5mm White LED Fade Test done!

JohnR66

Flashlight Enthusiast
Joined
Aug 1, 2007
Messages
1,052
Location
SW Ohio
I completed an 8 day (192 hr) test of 5mm LEDs driven at 30ma. Regardless of their rated current, I consider maximum current to be 20ma, so I am over driving them by 10ma. In fact, one LED rated at 30ma showed notable change.

fadetest.jpg


Results: (column 1 is over drive, column 2 is control - not driven and pre matched before test)

A. Asia Engineer's (ebay) 13,000 mcd white. These LEDs have a greenish blue tinge and faded very little.

B. Superbright LEDs 18,000mcd 15 Deg white. Faded a bit but had a notable purple shift.

C. Sylvania X-mas light string hacked flange-less wide angle white LED is only 3mm, but these are the brightest and fading was non existent (or very slight). The clear winner!

D. Warm white hacked from string purchased at Menards. These are nice and bright WW, but suffer significant fade in this test.

(Not Shown) From ebay seller cece718. These had a blueish green tinge and it faded. Not nearly as bad as the WW LED, but quite noticeable. Another batch from the same seller is more of a purplish tinge and faded the same. (I tested several from each batch for color. Each batch consistent.

Conclusion. I don't have the equipment to post actual numbers, but all I really need to see is the visible fade. The camera actually shows it more than by eye.

I was very surprised to see a x-mas string 3mm LED show no fading. These are also brighter than any of the others. I expected a small 3mm LED to fade faster due to lower thermal dissipation. I guess sylvania puts good quality LEDs in their strings. Recommended!

Next I'm going to turn up the heat:crazy:. They are at 40mm to make them suffer.:popcorn:
 

JohnR66

Flashlight Enthusiast
Joined
Aug 1, 2007
Messages
1,052
Location
SW Ohio
Re: 5mm White LED Fade Test Update

Test update
I know I said I was going for 40ma, but I'm cooking them at 50 and 60ma for the last couple days.

The little 3mm Sylvania X-mas string LED hung in there at 50ma, but at 60ma it started to fade over a day. No color shift, just reduced output.

The superbright LEDs 5mm 18,000mcd 15 Deg. has turned a stronger violet color. It holds on to second brightest.

The Cece718 LEDs are both have faded more and are rather dim now. One stopped outputting light below .5ma while the others light.

The warm white faded a bit more, but seems to have stabilized.

And now the amazing. The "Asia Engineer" LED has not color shifted or faded much at all. Almost imperceivable. It still belts out the bluish green white as if new. Too bad about the color cast of that batch and it being only 13,000mcd as the seller rates it's output. Otherwise a very rugged LED for overdrive.

Now for the final roasting. They are all in series with a 100 ohm resister. My regulated PSU is cranked up to its max 30 volts. The LEDs are getting 80ma!

I have some "Light of Victory" LEDs on the way. They cost much more than the 100 pcs for $10 free ship deals on ebay, but are claimed to be 55,000mcd. I also have some 280kmcd multichips on the way. I'll see how they compare to my other LEDs and pass on the info.
 
Last edited:

JohnR66

Flashlight Enthusiast
Joined
Aug 1, 2007
Messages
1,052
Location
SW Ohio
I finished roasting them at 80ma. There wasn't much point to this other to see what would happen. No one would run 5mm LEDs continuously like this in an application:
80ma.jpg


The top row is the unused control group, the bottom row is after all the tests performed earlier in this thread plus the 24 hour 80ma run. Perhaps I underexposed too much, because they all still work, but suffered heavy lumen depreciation except for the "Asia Engineer" LED on the far left that did fade, but not nearly as much. The little 3mm started to turn brownish over the die due to burned epoxy, I'd guess. I'd still give it the crown for overall output and longevity for moderate "push" currents (30ma).

I received my ebay "Light of Victory" LEDs. Here are my first impressions.

55,000mcd 5mm white. Much brighter than the cheap 100qty for $10 ebay LEDs I've tried, but they very in brightness somewhat, although the dimmest ones are still brighter. Color cast varies from a nice neutral white (5,000K perhaps) to a bluish cast. I have not run across any purple or blue-green cast yet. 55kmcd? I think that is a bit of a reach.

280kmcd 10mm white, 100ma. I count 4 bond wires on these. Lead frame is heavier to dissipate heat. They very in brightness and color like the 5mm ones above, but they are very bright. Beam is 15-20 Deg and free of artifacts, but the bluer ones have more of the bluish light near the center, although not as bad as the older 5mm used to be. Cost wise, it is cheaper to use one of these rather than several 5mm LEDs.

I plan to test lumen depreciation in another "fade" test.
Thanks for looking and HTH.
 

lumengeek

Newly Enlightened
Joined
Oct 18, 2008
Messages
17
Interesting to see how you have recorded both phosphor degradation (with tint shift) and epoxy degradation in these tests. Very interesting to see some real results here, goes to show the importance of not overdriving these small devices! Thanks for sharing.
 

Oznog

Enlightened
Joined
Dec 2, 2006
Messages
595
How'd you mount them? With these the leads dissipate the heat, but the full length of the lead would never actually be left on in most applications. Then again the PCB trace can dissipate more than what the mission portion of the lead did.

If the clip leads are high up on the lead then I'd expect the LED to dissipate better and perform better. I suppose if they were just stick into protoboard that should be fair.
 

JohnR66

Flashlight Enthusiast
Joined
Aug 1, 2007
Messages
1,052
Location
SW Ohio
How'd you mount them? With these the leads dissipate the heat, but the full length of the lead would never actually be left on in most applications. Then again the PCB trace can dissipate more than what the mission portion of the lead did.

If the clip leads are high up on the lead then I'd expect the LED to dissipate better and perform better. I suppose if they were just stick into protoboard that should be fair.

I clipped off 2/3rds of the lead before sticking in the breadboard.
 
Top