LED's glow when off

malcolmcro

Newly Enlightened
Joined
Jul 20, 2010
Messages
1
I notice there is a similar problem reported in flashlights but not quite the same.

http://www.candlepowerforums.com/vb/showthread.php?t=277357

I have renovated my house using LED's for most of the lighting and of the 50 or so fittings 2 glow faintly all the time they are switched off.

When the lighting circuit on the fuse board is switched off they do go out. This suggests a faulty switch but they are on different switched circuits and the other LED's on the same circuit do not glow.
Could it be inductance from a nearby lighting circuit or is there some other explanation?

Malcolm
 
most likely it is circuit leakage of the light itself. A fraction of a milliamp can cause an LED to glow but not put out useable light and the amount of energy used over a year is probably just a few pennies so I wouldn't put a lot of effort in trying to fix it if that is the issue.
 
It's the light switch. Some let through a few milliamps - nothing you'd notice with a normal light. Also, the driver getting the voltage right can have a capacitor that discharges very slowly through the LEDs. I think the first is more likely for 5mm LEDs and the second for power LEDs.
 
It's the light switch. Some let through a few milliamps - nothing you'd notice with a normal light. Also, the driver getting the voltage right can have a capacitor that discharges very slowly through the LEDs. I think the first is more likely for 5mm LEDs and the second for power LEDs.

if it is the light switch putting the LED in a different fixture would eliminate that possibility for sure
 
Actually, LEDs glow easily enough that, in the dark with your eyes adjusted, an LED may glow when you hold one lead and touch the other to a piece of metal just from the 60Hz radiation around the house.

Did it under the power lined behind the house. Now a lot of people know you can make a fluorescent tube glow weakly by bringing it under the power lines, but don't know you can do it with LEDs too. Got a pretty nice glow off that in the right weather. But, you kinda gotta be looking for it. You're not gonna light a path with that.
 
most likely it is circuit leakage of the light itself. A fraction of a milliamp can cause an LED to glow but not put out useable light and the amount of energy used over a year is probably just a few pennies so I wouldn't put a lot of effort in trying to fix it if that is the issue.
Would this be akin to parasitic drain in a flashlight?
 
Would this be akin to parasitic drain in a flashlight?

nope, parasitic drain doesn't cause LEDs to glow, it is the drain usually by electronics circuits that detect you pressing a momentary switch to turn on things and change modes. What you have is leakage not drain. Drain is power used normally leakage is power getting where it should not be usually caused by substandard components and/or design.
 
if you have lighted switches leds will glow, there is still some current going thru the bulb in the switch so it glows when off, you can actually get zapped when touching wires with lighted switch off, but it wont happen with regular switch.
 
if you have lighted switches leds will glow, there is still some current going thru the bulb in the switch so it glows when off, you can actually get zapped when touching wires with lighted switch off, but it wont happen with regular switch.

If the lighted switches are designed properly they should not send any power through lights at all. You can wire a light directly to the source separate of the circuit path of the main bulb so that it has no effect on it.
 
If the lighted switches are designed properly they should not send any power through lights at all. You can wire a light directly to the source separate of the circuit path of the main bulb so that it has no effect on it.
they do. apparently that is how they designed, small neon bulb drawing few mA, disconnect load and they wont have that little light on,
i,m talking about regular light switches, not digital one touch dimmers with tiny blue leds.

in most houses\buildings i worked\lived, you have no option to rewire load, most of the time you only have hot and switch leg.
 
I would just replace the light switch with one without a neon light in it then if the lighted switch doesn't have separate power leads and access to voltage, ground, and light circuit (3 inputs).
 
Thats because if it is dimmed it will heat up the driver, maybe burn it out.

I am not sure if it is ok for LED bulbs to be driven at such low levels, some of them say they are not for dimming.
 
I am not sure if it is ok for LED bulbs to be driven at such low levels, some of them say they are not for dimming.
there was a thread here on led retrofit bulbs, someone said they were flickering when dimmed.
i don't see why led or driver would be damaged if you supply 1\1000 of the current, but anything is possible. call light manufacturer, ask them if it is ok to use the bulbs like that
 
Most LED "bulb replacements" have their driver in them (in theory you could make one "driverless" using 40ish LEDs at 3-ish Vf = 120ish Vf total (to match 120V US supplies). Those drivers are typically only designed to operate either "off" or "120ish" volts, and they boost or buck accordingly for the desired output parameters. Their compents are rated accordingly, if you have designed around a controller chip designed to operate at 10% duty cycle at 120V input, then if you operate them too far out of range (really low for instance), the chip might be running 100% while trying to "boost" the circuit (just a for-instance)... it never gets enough voltage of course to generate the desired output voltage/current, so although the LEDs themselves will be fine (LEDs don't care about "too-low" voltage), the controller IC may burn itself out trying to make it work, making your bulb worthless.

That's just an example possibility, and if it's just induced low voltage/hig-millivolt levels, it's unlikely the IC will be able to run at much of a duty cycle either (you can "boost" from below 1V in some circuits to a few volts)... so basically the "start" circuit boosts the millivolts to 3.0V or so (which typical ICs use nowadays)... which wakes up the IC, it tries to do it's thing, gets a little boost out to the LEDs, and then the whole circuit probably stops/collapses electrically until the next "start" cycle begins, so it'd be a fractional duty cycle for the whole circuit, not going to hurt anything.

The "no-dimming" problem is totally different, many dimmers work by "chopping" the current to the fixture (or they may cut the voltage level, with the same result/problem below), if it's a smooth 60hz sine wave, and you dim 50%, it "chops" out every other cycle for instance. Incandescent bulbs don't care much, they just cool a tiny bit in the off part then back up in the on part, but it's not a major percentage of their operational range.

However, your LED circuit cares a LOT about this (either chopping or dropping voltage)... if it's chopped 50%, then the driver has only 50% of it's operating power to work with to the LED... remember, it wants to drive the LED 100% (unless designed for dimming), so the built-in circuitry has to work twice as hard to boost the output. If the parts are designed around a 120V/1A input, now it has to take 2A during the "on" cycle to make up for the "off" cycle, potentially burning out power controllers, etc. AND/OR the controller IC runs twice as hard as designed, burning itself out, or all the above... bottom line, it's working outside it's design specs and going to die a premature death.

<edit> A driverless (say 140Vf worth of LEDs (or LEDs and resistors) in series (to prevent surge voltage damage) fixture on the other hand would have no issues with dimming, if the dimmer chops the current, the LED runs at a fraction of it's full "on" state, if it drops voltage, the LEDs run below their max Vf, but are fine as well. It's the control circuits that are the issue (although a dimmed control circuit could potentially cause damaging surges as a consequence).
 
Last edited:
Most LED "bulb replacements" have their driver in them (in theory you could make one "driverless" using 40ish LEDs at 3-ish Vf = 120ish Vf total (to match 120V US supplies).

Actually, I recently learned that this isn't quite the case. 120VAC is called "120 volts" because it effectively averages to the same power in a given resistor as 120 VDC. Rectified AC power (What we'd want for LEDs) alternates between 0v and a peak - The sine wave becomes a series of lumps. These lumps average to 120 volts, but the peaks are rather higher. I can't recall exactly how much higher it goes, but this wiki goes into more detail. It seems that half-wave rectifying (a pulse then an equal-length 'off') gives 85VDC, and full-wave gives 170VDC. Adding a smoothing capacitor would reduce the ripple but you still get this voltage.

Such a design would have full-wave rectifying with a capacitor to smooth any bright-to-dim cycling. If the capacitor size isn't chosen with dimming in mind, you will get two failure modes: Flickering (Capacitor discharges 60 times per second) or capacitor failure - filling an empty capacitor involves large current surges, which could damage the capacitor if it has no limiting resistor.
 
It's the light switch. Some let through a few milliamps - nothing you'd notice with a normal light. Also, the driver getting the voltage right can have a capacitor that discharges very slowly through the LEDs. I think the first is more likely for 5mm LEDs and the second for power LEDs.

Beat me to it.
 
I'll bet the LED's that glow are on 3 way circuits. i.e. lights that are controlled by two switches. The red and black wires in the diagram below run in the same cable. The two conductors separated by an insulator make up a capacitor. Any capacitive leakage current from red to black will flow through the light fixture. It won't be much, but it might be enough to light up an LED a little.


3wayswitchdiagram.gif


If it's not the 3-way circuits, are the switches strictly mechanical? i.e. not dimmers, not electronic, not times, not lighted switches, etc.

You could get capacitive leakage like the 3 way circuit if an unswitched hot wire runs next to the switched hot wire between the switch and the fixture, but that would seem less likely and would probably be less tightly coupled than the "3 way cable," which will probably be two wires in the same sheath.
 
I agree with Apollo Cree that it's probably a three-way circuit, or a hot runs parallel to the switched leg for some reason. I have the exact same thing going on in my house. (3-way circuit). It involves a diffused-torpedo-cover lamp with about 20 or 25 5mm strawhat LEDs in series, probably with some combination of resistor/capacitor in series with that to limit the current. (I haven't opened the base, but the cover got broken, hence I know some facts and surmise others.) There is enough leakage (stray) capacitance to light many of the LEDs in this lamp dimly, whereas an incan would just shrug it off without even beginning to get anywhere remotely near the point of glowing!

If there are other LED lights of that type on the same circuit, most probably only the one with the lowest overall Vf is drawing all the available current. Try unscrewing that one and see if another now starts glowing.

It seems less likely that a lamp involving an active driver circuit would glow like this.

I agree with AnAppleSnail regarding dimming a capacitive-input lamp (one that uses a filter capacitor to smooth the AC waveform). Most common incandescent dimmers work by turning on at some delayed point in the normally-curvy-smooth sine wave AC waveform. This delay ideally happens 120 times per second (for 60Hz power), 60 times when the voltage is positive alternating with 60 times when it's negative. This dimming technique basically amounts to PWM (pulse width modulation) for the AC line. Just like PWM dimming in a flashlight, the reduced on-time of the power results in less light being produced. (Yes it is true that the mechanism of perceived light reduction is different for an incan than an LED, but the effect is the same- a perceived dimming.)

Now a side-note:
Please note that the peak-to-peak voltage of a normal 120VAC line is +/- 170V, or 340V. The voltage swings to +170V, then drops back to zero and continues to -170V, returning again to 0. It repeats this process 60 times per second. Half-wave-rectifying results in "85V" pulsed DC, although the peaks are still 170V. (Since only every other peak is getting through the "perceived" voltage winds up being 85.) If you then filter this with a capacitor, you approach pure 170VDC as the capacitor gets large enough. Full wave rectification results in all the peaks getting through, so pulsed 120VDC is obtained. If you filter this with a capacitor, once again you approach pure 170VDC as the cap gets large enough for the current being drawn, which isn't nearly as big a capacitor as for the half-wave. Notice I have rounded, and neglected rectifier diode Vf.
/side-note

Back inside the dimmer, the delay viciously chops the AC waveform, causing a sudden climb from zero to a high voltage (as much as ~170V) in a very short time, happening 120 times per second. Even a dimmer running at 100% will still cause some amount of this sudden voltage jumping, depending on dimmer design.

The capacitor charging surge that happens every half-cycle (for a full-wave rectifier) when the solid-state switch in the dimmer suddenly turns on will be much greater than if the capacitor is allowed to charge normally every half-cycle by not having a waveform-chopping dimmer installed. It seems that there is almost always a resistor of some sort in series with the filter capacitor in these sorts of lamps (to limit inrush current when the lamp is switched on?) so that device does try to limit the current when the lamp is dimmed. Unfortunately the poor little thing overheats, because it is only designed to see the occasional lamp-switching-on inrush, not 120 inrushes a second! That resistor could also be a failure mode, in addition to the capacitor possibly failing. Hopefully they used a flameproof resistor!

I almost burned out a poor little LED night light once by running it on a "modified sine wave" inverter in my car. Boy that limiting resistor got hot! A "modified sine wave" inverter puts out a square wave with sections of zero voltage in between the pulses. (That's a LOT of "modify"ing if you ask me! I think it would better be called a "modified square wave" inverter! So there!!!)

That's the kind of waveform shape we're talking about with most dimmers.
If a lamp has this sort of input circuit, you could however dim it with an autotransformer dimmer or a rheostat (variable resistor), since those leave the AC sinewave waveform pretty much alone and instead adjust the voltage level. If however the lamp has an active driver not designed for dimming you may still have problems. As the input voltage is reduced, as Christexan pointed out, a non-current-limited driver might try to draw more and more current to keep the output constant, eventually burning out.

OK i gotta get going now! :shakehead
 
Last edited:
Top