LED light in ceiling fan glowing faintly - workarounds?

[optilude]

Hi,

I have just is installed several Philips ceiling fans with LED lights – three Atlas models and one Bliss – in different bedrooms. These models are basically the same fan and light source, they just have different aesthetics. Significantly, the Bliss design lets more light out of the frosted plastic cover that installs over the light fitting.

The units each have a remote control with a receiver that connects to the fan motor and light separately. The light fittings consists of dozens of fixed LEDs in a spiral pattern i.e. it does not use removable bulbs. There is no power at all to the unit when the wall switch is off (i.e. only earth, neutral, and switched live arrive from the ceiling). When the wall switch is on, it is possible to turn on or off the light and fan separately via the remote. The light is not dimmable and there is no dimmer on the wall either.

Last night, we turned on the Bliss fan and then turned off the light via the remote to sleep. However, in the pitch black there was still a faint glow from the light. With the wall switch off, the glow disappeared, but obviously the fan was then also off.

I have now confirmed that with the light on, 246V AC comes out of the remote receiver and powers the light to full brightness. With the wall switch on but the light switched off via the remote, 0.5V AC is still present on the wires going to the light. With the wall switch off, I read 0V. If I take the frosted cover off and look very closely, I can see the glow even in the daylight (more accurately, I can see the glow disappear when the wall switch is turned off).

This is the case for all four lights, but it's not noticeable on the Atlas models because they let less light through the plastic cover. However, on the Bliss light it's bright enough to disturb sleep.

In our previous house, we had some switches without neutrals and lights with dimmers, and we had both Fibaro and Shelly smart dimmer units. These are powered by running a small current through the light circuit, and require the installation of a "bypass" module (as far as I understand, these are just a high value resistor – I read about 15M ohm on mine) to avoid a similar problem with LED lights glowing dimly even when off . They were installed in parallel with the light fitting, i.e. across its live and neutral wires.

As an experiment, I have just put an old Shelly bypass across the wires going to the light from the receiver unit. I can still see a faint glow up close, though it's hard to do a definite "before and after" in the daylight to confirm if it made things better. I will try again in the dark to see if it's made enough of a difference, but it certainly hasn't made the "wall switch on, light off via remote" setting completely glow-free.

I'd like to make sure there is zero glow. Given it's happening on all four lights, I assume this is a design flaw of the receiver unit i.e. a little bit of AC leaks even when the relay to the light is off, rather than a defective unit. Some options might include:

- A different (smaller?) value resistor across the light or even a potentiometer, basically making a more "greedy" bypass? If so, what values to try? I assume also I'd need to ensure this was rated for enough wattage (the light is 28W I think: specs here). I have plenty of pots and resistors from making guitar amps and pedals, but most of them are in the 1/4W to 5W range.

- Some kind of relay that would make the light connection be definitively cut? I have confirmed that if I disconnect the light fitting wires entirely from the remote receiver unit, there is no glow at all.

- Something else?

Grateful for your thoughts!
Martin

 

[Adam_151]

I think you are almost along the right lines, but I doubt 15Meg ohm is unlikely to do anything, and of course you should be aware that putting a resistive load across the supply will waste some energy as heat, likely not significant for your bill, but you do need to be careful of where its going. Also note that for a resistor across the load (rather than in series with it) the draw of the load is irrelevant- the current through the resister will be governed by its resistance and the supply voltage.

Consider some reactance in parralel with the load, rather than resistance, such as RC switch suppressor https://www.rapidonline.com/rc-network-suppressor-26-5856 should give a bit over 30k of reactive impedance at 50hz

 

[bernardgreen]

There is capacitive coupling between Live and Switched Live in the wiring to the switch, this allows a small amount of power to bypass the switch ( when OFF ) and an LED lamp can glow on this small amount of power.

This leaked power can be absorbed by a snubber connected across the lamp

Capacitor 0.047uF micro Farads 250 volts AC in series with Resistor 100 ohms

Which come pre-packaged as a contact suppressor from RS Components.

As a contact suppressor it would be fitted between Live and Switched Live at the switch BUT as a snubber it has to be fitted between Switched Live and Neutral at the lamp..

( If there is a Neutral at the switch it may be more convenient ( space wise ) to fit it in the back box of the switch )

KEMET RC Capacitor 470nF 100Ω Tolerance ±20% 250 V ac, 630 V dc 1-way Through Hole PMR209 Series
RS Stock No. 206-7869
Manufacturer Evox-Rifa
Manufacturers Part No. PMR209ME6470M100R30

 

[optilude]

This is interesting. I actually have the components to make this, but I’d likely want to buy one to have it be tested / safer.

I am confused about fitting between switched live and neutral though. Wouldn’t this in essence create a short when it’s such low resistance?

 

[bernardgreen]

The capacitor has a high impedance ( resistance ) at 50 Hz, much higher than the 100 Ohm resistor.

Calculated impedance for 470 nF at 50 Hz is 6770 Ohms

 

[optilude]

Can you tell me why you chose those values (470nF + 100r)? I will order the one you suggested from RS, which comes in a through-hole package. They also sell https://uk.rs-online.com/web/p/rc-network-capacitors/6167682 which is 100nF + 100r and comes with flexible leads which would be easier to fit and save me soldering some on.

And just to confirm, I then just connect that in parallel with the load?

 

[Taylortwocities]

And just to confirm, I then just connect that in parallel with the load?

Yes, the cap and resistor are in series with each other. Connect the series pair across the load (effectively in parallel with the lamp).

Here's one, fitted.

 

[ericmark]

This is the standard idea, however it would require you to modify the lamp.

There is capacitive coupling between Live and Switched Live in the wiring to the switch, this allows a small amount of power to bypass the switch ( when OFF ) and an LED lamp can glow on this small amount of power.

That is normally the problem where a cable comes from lamp to switch, but in this case, the switch is not used if I am reading it correctly, but left on, and the switching is done inside the lamp, so the lamp is faulty, same as using a smart bulb, and that staying on dim.
I had the problem with a chandelier in my wife's bedroom, I was using an electronic switch, I tried using the capload shown, it did stop the lamp staying on dim, but it then started to shimmer when turned on, since using G9 bulbs, I could change them, the much larger bulb shown here worked one failed, and I opened it to see how it worked, the smoothing capacitor was nearly as big as the whole of the smaller bulb, on latter units I used a TP-Link (Tapo) switch that uses batteries, that also worked, but in your case you're leaving the switch on, so not down to the switch or cables, it is simply a faulty lamp.

 

[bernardgreen]

Can you tell me why you chose those values (470nF + 100r)?

It was more by trial and error and what was to hand at the time in the "junk box"

 

[optilude]

View attachment 351221 This is the standard idea, however it would require you to modify the lamp.

That is normally the problem where a cable comes from lamp to switch, but in this case, the switch is not used if I am reading it correctly, but left on, and the switching is done inside the lamp, so the lamp is faulty, same as using a smart bulb, and that staying on dim.
I had the problem with a chandelier in my wife's bedroom, I was using an electronic switch, I tried using the capload shown, it did stop the lamp staying on dim, but it then started to shimmer when turned on, since using G9 bulbs, I could change them, the much larger bulb shown here worked View attachment 351222one failed, and I opened it to see how it worked, the smoothing capacitor was nearly as big as the whole of the smaller bulb, on latter units I used a TP-Link (Tapo) switch that uses batteries, that also worked, but in your case you're leaving the switch on, so not down to the switch or cables, it is simply a faulty lamp.

Thank you!

I don’t think it’s faulty (but it might be a design flaw) because I have four of them with two different lamp shade designs bought at different times and they all exhibit the problem to some extent.

I did wonder if something is odd about my wall (ceiling) voltage actually. But it’s not obvious what. Maybe it’s slightly high at 246VAC when I last measured.

The light is on a relay that is controlled by a remote control receiver unit. There is a wire between the receiver and the light fitting (live, neutral, earth). I can install the snubber there with minor modifications.

 

[Ragnar_AT]

You can get pre-packaged „load modules“. If I remember correctly, these use a 330 nF capacitor, so 470 isn‘t crazy far off. The higher the capacitance, the lower the reactance and the more current at 50 Hz you get across the cap, so you can start small and go up if the light still glows.

It does make a difference. Some record players have a 10 nF X2 cap across the mains switch to protect the switch contacts. If you accidentally replace that with a 100 nF the motor will spin slowly with the switch in the off position.

BTW, those Rifa caps in the picture are known to fail in a very startling and smelly fashion after 20-30 years. They either go with a bang or with strange fizzing and crackling but either way lots of very smelly smoke. I‘d only ever use MKT or MKS (polyester film) X or Y safety capacitors, not MP (metallised paper) like the Rifa.