine I don't have the electrical engineering knowledge to know exactly how capacitive touch works and how adding capacitors in parallel would reduce the sensitivity of the sensor.

If you don't mind, could you point me towards resources that could at least give me an elementary but sufficient understanding of these topics?

If you don't mind, could you point me towards resources that could at least give me an elementary but sufficient understanding of these topics?

Not really, no specific "resource" here, it's just an application of general principles.

To measure capacitance, you have some logic that charges the capacitance to your working voltage - say 5 V - then connects a (large) resistance across it and monitors the voltage as the capacitance discharges. How long it takes depends on the product of the resistance and capacitance - the higher either is, the longer it takes. For very small capacitances, this is a matter of microseconds.

These touch sensors self-calibrate - they determine what capacitance is connected when they start up, and use that as a reference. They are designed to detect a proportional change in the calibration figure. If you add a parallel capacitance to the circuit, then whatever capacitance you are wishing to detect becomes a smaller proportion of the total capacitance, so it must then be a greater capacitance change in order to be detected.

Reference to the TTP223 datasheet suggests adding capacitances up to 50 pF in order to de-sensitise the device, but I gather that you can use more capacitance if really necessary.

Note that you should always mount the sensor board itself - by its three connections - directly adjacent to the touch assembly in order not to be influenced by anything close to the wire connecting the touch point which wire is after all, a touch or proximity sensor in itself. Similarly, the actual size of your feeding tube determines how sensitive it is to nearby proximity - the smaller the better. You can also shield it within a larger metal tube (with plenty of clearance - several mm) connected to the ground point on the sensor.

I have to say that all these considerations apply equally to your original sensor - I just feel it is over-complicated for the purpose.