Can't Switch Light Off

[BlueDragonfly]

We have just installed a series of lights illuminating an internal stair. The lights are small rectangular banks of LEDs (connected in parallell) operated by a two way switch (one switch at the top of the stair, one at the bottom). The lights work wonderfully however one of the lights (the one at the top of the stair and closest to the power supply) only dimms when either two way switch is thrown, whilst the other two lights switch off successfully.

I have undertaken the following diagnostics already without success:

1. Called the lighting supplier to ask if this was a common problem with these lights

2. Tried replacing various wires and switches

3. Tried bypassing the first light - intersetingly this resulted in both of the remaining lights only dimming when they should have been turning off.

4. Checked some of the more accesible junction boxes and switches for a short or bad earth.

5. Connected the circuit to an independent circut from the same main swiich board.

I have successfully connected the LED fitting directly to the same power source using a simple single switch setup and the light worked successfully. When I reconnected the single light to the two way switch system and flicked it off, the light only dimmed again.

Any assistance would be appreciated.

 

[TicklyT]

Assuming the 2-way switching works correctly with an incandescent light connected in place of the LED lights, I would begin to suspect capacitive coupling between the conductors in the switch wiring.

It is caused by the voltage in a 'live' conductor in a cable inducing a small voltage into other nearby conductors. It happens on all cables to a greater or lesser extent.

It is not usually a problem with a load like an incandescent lamp, which easily dissipates the induced voltage, but as LEDs and compact flourescent lamps require so little power to run them, it can cause them to light when you don't want them to.

 

[BlueDragonfly]

Thanks for your swift reply Tickly T. I feel much relieved now.

I tried an incandescent and, as you speculated, it did not glow in the off position. I then decoupled the two core switch cable by adding an temporary cable and ensured that the two were spaced well apart. I recall from high school physics that the induced current drops off something like 16 times each time you double the separation distance.

There is still some glow. Not as much as before so we are definately onto something. What I don't understand is that in a live switching loop, when either switch is thrown, there shouldn't be any current in either of the two switching wires. So if there isn't any current why is there an induced current?

The house does have a residual current device (safety cut out switch) which would pass a small current through all circuits even when dead so perhaps that is the explanation and perhaps that means I can't get rid of the glow.

I could try to add another isolating switch but I would strongly prefer not to. Is there some kind of solid state device that I can add to the circuit to rectify the situation locally?

 

[breezer]

can you swap one of the led lights for another one?

i.e take the furthest one and swap it with the glowing one?

and as its dc, how about trying a diode across the supply? (to the actuall led)

you have to put the diode across the supply backwards, or else it will be a dead short

 

[BlueDragonfly]

Thanks Breezer,

I could do the swap but I am assuming from my first test where I took out the first one entirely and the others glowed that they are all the same and that the problem is with the circuit and not the lights.

I think the lights are AC as there is no visible transformer and the housing reads 240V 50Hz MAX1.2W.

Thanks for the diode suggestion. Will it still work if the lights are AC and how do I work out the diode rating (impedance is it?)? The lights are connected in parallel.

 

[breezer]

you should not light leds with direct ac, i bet your little box has a rectifier in it.

any cahnce of a pic or link to where it came from?

 

[bernardgreen]

Ad a low wattage incandescent bulb to the circuit to drain the capacitive coupled energy from the circuit.

There are alternatives ( such as carefully selected resistor and/or capacitor across the LEDs ) to drain the energy but the above is the easy DIY solution.

 

[BlueDragonfly]

Thank you all. A picture is available at http://web.aanet.com.au/dragonfly/Transfer/led housing.jpg I am guessing that the brown thing is the rectifier.

The low wattage bulb is an ingenious suggestion although we have selected LEDs for the stair as they use little power and we may elect to leave them on at night as our children get to the stage of wandering around the house at night. Adding a globe or resistor to the circuit seems to defeat that idea somewhat.

 

[breezer]

url doesnt work (you can use the edit / preview buttons to correct it

 

[BlueDragonfly]

Sorry. url should work now. Thanks for the edit tip.

 

[bernardgreen]

Thank you all. A picture is available at http://web.aanet.com.au/dragonfly/Transfer/led housing.jpg I am guessing that the brown thing is the rectifier.

You guess wrong. The brown thing is a capacitor which will be in series with the mains to pass a small amount of charge onto the LED circuit each hafl cycle of the mains 50 Hz.

The rectifier is probably the little black thing

 

[breezer]

now i see.

a diode will be of little / no use

but just humour me swap 2 fittings over

 

[BlueDragonfly]

I have now placed the last in place of the first. Both with and without the last two lights connected.

The problem persists, although interestingly in the case of all lights connected the problem does follow the first light fitting.

I can sumarise all of the test configurations as follows. If the original configuration was [1] [2] [3], numbers being the fitting and the first position being closet to the power source, performance with either two way switch thrown to the off position is given below. [x] is a disconnected fitting

a [1] [2] [3] : [dim glow] [off] [off]
b [x] [2] [3] : [off] [dim glow] [dim glow]
c [1] [x] [x] : [dim glow] [off] [off]

With breezer's cue I have now found the following.

d [3] [x] [x] : [dim glow] [off] [off]
e [3] [2] [1] : [off] [off] [dim glow]

conclusions
(i) I should have listened to breezer's original sugestion;
(ii) maybe fitting 1 is defective, although that does not explain configutation b

 

[BlueDragonfly]

I have some new developments.

The supplier of the lights was good enough to offer some technical advice and three replacement fittings, which I will call 4,5 and 6. They initially suspected the capacitors for holding charge. This was studied by waiting for them to dim - they don't even after about 15mins and also switching off the mains - the lights do go out. Hence it does not seem to be the capacitors.

I have placed a schematic wiring diagram at http://web.aanet.com.au/dragonfly/Transfer/2wayswitchedLEDs.pdf

The following observations were made whith either switch in the off position:

f [3][2][1] : [dimglow] [dimglow] [dimglow]

not sure why this is diferent to e

the suspect number 1 light was then substituted for a new light, 4. Result is the same ie

g [3][2][4] : [dimglow] [dimglow] [dimglow]


This was repeated four times with four different sets of switching wire (1,2 to 1,2 on the diagram). Three of the switching wires were the required length however the switches were placed closely together for the purposes of the tests. Result is the same for all of these ie. [3][2][1] : [dimglow] [dimglow] [dimglow]

The fourth switching wire was only about 400mm long. With this wire it all works At last
[3][2][4] : [off] [off] [off] But why only for the short wire T his doesent really help me.

We then did some more studies with the troublesome long switching wire connected.

Power was taken for a completely separate circuit in the house. Same result.

h [3][2][4] : [dimglow] [dimglow] [dimglow]

Finally a completely separate circuit was built, powered via a previously unused power circuit in the house, using the long switching wire. Result is the same, ie

i [5] : [dimglow]

Please help.[/img][/url]

 

[breezer]

you only have one thing left to try

take them all out (really, all of them)

put them some where else, i.e in your shed / garage (just sitting there in a line) wire them up, and plug them in, see what happens