LED T8 Tube light Many failures, WHAT????

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
The lights are 48 inch T8 size, each with 96 LEDs on a circuit board strip, divided into 3 series strings of 32 LEDs in series across the supply, which is also part of the assembly. The supply voltage is listed as 100 to 277 volts, output power 20Watts.
The tube assemblies are failing! so far we have had to replace 20 of them, and 3 more have failed. The failure starts as one of the groups flashing, then 2 or all 3 groups flashing, and then the whole thing either goes out completely or else glows vary dim, just barely visible.

These units are mostly not available because the AC power is fed at opposite ends, and so if a person were to put a hand on one end while installing it they might get a shock. So the result is that only one store sells them. LED lamps are supposed to last many years, these have failed after less than one year of only being on 6 hours each evening.
Does anybody have a clue? The supply voltage is 123.9 to 124.1 volts so well within the range.
 

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
Max,
None of the lights in that link are similar. These have an aluminum extrusion and a clear colorless plastic cover. The LEDs are on a narrow strip of PC board that slides into the heat sink, the power supply is in the hollow back section of the heat sink. AND all of those lights in that link cost a lot more than the ones that are failing. The new ones were installed by an alleged electrician who did OK work but charged outrageous prices. So they are in new fixtures with new wires. The problem is the early failures and how to prevent them. Replacing 20 of them is not cheap.
 

Wolframore

Joined Jan 21, 2019
2,609
part numbers? brand? any information about them?

For your immediate problem is there a warranty that covers them lasting only 1 year? I have 4 lithonia branded lights in my garage and they sent me replacement LED boards for free when I had a couple dim LED's on a board. They have a 10 year warranty and could verify that the model hasn't even been out that long sending them pictures of the failed boards with their part numbers.
 

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
part numbers? brand? any information about them?

For your immediate problem is there a warranty that covers them lasting only 1 year? I have 4 lithonia branded lights in my garage and they sent me replacement LED boards for free when I had a couple dim LED's on a board. They have a 10 year warranty and could verify that the model hasn't even been out that long sending them pictures of the failed boards with their part numbers.
These LED tubes have NO BRAND NAME listed at all. And the people at the store are rather hostile, and if these tubes were available anywhere else I would go to that store instead. The packaging contains no printing at all. Lithonia is a reputable brand with a good reputation and a lot of integrity, totally different from this brand.
Any warranty expires as the cash for them is paid to the cashier. So really they are not the high quality $15 to $18 tubes. I am asking about what anybody may have discovered about how to prevent failures, or if there is any simple trick to repair them. These look different, since with the crystal clear tubes the individual LEDs are visible, while with those other types the look is a uniform glow, which is not what we wanted for the outside lights.
 

oz93666

Joined Sep 7, 2010
739
Sure .... the units , like most , are DESIGNED to fail .... The company won't get rich if their product lasts 10 years ....

After trying a wide range of items I realized none are designed to last ....

You have to build your own . I use self adhesive strip led as the basic component ....

But even these don't last long if powered as suggested with 12V ... running at half recommended power , about 10.8V keeps them cool and so long lived ...

If you like the package type you have then you have to open it up and reduce power to led by half some way .... eyes work logarithmically so won't perceive the light output too much reduced.
 

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
Interesting concept indeed, Oz. I could bypass the internal supply and run all 50 of the lights from a single DC supply, with a dropping resistor for each device. I would also need to add a blocking diode to prevent damage if a bulb was installed backwards. 50 lamps at 20 watts each, which is what they are rated, is only 500 watts, or 5 amps at 100 volts. The challenge will be to find a transformer to deliver the right voltage, since the peak voltage is what will be what I need to limit. But 5 amps is not that much current to regulate. Now you have a really good idea started. Thanks!
 

Tonyr1084

Joined Sep 24, 2015
7,853
I've had a few LED lamp failures myself. They've been different brands and such. The LED's are going to last 10 years, but what is questionable is the electronics that power them. One of my early failures - I opened it up and found a main power resistor (low resistance) had burned out. That's why the unit failed. Another dimly lights when switched on, but when switched off it glows brightly for a moment then slowly dims out to nothing. LED's themselves do not behave like that. It's "Probably" most likely an electronics failure rather than the LED chips themselves failing.

I'm not sure how 10.8 volts is half the power of 12 volts. Maybe it is, maybe @oz93666 knows more about it than me. Your failures may be due to heat. Especially if they're in a poorly ventilated location. But even that's an unlikely scenario. I would wonder about power spikes. When heavy machinery kicks on - air conditioners - air compressors - etc. could be causing the problem, especially for cheaply designed units.

Replacing them may be expensive. Even more so if you go with the more expensive units. But I've found that when you replace with 10¢ parts you have 20 or more failures than when you replace with $1.00 parts. In the end the cheap parts can be more expensive. I said "CAN BE". Someone else may disagree. OK, if that's so then - - - .
 

Tonyr1084

Joined Sep 24, 2015
7,853
The challenge will be to find a transformer to deliver the right voltage, since the peak voltage is what will be what I need to limit.
I have a transformer in my bin rated at 68 VAC (with center tap). That comes to a peak of around 95 volts. It came out of an old stereo. Just saying.
 

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
I have a transformer in my bin rated at 68 VAC (with center tap). That comes to a peak of around 95 volts. It came out of an old stereo. Just saying.
I also have a collection of transformers, but 500 watts is more than many of them can provide. So I may need to think very carefully. The lights that fail are mounted on the underside of a wide roof overhang around 3 sides of a take-out restaurant, so they are well vented and rather well out of the weather..
 

Sensacell

Joined Jun 19, 2012
3,432
The typical LED failure you describe is Die Bond / Wire Bond failure.
Companies making LED's often don't make the wafers, they just buy them, dice them up and package the dies.
This fault can also be caused by poor temperature profile control during assembly/reflow.

A common misconception is that the "flashing LED" failure is always caused by the driver circuit failing, when in fact it's often the LED's themselves causing this. Of course, there are other possible driver faults that can also cause flashing.

A "thermal oscillator" can form when a wire bond fails, the poor connection heats up, the circuit opens, things cool, the cycle repeats.
The physical feature size is small enough so that the frequency of the flashing can be quite high, up to several Hz.

Poorly made LEDs like this are a disaster waiting to happen, trying to replace individual LED's in a fixture is a waste of time, the next crappy LED in the string will just fail a week later.
 

oz93666

Joined Sep 7, 2010
739
..... The LED's are going to last 10 years, but what is questionable is the electronics that power them. .
Well the leds can last ten years with an exceptionally good heat sink and not pushing them too much current wise ....

But a good heat sink is expensive and not easy in an enclosed device ... so the manufacturers cut corners .

My experience is that leds are the main issue , in the led strip pictured above the leds do fail in blocks of 3 (that's how they're wired ) even if the power is fine.

I'm sure if run in Alaska with 0C ambient temperature they would last 10 years


I suspect because of corner cutting design most leds are around 100C giving a life close to 1000 hrs.
 
Last edited:

Sensacell

Joined Jun 19, 2012
3,432
Well the leds can last ten years with an exceptionally good heat sink and not pushing them too much current wise ....

But a good heat sink is expensive and not easy in an enclosed device ... so the manufacturers cut corners .

My experience is that leds are the main issue , in the led strip pictured above the leds do fail in blocks of 3 (that's how they're wired ) even if the power is fine.

I'm sure if run in Alaska with 0C ambient temperature they would last 10 years
Good design and good quality- two things that don't always converge.
 

oz93666

Joined Sep 7, 2010
739
...I'm not sure how 10.8 volts is half the power of 12 volts. Maybe it is, maybe @oz93666 knows more about it than me. - .
I got that figure from experiment , Tony.... powered the strip with my bench supply at the recommended 12V , read the current , hence the power at recommend voltage ..... then turned the voltage down till the new V x I figure was half ... usually about 10.8 V ...

The strip is composed of many units of 3 leds and resistor in series.

I figured by halving the power (heat) input should increase the life of the leds by about 5 times .... and it won't reduce light output by a half !! Leds are more efficient at lower currents...half power gives about 60% lumens you get at full power.
 
Last edited:

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
OK, and probably good news that the flashing could be the LED and not the supply. That does explain why one section fails while the other two remain lighted. So now the challenge will be to find a way to rob good ones off of a failed string to replace the ones that have failed. These LEDs are surface mount with very little solder showing. So unsoldering without destroying will take some practice.
 

Tonyr1084

Joined Sep 24, 2015
7,853
well vented and rather well out of the weather..
OK, now I'm wondering if temperature fluctuations may be at play.

I've not messed with long strings of LED's except for the color changing strip I put in my spa/bathroom (has a jetted tub). Those are powered from a 12 volt source along with a controller that changes the current to each color group causing sweeping through the color range. The LED's I've opened after failure have been the standard lightbulb replacement type, and every one of those has failed due to power supply. I can only speak from experiences I've encountered.

Logic would seem to agree that if a segment in a lamp were to go out while the rest remained lit - I would agree it's not power supply.
 

SLK001

Joined Nov 29, 2011
1,549
The problem is with the LEDs themselves. In order to compete with incandescent and flourescent lights, an LED has to be pushed to its limits and many just die of fatigue. Instead of putting more LEDs in a strip and driving them at 65 to 75% of rated max, manufacturers are putting 80 one watt LEDs in a strip and driving them all to 1 watt, or their maximum rated output. This virtually guarantees that weaker LEDs will start to fail. Since a lot of LED chips have a parallel zener that turns on when the chip fails, if the controller isn't "smart", more current is now fed to the remaining LEDs, pushing them above their rated output. This forces more LEDs to fail, freeing up more current to be pushed to the remaining, etc, etc - until you have total failure.

TVs (with LED backlights) have this problem. An LED backlight is not near as reliable as the old CCFL units.
 

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
OK, now I'm wondering if temperature fluctuations may be at play.

I've not messed with long strings of LED's except for the color changing strip I put in my spa/bathroom (has a jetted tub). Those are powered from a 12 volt source along with a controller that changes the current to each color group causing sweeping through the color range. The LED's I've opened after failure have been the standard lightbulb replacement type, and every one of those has failed due to power supply. I can only speak from experiences I've encountered.

Logic would seem to agree that if a segment in a lamp were to go out while the rest remained lit - I would agree it's not power supply.
Temperature changes may be involved, for sure, since inside the building are several fixtures with the same tubes that burn for a lot more hours every day, and those tubes have no failures. So even though these tubes have an aluminum heatsink back for the LED board it is true that outside temperatures vary far more. But how does one control the temperature swings of lights mounted outside?
 

Tonyr1084

Joined Sep 24, 2015
7,853
how does one control the temperature swings of lights mounted outside?
Coping with outdoor extremes may require a more robust designed lighting system. it may cost more but in the long run it may be worth it. But nobody can tell for sure; unless you can get some sort of guarantee. Perhaps class III (class three) electronics such as is employed by the Aerospace industry, Medical and Automotive.
 

Thread Starter

MisterBill2

Joined Jan 23, 2018
18,179
The problem is with the LEDs themselves. In order to compete with incandescent and flourescent lights, an LED has to be pushed to its limits and many just die of fatigue. Instead of putting more LEDs in a strip and driving them at 65 to 75% of rated max, manufacturers are putting 80 one watt LEDs in a strip and driving them all to 1 watt, or their maximum rated output. This virtually guarantees that weaker LEDs will start to fail. Since a lot of LED chips have a parallel zener that turns on when the chip fails, if the controller isn't "smart", more current is now fed to the remaining LEDs, pushing them above their rated output. This forces more LEDs to fail, freeing up more current to be pushed to the remaining, etc, etc - until you have total failure.

TVs (with LED backlights) have this problem. An LED backlight is not near as reliable as the old CCFL units.
All of the computer monitors that I have had fail to light had CCFL backlights. Probably TVs are a different quality.
 
Top