LED lamps glowing

[JohnJ0906]

I have a customer who called me about a strange situation, and I wanted to run this by some people before I go take a look at it. He has a ceiling fan with (4) lamp light kit, standard single pole switch. There were 3 LED lamps and 1 incandescent lamp installed. All worked fine, until the incandescent burned out. After that, the LEDs would glow when the switch was off. If an LED is removed, the remaining ones glow a little brighter .

Has anyone heard or seen this before?

 

[JFletcher]

Ive heard of similar things with older dimmers. wag, but if there is no dimmer, maybe the neutral is being switched and there is some small fault in the fan/light? maybe a 3-way switch to a dimmer elsewhere? bad switch? in any event, I'd plan on pulling the switch first (after observing the problem).

 

[GoldDigger]

There is clearly current bypassing the switch. And it is a case of limited current, not limited voltage, based on the increased brightness when one is removed.
My guess would be capacitive coupling between wires of the switch leg.
With an incan or a CFL the "phantom" voltage would not cause any visible light output and the one incan would drop the voltage below the threshold of the LEDs.

 

[Sierrasparky]

There is clearly current bypassing the switch. And it is a case of limited current, not limited voltage, based on the increased brightness when one is removed.
My guess would be capacitive coupling between wires of the switch leg.
With an incan or a CFL the "phantom" voltage would not cause any visible light output and the one incan would drop the voltage below the threshold of the LEDs.

Capacitive coupling causing light to glow. Please.

The switch is probably not a make break and some sort of solid state or other that does not have a air gap when switched.

 

[GoldDigger]

Capacitive coupling causing light to glow. Please.

The switch is probably not a make break and some sort of solid state or other that does not have a air gap when switched.

OK, let's run some back of envelope numbers.

Suppose you have a phantom voltage of 90V on a wire when measured with a meter with an input impedance of 1 megohm.
That is sourcing a current of about 90 microamps while being measured.
If the lamp has an LED array with a voltage threshold of about 90V rather than using a buck driver down to individual LED levels that is .09ma.

With a voltage of 90V that is 8 milliwatts. The normal power of the LED bulb might be 8 watts.
So a factor of 1000 down from full brightness. That is well within the light level visible as a glow in a darkened room and looked at up close.

I may have made some wrong assumptions, but probably not far enough off to be outside the two or more orders of magnitude that would still let the light be easily visible. Remember that a fully dark adapted human eye can detect the light of a single candle at a distance of 40 miles.

We are used to devices that cannot deliver any light at all at low power levels because they are based on incandescence.
LEDs introduce a different mechanism.

Now the OP may have some additional information for us that knocks my analysis for a loop, like the glow being visible in a daylit room.

 

[Barbqranch]

Perhaps others have, but I have never measured 90 Volts w/ even a 10 megohm input meter. My bet is it was a toggle switch type dimmer that burned out most of the dimming capability, but could still turn it on and off.

 

[K8MHZ]

Is the switch illuminated?

I have seen illuminated switches connected to LED lights do strange things when they were off.

 

[GoldDigger]

Is the switch illuminated?

I have seen illuminated switches connected to LED lights do strange things when they were off.

:thumbsup:

Most illuminated switches are lit when the load is off and are wired in parallel with the switch contacts.
The ones that illuminate or drive a pilot light when the load is on need access to a neutral for proper operation and are wired in parallel with the load.

 

[JohnJ0906]

Thanks for the responses. :thumbsup:

When I get a chance to get over to see the fan, I'll let you know what I find.

 

[gar]

150939-2048 EDT

Your illumination is not from some form of capacitance coupling unless the capacitor is large.

Experiment on two bulbs, one a Cree 120 V 9.5 W, the other unknown 120 V 9.99 W.

Test circuit 120 V 60 Hz line source, usually about 125 V. An adjustable capacitor box 0 to 1.1 ufd in 0.01 increments. The capacitor was in series with the bulb.

Unknown:
Required 0.3 ufd to start light, and dropped out at 0.07 ufd.

Cree:
Would not start up to 1.1 ufd.


Next experiment was to replace the series capacitor with a Potter & Brumfield OAC5. The OAC5 includes a snubber that leaks some current around the solid-state switch.

Neither blub glowed when the OAC5 was off, nor was there a glow after the bulb was on and the OAC5 was turned off.


With a series 5000 ohm resistor unknown glows fairly bright, Cree does not glow at all.

.

 

[gar]

151004-1435 EDT

JohnJ0906:

Have you determined the cause of your problem, and/or found a solution?

.

 

[don_resqcapt19]

Perhaps others have, but I have never measured 90 Volts w/ even a 10 megohm input meter. My bet is it was a toggle switch type dimmer that burned out most of the dimming capability, but could still turn it on and off.

I have seen it with in a volt of two of the supply voltage using that type of meter.

 

[big john]

I have seen it with in a volt of two of the supply voltage using that type of meter.

Ditto. That said, that has always been very long commercial circuits, not in homes

I've got a box of donuts says this is a neon illuminated wall switch or some other controller with no true "off" state.

 

[gar]

151004-2128 EDT

A simple experiment, or a calculation will show that you can get a relatively large voltage reading from only a small amount of capactive coupling.

Experimentally use a variable capacitor box. I used one that increments by 0.0001 mfd (100 pfd), and has a maximum value of 0.011 mfd. The capactive reactance of 0.011 mfd at 60 Hz is about 220,000 ohms. At 300 pfd the reactance is 9,000,000 ohms. Note, harmonics can cause a higher reading.

With 125 V and 300 pfd in series with a 10 megohm meter I read about 88 V. Use a vector diagram calculation with 9 M and 10 M fed from 125 V and the voltage across the 10 M resistor is 0.743 * 125 = 92.9 V . But the the high impedance meter has some shunt input capacitance that is significant when the series capacitors is only 300 pfd. Thus, one expects a slightly lower reading than 92 V.

The capacitance per foot between the two current carrying conductors in a #12 Romex cable is about 18 pfd. 300 pfd would be about 16 ft. This does not take into account any shielding effect from adjacent conductors, for example the EGC.

.

 

[Dennis Alwon]

We had a similar issue with an LED uc light. There were 2 lights on the switch and one would stay one for quite some time after the switch was turned off. The other worked fine. We change out the une light and all was fine.

To the op- that problem with the incandescent is a fix for that situation. The problem is the dimmer switch which I believe is installed.

 

[GoldDigger]

We had a similar issue with an LED uc light. There were 2 lights on the switch and one would stay one for quite some time after the switch was turned off. The other worked fine. We change out the une light and all was fine.

To the op- that problem with the incandescent is a fix for that situation. The problem is the dimmer switch which I believe is installed.

I had one candelabra type bulb with multiple (10+ small LEDs inside which I got from Costco several years ago. It apparently had a large filter capacitor after its rectifier and it would glow to where you could still see the LEDs in a darkened room for more than 20 minutes, even after unscrewing it.