Osram Ledvance flood light - flashes on and off about twice a second

[SolderSucker]

Osram Ledvance, IP65 (50W). One of the models made in Germany before they were, I think, at least partly taken over by a Chinese company (more recent models are pretty much impossible to dismantle, but my older model isn't too bad).

Just had this fault develop - switch it on and half a second later it switches off, then it switches on again, then off ..........

I suspect a problem with the power supply/driver board but first I checked the 144 LEDs with my multimeter set to diode test - eight LEDs are dead and six are very dim, the rest are okay. Marked each bad one with my marker pen.

LED board markings:

551032371

2835LED144PCS 24S6P (so that'll be 144 of the SMD 2835 LEDs)

I doubt that any of the LEDs are the cause of the issue, or does anything think otherwise?

If they could be the cause then is my best bet to remove each of the dead and dim ones and bridge their solder pads? (or replace with good LEDs if I can get any cheap)

Meanwhile I'll dig deeper into the power/driver board.

Just FYI, here's an image I found online showing the internals of the same model as mine:

https://a.d-cd.net/-xAAAgActeA-960.jpg

[Psi]

Power cycling is very common on failing driver boards.
Often caused by failing caps, a quick check can be to just solder a new cap across the old one and see if that fixes anything. If it does you can go to the effort of pulling the old cap out and replacing it.

The dead LEDs should just reduce the current draw a little.
I cant see that causing a problem unless the driver has some sort of overvoltage shutdown which is getting tripped without enough loading,
But that seems pretty unlikely. The drivers are usually just a simple constant current switchmode circuit.

[RoGeorge]

LED lights drivers often run hot, and fail because of dried/leaked/bulged electrolytic capacitors.
I wouldn't bother testing them one by one, replace all the electrolytic capacitors with new ones.

[SolderSucker]

Thanks both of you - no bulging electrolytics but of course one or more of them could still be bad. Will replace and report back.

[SolderSucker]

Before I replace any caps, some additional info - now that it's dismantled I've attached a plug so that I can power it on with the cover off, etc.

As it flashes I'm noticing that only the top and bottom rows of LEDs are fully lighting up, the other rows are very dim as they flash.

Could that be because of a lack of current draw due to faulty LEDs (the top row has no faulty ones but the bottom row has two, the rest are spread among the other four rows). Or simply because it's powering off before the middle four rows have a chance to power up completely?

[RoGeorge]

Usually the LED plates are in y parallel strands of x series LEDs and no series resistors.  For example, the PCB plate might be marked with something like 8s3p (meaninf 8 series 3 parallel), together with another number like 2835 (meaning 2.8x3.5mm, one of the common sizes for a SMD power LED).  Sometimes the total number of LEDs might be written, too. 
https://electronics.stackexchange.com/questions/422841/how-to-find-a-particular-led-strip

Should be possible to see how the LEDs are grouped from the PCB traces on the aluminium plate.  If the plate lights only partially, that's a bad sign.  Usually it means some LEDs got interrupted (or shorted).

If you turn them off and look at each LED with a magnifier, should be possible to see a black spot present on the face on damaged LEDs only.  If the series string is long, might be possible to just short circuit the interrupted LED with a wire jumper, but that's only a workaround, not a proper fix.

It's almost impossible to desolder a broken LED without a hot plate, but in case you happen to have spare LEDs, you can crash the defective LED with a long nose pliers then solder a new one on top.  In my experience, repaired LED strips will fail again after a while, in another spot.

For the most common lights, you may find to buy the plate with LEDs only, without case and driver.

[SolderSucker]

Can't see much of the tracks on the plate, just those going from the positive and negative inputs. I think they may be in parallel strands across the six rows but not certain.

What voltage should the plate be getting? Decided to measure it with my multimeter and it's fluctuating a lot but not in time with the flashing on/off.

That makes me think it's the driver board at fault, would you agree?

Also, I assume that the plate should be getting DC and not AC? (I know that some LEDs can run on AC). What kind of voltage should I be seeing?

I can see which LEDs are dead and which are just dim by using the diode tester on my multimeter. Not all of the dead ones have a black spot.

Here's a thought - is there anything that I could hook up to the positive and negative wires (when desoldered from the plate) to see if it too will flash on/off?

I don't have any suitable caps right now to replace those on the driver board - I have ordered some but for now I'm trying to be certain whether it's the LEDs on the plate contributing towards the issue.

[RoGeorge]

By plate I mean the Aluminium plate on which the PCB traces for LEDs are glued.  The metal Aluminium is isolated from the LEDs and from the rest of the circuitry.  Whatever voltage is there it's capacitive coupling, and irrelevant for the LEDs.

PCB traces between LEDs are only on the LEDs side of the plate.  The traces should be visible as soldermask bumps/reflections when you shine an external light under a proper angle.  If not visible, test the connections between LEDs with a DMM on continuity.

Once you know how many series LEDs are there, you can disconnect the LEDs from the driver, then apply external DC from a lab power supply.


I would test each LED at half its nominal power or so, in order to find if there are any defective LEDs, because at very low current (when testing with the DMM), the light intensity might be misleading.

Unpower the lamp, and use a Li-Ion battery with a series resistor to test each LED, one by one.  Use some sharp wires or probes, and attach them on isolated handles (if you touch only one end, the LED may light by capacitive coupling to Earth, which means hundreds of volts and can even damage the LEDs by reverse breakdown).

Typical voltage drop on a white LED is about 3-3.2V, while a charged Li-Ion cell is 4.2V.  Read on the box the total power, then divide that to the number of LEDs to find LED power, so to find out max LED current, then to calculate the series resistor.  Or you can just go with a safe bet of 50mA (20 ohms in series with a 4.2V Li-Ion battery).  Power LEDs, they are very bright even at half power, so you may want to avoid looking directly to the tested LED, wear dark glasses or obdtruct the LED before testing.

Don't use a lab power supply on current limiting, use a battery.  Most lab power supplies are not really floating (they would light or fry a LED only by touching a single wire (by capacitive coupling with the ground), plus that the lab supplies have output capacitors that can discharge very high currents through a LED even when a current limit is set (because in many supplies the output capacitor is placed after the current sensing).  If you think your power supply has a true floating channel (most don't), then set it on 4.2V/50mA but also add the 20 ohms resistor in series with the lab supply, or else the output capacitor might damage the tested LED.

[wraper]

As 14 LEDs are bad, it's pretty much trash that is not worth repairing. I bet 100% of LEDs have cracks in the yellow compound and their further failure is only a matter of short time.

[SolderSucker]

As 14 LEDs are bad, it's pretty much trash that is not worth repairing. I bet 100% of LEDs have cracks in the yellow compound and their further failure is only a matter of short time.

Perhaps I should replace them with a single LED COB chip?

[wraper]

I don't think it's worth it. Cheap Chinese LED modules are garbage assembled from factory rejected LEDs and their power ratings are inflated. But for decent quality stuff you need to pay quite a bit more. Ledvance 50W floodlight costs about EUR 22 https://www.amazon.de/-/en/LEDVANCE-Floodlight-Light-Outdoor-Applications/dp/B07TTRHMW3 and less known brands can be found for half of that. While I myself tend to spent too much time/effort on repairing cheap things which are totally not worth it, I consider this repair totally pointless even by my standards.

[wraper]

And it's actually better to buy something with high efficiency like this https://www.amazon.de/-/en/HOFTRONIC-Outdoor-Spotlight-Floodlight-Headlight/dp/B08BHP3FGS which with 30W is brighter than most of 50W floodlights. And as another bonus it will heat less and should last longer.

[SolderSucker]

Thanks, got it working now - I removed every dead LED and put a solder bridge across each contact and it works great now (minus a few LEDs). Think I'll also remove the few that are very dim using the diode test.

Also checked the rest of the LEDs - diode test is fine and no apparent cracks. I guess it won't be long before it fails again but at least it's had its useful life extended.

Thanks all!

[wraper]

Also checked the rest of the LEDs - diode test is fine and no apparent cracks.

Use some sort of magnification. Those are hairline, so not very obvious on a first glance. Very obvious with microscope but magnifying glass should be enough too unless you have really poor vision. I have pretty good vision so can somewhat see them with naked eye too.

[SolderSucker]

Also checked the rest of the LEDs - diode test is fine and no apparent cracks.

Use some sort of magnification. Those are hairline, so not very obvious on a first glance. Very obvious with microscope but magnifying glass should be enough too unless you have really poor vision. I have pretty good vision so can somewhat see them with naked eye too.

My powerful stereo AmScope thankfully showed no cracks.

[wraper]

Then maybe they did not ran that hot and there was some other failure mechanism. Still if this many failed, they will continue failing as they have the same wear.

[SolderSucker]

Then maybe they did not ran that hot and there was some other failure mechanism. Still if this many failed, they will continue failing as they have the same wear.

I suspect that the LEDs have been failing for years but it reached a point where one kicked off the on/off problems.

This will be the beginning of the end for it of course but as long as it works and can be repaired I'll keep using it.

Maybe I'll replace the LEDs with something else one day if anything cheap and viable catches my eye.