I have 2-gang touch wall switches that I want to make Wi-Fi-controlled. I'm intending to use Wi-Fi-enabled relays connected to the switches so that the switch outputs are connected to the relay switch inputs.

The problem is that the switches do not use the neutral line and leak current through the load in order to power themselves. This means that I cannot simply connect these two kinds of devices as described above and I somehow need to allow for constant current flow through the switches.

For now, I installed 5W 47kΩ ceramic resistors and everything seems to work as expected, but the resistors waste a lot of power and heat up to 75°C:


(The 2W connection is for 2-way communication between the switches and only one of them actually drives the lights.)

I'm thinking about replacing them with metallized polyester film capacitors, as illustrated below:


I tested a range of resistances and found out that the maximum resistance for the switches to work properly is around 47kΩ. I was thinking about using 100nF capacitors rated for 630V, such as these jb 2J 104K ones, which should provide around 31.8kΩ capacitive impedance at 230V 50Hz.

Am I reasoning correctly and is it safe to use these capacitors in my case? Note that they would be under constant operation. Also, I'm not sure if the phase shift introduced by the capacitors would be an issue here.

  • 1
    \$\begingroup\$ Put a small reisstor in series with each capacitor to limit the worst case inrush current to something your switches and fuses can handle. \$\endgroup\$ – Neil_UK Jun 21 '20 at 13:53
  • 1
    \$\begingroup\$ Remember Zc impedance of cap and current depends on dV/dt which may be much higher than 50Hz sine unless these are zero-crossing type switches and there is a line filter. \$\endgroup\$ – Tony Stewart EE75 Jun 21 '20 at 14:08
  • \$\begingroup\$ @Neil_UK Thanks, will do that! \$\endgroup\$ – pajac Jun 21 '20 at 16:48
  • \$\begingroup\$ @TonyStewartSunnyskyguyEE75 I'm not sure if I have an easy way of knowing if these are zero-crossing type switches. Assuming they are not, how could I address this issue? \$\endgroup\$ – pajac Jun 21 '20 at 16:48
  • 1
    \$\begingroup\$ Without details on the design, I can imagine if you are concerned about reliability, you will measure the voltage or current stress across Neil's current-limiting R and examine the internal test points with twin probes for differential readings. \$\endgroup\$ – Tony Stewart EE75 Jun 21 '20 at 19:56

I ended up using the capacitors but also added 4.7k resistors in series, as suggested in the comments:

Final circuit

Everything seems to work fine so far. The heat issue appears to be resolved as well.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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