My setup works fine with following combination of hardware: Raspberry Pi4 connected to a TCA9548A I2C Multiplexer from Texas Instruments, whereas each channel is connected over 4wires (GND, VCC, SDA, SCL) with 140cm 30awg silicone wires to two more I2C sensors.

In the office, the setup works perfectly. When the cables are immersed in water, which they need to be for the application, I get errors. Wrong addresses are displayed for the I2C sensors and it happens very very rarely that the sensors are recognized under their correct address and can also be read out.

My question is basically: How can I model the wire capacitance with regard to the surrounding water and maybe other relevant water parameters such as the electrical conductivity?

  • \$\begingroup\$ If the wires are in the same insulating sleeve, and there is no water between them, there should not be any capacitance change. You are probably having some other issues \$\endgroup\$
    – Eugene Sh.
    Commented Dec 6, 2021 at 18:27
  • 2
    \$\begingroup\$ Look at the signals on a scope in both scenarios \$\endgroup\$
    – Ben Voigt
    Commented Dec 6, 2021 at 18:29
  • 1
    \$\begingroup\$ It's probably down to grounding and the presence of water supplies an extra (interfering) ground electric field connection to all sorts of random other bits of equipment. Use a differential form of signalling is my advice. \$\endgroup\$
    – Andy aka
    Commented Dec 6, 2021 at 18:36
  • 1
    \$\begingroup\$ Have you verified that immersing the cable in water has anything to do with capacitance? Do you have oscilloscope capture to prove this? Is this the same issue than with your previous question why I2C does not work properly on your setup? \$\endgroup\$
    – Justme
    Commented Dec 6, 2021 at 18:43
  • 1
    \$\begingroup\$ if you're in a place where you need to worry about such things, I2C, which is unbalanced open collector, is not what you want to be using. Something like RS422/485, which is low impedance and balanced would be my suggestion. \$\endgroup\$
    – danmcb
    Commented Dec 8, 2021 at 7:37

2 Answers 2


Polymers' water uptake can vary a lot (e.g. https://omnexus.specialchem.com/polymer-properties/properties/water-absorption-24-hours#A-C ).

This water ingress changes the relative permittivity as already mentioned. This is even used to build sensors (e.g. https://www.mdpi.com/1424-8220/18/5/1516/htm ). The final value strongly depends on the exact conditions and material composition but to give an idea, one can take a look at research papers like DOI:10.1109/TDEI.2015.005291 i.e. figure 3 (For a sneak peek look here: https://www.semanticscholar.org/paper/A-capacitance-study-of-anomalous-diffusion-of-water-Gao-Liang/89f65da7fe71ba55a923a3876e801587781f1799/figure/4 ). In the mentioned publication an increase of the relative permittivity from 3.5 to 6.0 is reported for some of the investigated materials.

So how to deal with that?

  1. Select the cable (isolation) carefully - ideally something with a very low water uptake and enough thickness
  2. Keep the cable short as possible
  3. Lower the switching frequency

EDIT: thinking about it, a coax cable might do the trick if it is hermetic.

EDIT2: if none of the approaches works one can try to use differential I2C bus extender. The trick is that differential transmission is less susceptible to signal perturbances. Downside is the additional HW requried.

  • \$\begingroup\$ I am using 30AWG silicone wires with a silicone thickness of 0.24mm around the 0.08mm diameter copper wires. Thanks for the references! \$\endgroup\$ Commented Dec 7, 2021 at 7:24

Water has a dielectric constant of about 80 (compared to about 1 for air and perhaps 4 for plastic insulation), so the cable capacitance will be greater. How much depends on the construction of the cable.

You can measure it by dunking the cable and using your multimeter capacitance range.

  • \$\begingroup\$ thank you! That is a good approach. Is there also a way to reduce the capacitance? use thicker silicone around the cables or shield it with sth.? \$\endgroup\$ Commented Dec 6, 2021 at 18:46
  • \$\begingroup\$ It would only matter if the wires are immersed in the water without any insulation. If these are bundled inside a sealed sleeve, or even if just bundled tightly, so there is gaps between them I would not expect any capacitance difference. \$\endgroup\$
    – Eugene Sh.
    Commented Dec 6, 2021 at 18:47
  • \$\begingroup\$ @EugeneSh. If you think about the electric field lines I think it could be double or triple depending on the jacket thickness. \$\endgroup\$ Commented Dec 6, 2021 at 18:49
  • 1
    \$\begingroup\$ Ok.. I guess it worth an experiment \$\endgroup\$
    – Eugene Sh.
    Commented Dec 6, 2021 at 18:49
  • 1
    \$\begingroup\$ @EugeneSh. I think you are correct if the wire is shielded. \$\endgroup\$ Commented Dec 6, 2021 at 18:52

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.