I'm using a capacitive sensor CAP1188 with Arduino, connecting 8 of its touch sensor inputs to 8 copper boards that I want to be the actual touch surfaces.

2-line capacitive sensor and copper board

The wires to the sensor are quite long (100 to 500mm depending on the location of the copper board in the system) and approaching them also triggers the sensor, which I do not want.

How to remove the noise on these wires?

To fit the setup the wires need to stay thin (2mm of external diameter max). I am currently using phone wires (from R11 male - RJ male cable)

I've already tested surrounding the wire with a large piece of aluminium foil connected to the sensor's GND, but the detection seems even more noisy. For simplicity I grouped the wires together when possible (i.e. protected with a single aluminium foil), but I presume this is a bad idea?

  • 5
    \$\begingroup\$ Shields are capacitors, too. Too much shielding will swamp the changes you're trying to detect. Better to put the active circuit close to the sensing surface. Make copies if necessary. \$\endgroup\$
    – Dave Tweed
    Mar 2, 2016 at 3:04
  • \$\begingroup\$ Yeah OK so there is absolutely no other solution than making the active circuit closer to the surfaces? What about the insulation around wires (yellow/black plastic on the illustration), does it influence the detection somehow? Should I remplace them by nude copper wires? \$\endgroup\$
    – myoan
    Mar 16, 2016 at 23:33
  • \$\begingroup\$ Keep the insulation; it doesn't matter as far as the capacitance is concerned. \$\endgroup\$
    – Dave Tweed
    Mar 17, 2016 at 1:28

1 Answer 1


Adding a grounded shield around the wire will significantly increase the capacitance of the wire, resulting in even worse signal quality.

There are two ways you can solve the problem:

  1. Use a driven shield instead of a grounded shield.
  2. Add measurement circuitry at the sensor and communicate with it via a serial link.

Option 1. Use a driven shield

The first solution requires a circuit to drive the shield to the same voltage as the sensor conductor, but to drive it with a low-impedance driver. For example, see TI application report SNOA926A section 2.2: enter image description here

See section 6 Driven Shield (E-field Directivity) in Atmel QTAN0087 Proximity Design Guide. Here is the driven shield circuit as an example:

Figure 6-3 Driven Shield Circuit from QTAN0087

and to apply that to the wires connecting the sensor to the measurement circuitry, Figure 6-4 provides a diagram:

Figure 6-4 Coaxial Sensor Cable with Driven Shield

Option 2. Add a measurement device at the sensor

The best results will be obtained if the measurement circuitry is local to the sensors. You can put a dedicated capacitive sensing IC (for example, from NXP or TI) or even just a cheap microcontroller ($1-$2 max) in the same assembly as the sensors. Then use a digital communications link like I2C or SPI or UART to read the capacitive measurements from the sensor module.


Using a driven shield requires use of a coaxial cable for each capacitive sensor, and added electronics to drive each shield. This adds significant cost and bulk when there are multiple sensors.

Moving the sensing circuitry local to the sensor itself, whether you use a microcontroler or dedicated capacitive sensing IC, costs less than even a single length of coaxial cable and associated connectors. Only about 4 conductors will be necessary (power, ground, clock/data (I2C) or TX/RX (UART) for instance), and since no sensitive analog signals are carried, the cablign is not critical. Also the number of capacitive sensors can be increase almost without limit, by simply adding more sensing chips.

  • \$\begingroup\$ Thus, Option 1 assumes that the capacitive sensors provides a PIN to be connected to the shield, that is properly driven to get rid of interferences? \$\endgroup\$
    – myoan
    Jul 29, 2016 at 14:06
  • \$\begingroup\$ Option 1 (driven shield) has less to do with the sensor itself than about the cabling from the active measurement circuitry. The driven shield is a conductor driven with a low-impedance buffer on the measurement circuitry so it is at the same voltage as the capacitive electrode sensor output. This nullifies the electric field within the shield, so outside influences near the cable don't affect the measured capacitance. I will add some more references to the answer in case you want to dive deeper into that topic. \$\endgroup\$ Jul 29, 2016 at 15:04
  • \$\begingroup\$ @ColinDBennett If having the sensor module closer to the proximity sensors means having the same sensor module further away from the MCU (and therefore longer I2C lines communicating the sensed data from the sensor to the mcu), wouldn't that cause more harm than good? In other words, isn't it better to have I2C lines as short as possible? \$\endgroup\$
    – Geo
    Aug 15, 2020 at 17:37
  • 1
    \$\begingroup\$ I2C can go a considerable distance without problems, as long as the clock rate is not too high. E.g. 1 m at 100 kHz. The capacitive sensing conductors are much more sensitive to interference than I2C because of the high impedance, so that's where I would be focused. \$\endgroup\$ Aug 17, 2020 at 18:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.