Considering a smart switch based on a power greedy WiFi ESP8266 (800mA spikes and 100mA on idle, @ 3.3V), there is the common situation to have the neutral wire missing from the wall switch box (left with 2 hot wires, maybe some earth wire that I don't want to consider in any way). Powering the electronics in the smart switch from the 2 hot wires will draw current through the light bulbs, making them either flash or light up (maybe except for some incandescent ones, but I can't rely on the light bulb type in any way; all usual types ranging from 1W to 100W, 110/230VAC should be considered).

At this point you might have already guessed the problem: what kind of smart switch design can bypass this issue? Some suggestions I read were to use an extra capacitor in parallel with the light bulb so that the current drawn by the switch won't get through it. Although I'm a beginner with electronics and I don't fully understand how exactly should that work, or how to scale the capacitor (maybe someone could explain me), I only wonder why didn't the LED/CFL bulbs manufacturers just include this feature in their products, since this seems to be an universal issue. Are there any reliable solutions to this problem?

  • \$\begingroup\$ "I only wonder why didn't the LED/CFL bulbs manufacturers just include this feature in their products" if it costs 10 cents a bulb and they make ten million bulbs a year, that's $1,000,000 not in their pockets. \$\endgroup\$
    – Trevor_G
    Commented Mar 12, 2017 at 16:41
  • \$\begingroup\$ Yes. Use Bluetooth low energy. \$\endgroup\$
    – bobflux
    Commented Mar 12, 2017 at 16:51
  • \$\begingroup\$ @Trevor: true enough, but a slightly different product line marketed as "smart switch compatible" that costs some extra 50c/bulb would translate to easy profit. An existing solution is better than none and people expect it to be more expensive. Or might rather the capacitor not be a real solution? I lack the electrical knowledge to answer that... that was part of my dilemma. \$\endgroup\$
    – i2r
    Commented Mar 12, 2017 at 16:54
  • 1
    \$\begingroup\$ Yup but users is smart switches is STILL a very small marketplace. \$\endgroup\$
    – Trevor_G
    Commented Mar 12, 2017 at 16:56
  • \$\begingroup\$ Check out this page electronics.stackexchange.com/questions/161708/… \$\endgroup\$
    – Trevor_G
    Commented Mar 12, 2017 at 16:58

3 Answers 3


What kind of smart switch design can bypass this issue?

I have not seen a Wi-Fi switch that doesn’t need neutral. Perhaps this can be done, but I guess the power consumption is such that manufacturers choose different protocols.
There are popular generic RF (433 mhz, etc.), z-wave and ZigBee switches that work without neutral. All those protocols are designed around low-energy standards. Some can be used with a generic X to Wi-Fi gateway which is powered from mains.

Why do some bulbs require a capacitor in parallel for non-neutral-wire switches to work?

The switch requires some current to pass even in “off” state. If the bulb doesn’t let the current pass or if it starts to flicker as a result of this current, then you will be instructed to install a capacitor in parallel to the bulb.

Why does it solve the problem?

The capacitor forms a capacitive dropper (together with some components in the switch) that bypasses the bulb. A capacitor in an AC circuit forms a current limiter because when current flows one way the capacitor lets it pass until the capacitor is charged in one polarity and then when the current is reversed the capacitor discharged and lets the current flow the other way. The amount of current depends on the capacity of the capacitor and the frequency of the AC.

Also see this relevant discussion.

Why isn’t this standard in all bulbs?

Because it would waste energy. Unless you need some current to pass constantly to power this special kind of switch you shouldn’t install such a capacitor with the bulb.


When the switch is off, we can't allow any current through, or LED light bulbs will flicker. Therefore, we need a low standby current.

WiFi is therefore out of the question. This leaves:

  • Infrared
  • Bluetooth Low Energy

Infrared would be very nice, because standby power is ridiculously low. However, we might want to place the command button in another room, and it doesn't go through walls.

When the lamp is on, the circuit can be powered. For example, make the switch drop 3.3-3.5V or so, which is easily done with a triac and a couple diodes.

This voltage will be used to charge a small NiMH coin battery.

BTLE is great for stuff that transmits whenever it wants, but here, the device will need to periodically wake up and ask its master if it needs to turn on. It can't receive orders while sleeping. Therefore, its power consumption will be higher than expected.

OK, let's say your Bluetooth Low Energy device needs about 1mW to connect 10x per second in order to know if it needs to flip the switch or not. I fudged the figure from this paper.

Now, how to power the device at 1mW average...

  • CR2032 cell (3V 225 mAh) : 1 month. Too short.
  • 2x AAA batteries (1Ah) : 4 months, why not... Still sucks. Would fit in Euro size switch, probably not in US size.
  • Supercap 1 Farad 5V: 1.5 hours. No go.
  • Rechargeable 100mAh 3.6V cell: 15 days.

All these options suck. The rechargeables would need the switch to be turned on periodically to recharge.

Lowering the polling frequency 10x would lower power draw 10x also, but make the switch sluggish.

Another option would be to bite the bullet and allow some leakage current... but how much? It only needs 300µA at 3V3. That's enough to make a LED lightbulb go blink.

Therefore, I propose to change the problem:

If you put the received and controller inside the light fixture, then it can have its phase and neutral wires, and be powered constantly, as needed.

This leaves the problem of the old switch now being a hole in the wall. Simply replace it with a coin cell powered transmitter, which will have the same role as the other wireless switches in the house. No need to connect it to the wires at all.

  • \$\begingroup\$ Thank you for sharing this information, but unfortunately WiFi is a must have in my case (to avoid any other hub equipments along with some other reasons). The issue is how to make a power consumption up to ~1A @3.3V bypass the light bulb. I've seen manufacturers coming up with a device that does the trick when installed in parallel with the light bulb. How do they do it? \$\endgroup\$
    – i2r
    Commented Mar 13, 2017 at 14:00
  • \$\begingroup\$ If you install a device on the light fixture, then why not put the WiFi stuff in there too? \$\endgroup\$
    – bobflux
    Commented Mar 13, 2017 at 14:21
  • \$\begingroup\$ People still want to manually use the light switch (along with their phone or other WiFi command interfaces), you cannot remove the old switch or assume it's always gonna be ON. \$\endgroup\$
    – i2r
    Commented Mar 13, 2017 at 15:32
  • \$\begingroup\$ Look for eaton xcomfort. They have battery powered transmitters. \$\endgroup\$
    – Jeroen3
    Commented Jun 15, 2017 at 8:04

While the switch is off the voltage across the switch can be used to light up a LED and charge a supercap.

A triac can be used to close the switch. Once the triac runs out of power the triac will switch off, allowing the supercap to be charged again and presto...

The circuit can also be designed to switch off for half-a-cycle every now and then to re-charge the supercap before it runs out of power.


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.