Strange issue with LED Flood Lights

[jtsylve]

I've got a weird issue at home that is driving me nuts.  I've got a pretty standard LED flood light with one of those sensors that has a PIR and also photo diode to trigger the light on and off.  It was originally wired as in attachment [1], where the sensor array and the LED driver share a neutral line and the sense output from the sensor switches the live line to the LED driver.  After a while, I started to notice that the lights were blinking at night and you could hear an audible click from the relay in the sensor. 

I thought it might be feedback from the light reflecting confusing the sensor's day/night cycle or that the sensor array might be going bad, so I first bypassed the sensor as in attachment [2] and instead hooked the live input of the LED driver to a switch so I could manually turn the light on and off.  This works fine when the switch is on, but somehow, at night the LEDs were blinking on and off when the switch was turned off.  I could still hear the click from the relay in the sensors.  I was very confused by this, since when the switch is off there should only be the neutral wire connected to the LED driver with the Live wire floating.

I then cut the sensor array out of the picture entirely as in attachment [3].  I now don't get a relay click (because it's completely unpowered), but the LEDs still occasionally blink on and off at night when the switch is off.  The weird thing is it's also not consistent.  Some times it happens and sometimes it doesn't.  It's possible that the problem isn't limited to night, but that's only when I notice it.  I've got the light hooked up to my quick test right now with just the neutral and ground wires connected, and as you'd expect the device isn't powered.

What could be going on here? 

[jtsylve]

I also just tested the sensor array and the relay does click when power is added/removed.  It's got to be something to do with the wiring in my house, but I have no idea what would cause it to blink with the switch turned off, unless there's some induced current somewhere/somehow.

[Specmaster]

Have you checked the voltage between Ground and the Neutral and I take it that the Ground is also connected to the LED driver and the floodlight?. Test the integrity of the Ground connection and also check the driver and floodlight for Live to Ground leakage.
 

[jtsylve]

Will do that as soon as I get back home (multimeter was in my office).  The ground is only connected to the metal case of the light itself, and not the driver or sensor.

[capt bullshot]

I have no idea what would cause it to blink with the switch turned off, unless there's some induced current somewhere/somehow.

Yes, most probably. I'd guess on capacitive coupling somewhere in the wiring, e.g. a cable that has another live wire and your switched wire to the LED light. Some of these LED lights are just high impedance up to a certain voltage and then start operating at higher voltage. Since they rectify the line AC to DC and may have a bulk capacitor for smoothing the rectified AC, this capacitor slowly charges from the coupled current, and once the voltage reaches the start threshold, the light starts to operate and discharges the bulk capacitor, since the coupled current is way too low to operate the light continously. That's what causes the blinking. One can add a bleeder resistor in parallel to the light to suppress this behaviour at the cost of wasted energy when the light is turned on.

[onesixright]

Maybe far fetched. Checked for any crack (PCB, components pins)?

Maybe the wind or other form of movement trigger this? I would inspect the PCB and have a good look if anything is broken, you can carefully try to flex the PCB and see what happens.


Sent from my X using Tapatalk

[jtsylve]

PCB looks fine and shouldn't even be powered when the switch is off at this point.

[jtsylve]

Yes, most probably. I'd guess on capacitive coupling somewhere in the wiring, e.g. a cable that has another live wire and your switched wire to the LED light. Some of these LED lights are just high impedance up to a certain voltage and then start operating at higher voltage. Since they rectify the line AC to DC and may have a bulk capacitor for smoothing the rectified AC, this capacitor slowly charges from the coupled current, and once the voltage reaches the start threshold, the light starts to operate and discharges the bulk capacitor, since the coupled current is way too low to operate the light continously. That's what causes the blinking. One can add a bleeder resistor in parallel to the light to suppress this behaviour at the cost of wasted energy when the light is turned on.

Possibly, but in its original configuration the switch wire wasn't connected at the fixture, since it originally came as wired in [1] and the switch wire is not connected to the sensors, so it doesn't explain the relay clicking.  Something odd is going on indeed.  Hopefully I'll have more clues when I can take some measurements.

[capt bullshot]

Possibly, but in its original configuration the switch wire wasn't connected at the fixture, since it originally came as wired in [1] and the switch wire is not connected to the sensors, so it doesn't explain the relay clicking.  Something odd is going on indeed.  Hopefully I'll have more clues when I can take some measurements.

Then also check for live and neutral been swapped by accident - if the switch / PIR has live connected where the neutral is supposed to, all kinds of capacitive leaking may occur and cause the blinking and clicking.

[jtsylve]

Measurements seem ok

N -> G = ~250mV AC
L -> N = ~ 120V AC
L -> G = ~ 120V AC

The odd thing is connecting the meter only to the neutral line measures 0.8V AC between neutral and the air.  Could this difference lead to trickle charging capacitors and causing the blinking as described when the switch is tuned off?

All connections at the switches seem normal.  It's a 4 switch box.  All neutrals are tied together, all grounds are tied together and the live/switch wires are tied as you'd expect.  Nothing measures like there's a switchup between live and neutral.

[Specmaster]

Be careful here, what is the meter that your using to make these measurements, depending on the meter, those kind of voltages, especially in free air are understandable and may well be normal. I was going down the line of the high impedance on the LED's and or driver which if there was a leakage, charge up the capacitor sufficient to cause the LEDs to flash, discharge the cap and then the cycle repeats. Have you checked the resistance between the ground connection on the fitting and the neutral and also the ground and live? I believe that is more likely to slight tracking from ground to either the neutral or the live at the fitting and a bad earth connection somewhere along the line to the mail earth point.

Most likely is a build up of the leakage to ground of the various filter capacitors on many items these days could well be dumping a fair amount of mA to ground and if the main ground connection, be it a rod in the ground or the sheathing of your incoming supply

EDIT

I just did a test using my Brymen BM867 sitting here at my bench I can get 80mV AC with both the leads in free air, if I touch either of the probes this raises to 830 mV and the same if I connect with neutral so as I was saying earlier, those voltages are to be expected

[Psi]

When LED drivers wear out they sometimes start turning off randomly for a short time.

I think its the switchmode IC overheating and restarting.

[tooki]

When LED drivers wear out they sometimes start turning off randomly for a short time.

I think its the switchmode IC overheating and restarting.

No, the issue is that the LEDs blink on with their power turned off.

[Specmaster]

When LED drivers wear out they sometimes start turning off randomly for a short time.

I think its the switchmode IC overheating and restarting.

No, the issue is that the LEDs blink on with their power turned off.

That's why it just has to be something going between the Ground and Neutral and the leakage is slowly building the charge on a capacitor which then discharges via the LED's. My money is on the Ground connection not actually being 100% connected to actual Ground. You say thats all connected back at the distribution box, have you traced the main incoming earth to the box? does it bond onto a SWA if the supply to the building is fed that way or does it go to an actual Earth stake in the ground. If its the latter, throw a few buckets of water around it to give it a better connection with the ground.

You said that it does it at odd times, so maybe there is a higher impedance connection Ground somewhere and also you have something that is leaking to ground via filter networks to Ground. I once had a problem with a dealing room of a large financial company in London where they had loads of PC's being fed via 30mA RCBO's and when they were all active, the main breaker for the floor would trip out. I traced that down to the total amount of leakage via each PC's filters on their PSU's adding up to enough leakage on the Ground to trip the breaker. Replaced that breaker with a 300mA sensitivity trip cured the problem. Something similar may be causing your problem as the current draining away increases, because of the higher impedance somewhere on the Ground path, the voltage is being raised at odd times above zero. But for that to cause your issue with the LED flood, there has to a Ground connection between the Ground on the light / driver and the Live (Hot) connection so I would really investigate that area first. Even though the Live is switch off, it could be back-feeding slightly enough to charge the cap up via the Ground connection. If the driver and light are out of normal reach of people, you could isolate the Ground connection at that point and see if the problem manifests itself still. It would certainly prove or disprove  this theory.

[Psi]

hm.. strange.

I did once seen a LED light with built in dimmer where the 'off' state (dimmer set to zero) was not actually 0v but just low'ish volts a bit below the LED Vf.
Sometimes it would drift high enough to turn the leds on dimly.

[floobydust]

This is weird but I have seen a ghost flashing LED fixture when power was off.

It turned out to be the on/off switch was switching neutral, not hot.
So the SMPS always had hot (but not neutral) at the AC input. I found it was the Y-capacitor ended up slowly charging the main filter cap until the SMPS tried to start up and give a flash and it kept repeating that every minute or so. I found it by measuring the DC bus voltage and seeing it slowly climb up.

Cheap outdoor PIR sensors get false triggered by RF from smart meters 900MHz at the house or from the neighbor's place. I put an aluminium foil cover over it to narrow that one down.

[james_s]

Do you have an illuminated switch or one that has any sort of electronic feature? I had a CFL on an illuminated switch years ago and found the CFL would flash about once every 30 seconds. The current through the little neon bulb that lighted the toggle was just enough to gradually charge up the bulk capacitor in the CFL until the lamp would start, it formed a relaxation oscillator.

[jtsylve]

Everything seems to point to there not being issues in the wiring.  I'm assuming it's just a capacitive issues with the wiring (similar to [1]) still powering the circuit slightly.  I've added a 470K resister across the output of the rectifier circuit.  The light still works when powered.  Hopefully this will be enough to discharge the stray voltage when the switch is off.