Outside my house, at the front, I have a set-up of 3 lights attached to the wall. These lights can be switched on using a switch inside the house. Furtermore, they are attached to an infra-red sensor which can switch the lights on for 2 minutes at night time, when someone passes by. I have this same set-up at the back of my house. So far, so good.

I recently changed the light bulbs from incandescent to LED. At the back of my house, this worked fine. The set-up at the front of my house resulted in the LED lights flickering when switched off. If I place back 1 incandescent light bulb, leaving the other two lights LED, this problem does not occur. Following the post Mains LED light bulb flickers when switched off, I understand how LED lights work and why they can flicker. My remaining questions are:

  • Where does the energy come from leading to the flickering?
  • Can I change something in the switch or the sensor to make it stop?
  • Why does the addition of an incandescent light bulb stop the LED lights from flickering?
  • Is a set-up with all incandescent light bulbs using up the same amount of energy, without me noticing as easily?
  • \$\begingroup\$ I provided an answer to this question which addressed this - did you not understand my answer? \$\endgroup\$
    – Andy aka
    Dec 30, 2015 at 13:04
  • \$\begingroup\$ Thanks for your comment Andy. I read your answer to the other post. However, it does not really answer any of my questions above. You just say 'sometimes strange things happen'. Furthermore, this flickering occurs continuously, not only for a few seconds right after switching the electricity off. \$\endgroup\$
    – LBogaardt
    Dec 30, 2015 at 13:48
  • \$\begingroup\$ "As the AC voltage is removed, the DC voltage across Cin remains high" - that's the bit! \$\endgroup\$
    – Andy aka
    Dec 30, 2015 at 14:09

3 Answers 3


Apart from what Brian says, which may very valid in some cases, there are also other options for why this happens:

  1. The IR sensor hasn't got a relay but solid state switches: Solid state switches can always leak a tiny amount of current. This is why solid-state switch or dim packs in lighting rigs for shows need to be un-powered before people are allowed to modify cabling, as the leakage a "turned off" >10A triac can give, especially when aged a little can be dangerous or lethal in some cases. Depending on the design and Q.A. of the IR sensor the difference in leakage between one and another can be quite large.
  2. The sensor showing the problem has more contamination/dirt on the inside, due to wind or rain or even due to sunshine degrading the plastic joints letting water and dirt in over time. This then creates creeping current between the switch contacts.
  3. The IR sensor has a snubbering circuit that creates a leakage capacitance across contacts, which acts the same as the wire capacitance Brian mentions: A capacitor at AC (changing current directions) becomes a sort of resistance to the current flow, and thus allows a little leakage current through.
  4. One of the switches in the system has a NEON pilot light, either hidden or still visible, which leaks a few mA of current through the load lamp.

A pilot light is connected like this, for reference:


simulate this circuit – Schematic created using CircuitLab

With an incandescent lamp, when the switch is open, you can see that the 250 Ohm of the lamp does not add much to the 30kOhm already in the loop, so the neon light will turn on, and only 1 to 2mA will go through your ceiling light, which will not turn it on at all, won't even get warm, as across the 250 Ohm it presents 2mA is only 0.5V, way too little for a 115V lamp to turn on.

Now a leaking Triac will also be in the range of 2mA to 10mA, so will most capacitive coupling. Creep (before other paths will trigger earth faults that should cause a power shut off) usually stays under 50mA. All of those current are much too low to cause any noticeable effect in the wire of the incandescent.

But what happens when that current goes through a LED bulb? Well, a modern CFL or LED bulb usually has very low, to nearly zero internal leakage, so then the internal circuitry can be seen as this (they are much more complicated if well designed, but for this purpose, this is the representation):


simulate this circuit

When you have a hard switch, all is well, switch on: Light on. Switch off, light off.

Now, what happens if you have something "pumping 5mA" into the lamp continuously? 5mA is still not enough to power the LED, right, so... That should not turn it on at all, 5mA is only 500mW, and your LED bulb is, let's say 10W, so the maths is clear.

Well, because there is no leakage, the 5mA will just charge the capacitor. If we assume the current to be perfectly constant (not entirely true, but close enough), it will charge up over time like this:

dV = (dt * I) / C with C = capacity of the capacitor, I is the constant current, and dt is the time period, and dV is the change in voltage.

At some point the capacitor charges to the turn-on voltage of the controller, and the LED will turn on, using the energy in the capacitor. Because your bulb drains the energy in the capacitor much quicker than the current can replenish it it will drop to the turn-off voltage and turn off.

Depending on the size of the capacitor and the turn-on and turn-off voltages of the controller, the flickering can be quick, slow, or even so quick that it looks like an always-on bulb that flickers in brightness.

These flickering speeds also get influenced by the number of bulbs connected.

Funny thing is, that based on the flickering pattern of 3 or more lamps connected to one system it is possible to make some assumptions about what might be the cause. While no guarantee, it has some basis in analysis that is valid.

If they all flicker in an nearly synchronous pattern, the leakage current is highly voltage dependant, whereas if they flicker less in sync the leakage current is more constant, and this may give hints about where to seek when there is a full scope of the entire system.

As a side note: If the flickering is a real "fluid" flickering in an on-state it may be that the LED controller leaks the current through into the LEDs without actually turning on and the flickering is caused by much smaller on/off margins and inferring anything from a pattern is much harder.

Will this energy also leak away if there are incandescents? Yes! As I showed you above, the current flows, you just don't see it. So in a way with the LED lights you're already winning, if you can get past the flicker annoyance, because now at least the leakage is doing something for you.

The reason the flickering stops when you add only one incandescent is basically also already answered: If you add the low-resistance lamp in the loop, the leakage will only cause up to a few volts maximum on the wires, which is way too little for the High-Voltage LED driver to turn on, or to leak into the LEDs themselves.

  • \$\begingroup\$ Awesome! Still need to read through your answer a few times before I fully understand it, but sounds good. Also: the 3 LEDs (6.5W) flicker exactly synchronously, about once a second. \$\endgroup\$
    – LBogaardt
    Dec 30, 2015 at 14:14
  • 1
    \$\begingroup\$ @LBogaardt that means the current is -very likely- more strongly influenced by voltage change, allowing the lamps to synchronise their apparent voltage more stably, so it will be a small leakage, like the NEON pilot light or a parallel capacitance, possibly a slightly aged triac that would normally isolate relatively well. Caveat though: Sometimes patterns are just there and no hint to anything. \$\endgroup\$
    – Asmyldof
    Dec 30, 2015 at 14:38

I have seen this with fluorescent style eco lamps, and a 3-way switch (where 2 switches at each end of a cable can turn the lamp on or off from either upstairs or downstairs).

The 3-way switch diverts "live" or "hot" from one conductor to the other on a 2-way cable running between the switches. To switch a lamp "on" you set your switch to whichever conductor is "hot" according to the other switch.

I'm guessing you have a similar arrangement given you have both an IR sensor and a real switch?

Anyway, the two long conductors also act as a capacitor, and this couples a tiny current (microamps) from the "hot" conductor to the other.

This current does nothing to an incandescent bulb (a low value resistor) except develop a few millivolts across it (uselessly wasting a tiny amount of power).

But to an eco bulb or LED, it is rectified and slowly builds up a voltage across the DC reservoir capacitor until the internal circuitry turns on, flickering the lamp, and discharging the capacitor again.

Fixes : either avoid 3-way switching, or add one very low power incandescent bulb as a load.

There ought to be some other appropriate dummy load for this purpose, the question comes up enough there must be a market need for something of the sort.

You could wire up a simple resistor (designed to consume much less than 1W when powered, and maybe driving a LED) as such a load, but then safety would be entirely your own concern, and it could be illegal under your electrical regulatory codes.

  • \$\begingroup\$ Thank you for your answer Brian. I agree with you that 1 switch & 1 IR sensor should be similar to 2 switches. To sum up, with any type of bulb, this setup uses (a small amount of) power, even when 'off'. But you only notice it with LEDs. I still don't understand how a tiny current (microamps) can couple from the "hot" conductor to the other (?). \$\endgroup\$
    – LBogaardt
    Dec 30, 2015 at 13:55
  • \$\begingroup\$ A capacitor is two conductive plates, separated by an insulator. Connect a voltage across it, and it charges, storing electrostatic energy between the plates. At DC, it conducts no current (once it's charged). But at AC. it allows a small current to flow because you are charging and discharging it every cycle. The cable between 2 switches forms a small capacitor, probably about 100 pf for a run of 10-20 feet. \$\endgroup\$
    – user16324
    Dec 30, 2015 at 14:22


  1. change from a 2 wire sensor to a 3 wire sensor (need to ensure you've got enough cables at the existing sensor as runing cable will increase the cost.

  2. Add a resistive load, ie. Add a halogen at the first light of the circuit.

  3. Add a recitfiying capacitor to the circuit. Not if you have cbus relays and dimmers you'll need to buy a specific rectifying capacitor for those circuits (the cost more)

  • 1
    \$\begingroup\$ Welcome to EE.SE, Mike. I think you haven't really answered the questioned asked. Remember that you can Edit your answers to improve them. Capacitors don't rectify so there no such thing as a rectifying capacitor although there are rectifier capacitors which are added after a diode or bridge rectifier to smooth out the resultant DC voltage. In any case, you haven't explained where to wire this capacitor and how it would solve the flickering problem. \$\endgroup\$
    – Transistor
    Mar 23 at 23:06

Your Answer

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

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