Industry LED Luminaries, damage of some LEDs in...

[KrS1990]

Hello,

we have many LED luminaires with some problem (possibly caused by periodic work on generator of our factory), when dali dimming is set to to 100%, luminaries works fine, but after dimm them to 10-20% (dimming by dali protocol), some LEDs in one matrix glow darker or completely go out. They dont flickering. LED DALI driver method: constant current.

Below i will show you an example:

https://Upload the image directly to the thread.com/a/Ffu7x5c

The LED in green boxes works fine after dimming, the leds in red circles works badly

 

[pc1966]

I suspect they are not really designed for dimming. If they have a few sets of LEDs in parallel then once the overall constant current gets low you might find a significant imbalance forms between internal paralleled sets and so some get much darker than other.

What do the manufacturer's claim for dimming use?

 

[KrS1990]

@pc1966
At first, thank you for trying help me

And answer for your questions. 100% sure, they are designed for dimming, but they were damaged by AC generator which we used for 3 months when we had to wait for connection to local electricity network.

Perhaps the parallel led lines could has problems like you wrote but then all SMD chip should work badly in line.

It seems to be like, some chips were damaged by voltage drops or higher harmonics, and now i want to diagnose what happened with LED chips. Do they have destroyed junction, or p-n junction has been destroyed or has lost its properties and then Bang gap is higher than for good, healthy chips?



"What do the manufacturer's claim for dimming use?"
They have no idea what happened with smd chips.

 

[pc1966]

If the LEDs have been subject to overload it might change the operating point slightly whihc might just explain it, though usually they simply die!

 

[KrS1990]

Yes, that is some incredible fact, they are still working on 80-100% forward current but they light darker or generally go out on 1-15% nominal forward current.

In fact i'am looking for what is damaged, pn junction, micro resistance of wire bond, anode or cathode, damaged structure of cristal structure of pn junction. The manufacturer guarantee is one case, but the second, is abnormal LED chips work - I ve never seen before and any other engineer... So thats why we are trying
establish fact and reason.

The diode manufacturer said, the test in their laboratory will cost over 40 000 USD.

Of course, i sad that diodes works fine at 80-100 forward current but... The humans eye cant deceive us, and it is very possible that we can't register abnormal damaged LED work.

I will try to send luminaries to laboratory with goniophotometer to measure total flux, and will try to measure forward voltage of one diode. But it is very interesting, and looks like the damaged diodes have changed gap bands width of pn junction, so that's why phenomenon of electroluminescence is so weak with lower forward current. Maybe the material has changed their own special properties , and pn junction has much bigger resistance. This fact could explain why diode work, not flickering but glowing is darker than healthy diode - the LED diode resistance is not linear like resistors, has a greater dynamics of change like below:





Currently (i think) the characteristics has changed and resistance changing like voltage at lower current value - the changes are characterized by step change. It is linked with changed gap band what could have influence for characteristics and for example some part of electrons recombination is changed to auger effect, so we lost energy, diode emitting more heat, and much less photons.

So maybe it is simple way to check smd diodes - power the health and damaged diode with the same current 1/10 (nominal) and measure voltage drops, heat and flux. Of course the result wouldn't give an answer what was destroyed but give me some evidence, is it good or bad experiment direction.

** I don't know what is the energy of auger electron compared to photons in the same material, semiconductor type.