5mm LED fade testing continued

[JohnR66]

Thanks to AT&T discontinuing Web hosting, all the links to my charts will soon be broken.:thumbsdow With new hosting, I will continue the updates to the lumen depreciation tests here.

Nichia GS: After 3000 hrs at 30ma it has dropped to 76% of initial output. The drop started before 1500hrs and has been a steady, almost a linear decline. I expect it to drop below 70% at around 3500hrs.
3-18-10 update: 3384 hrs. The GS seems to have stabilized. Perhaps it is not done just yet. This will be interesting.
4-19-10 update: 4152 hrs. The GS dropped a little. it is now at %72 of initial intensity. It is in danger of dropping out of the test soon.
5-18-10 update: Nichia GS: Output has dropped to below 70%. It lasted 4,800 hrs at 30ma. At 15ma, I'd expect life to be at least 20,000 hr. Test completed.

Radio Shack 276-0017 neutral white 30 Deg LED. After 3000hrs at 30ma, there has been a small decline, however it is still at 102% of it's initial brightness. When a decline starts after thousands of hours of steady state output, means it has began its decent into the valley of the dim LED. It is too early to estimate when it will drop to below 70%.
3-18-10 update: 3384hrs. The RS LED also seems to have stabilized (slight brightness increase).
4-19-10 update: 4152hrs. Not much change. Steady as she goes!
5-18-10 update: 4848hrs. This LED shows no quit. Steady state.
6-18-10 update: 5592hrs. Slight drop but still above initial intensity.
8-19-10 : 7080hrs. No change from last measurement.
10-10-10: 8328hrs. No change. LED has been in overdrive for almost one year now. At half current, life is at least 4x, so it would last at least 4 years at 15ma. Since the LED has spent the last 4 months at steady state, I'd venture to guess that it will last 50K hours at normal drive currents.
10-26-11: 17472hrs. LED remains above initial brightness after 2 years. At this point I consider the test over.

Cree 503C-WAS-CBADA151 This part no. specifies a wide brightness and tint, but they all have been about the same brightness and tint (sample of 110 pcs). After 2000hrs, it has been a roller coaster ride of luminance, but at least it has not started to decline. The blue line is an older Cree LED. The newer 503 series is brighter and longer lasting.
3-18-10 update: 2328 hrs. The Cree just wants to get brighter. Interesting.
4-19-10 update: 3096 hrs. Notable drop, but it is back at initial output. Not sure if it is fading or just a momentary dimming. Odd how the LEDs vary over their life. I have seen this oddity in other tests not performed by me.
5-18-10 update: 3792 hrs. Trend seems to show it is now on the decline but still at 92% of initial intensity.
6-18-10 update: 4536 hrs. Falling fast and now at 70.5% of initial output. It is safe to say this LED will drop below 70% in the next couple of days. Interesting fade characteristic as compared to the GS. The GS started loosing intensity just before 1,500 hrs and slowly declined. This Cree gained intensity which peaked around 2500 hours and then started fading at a higher rate.
6-22-10 update: 4632 hrs. As expected, the LED has fallen into the 69% range. The reason for fading seems to be the discoloring (slight darkening) of the epoxy and/or the phosphors above the die. I doubt the die itself lost output. At 15ma, I'd expect the LED to last at least 20K hrs. Like the GS, this is excellent life for a 5mm LED. Test completed.

jtr1962 was kind enough to send me some surface mount LEDs for testing. These are tiny, but with masking tape, they are fairly easy to solder onto a perf board (the kind with the ring of copper around the hole).

The Osram LW E6SG PLCC4 (on Left in above photo) has been cooking at 39ma for 1680hrs and has been pretty much steady state. These have a nice near neutral white tint.
3-18-10 update: 2088 hrs. Very slow but steady decline
4-19-10 update: 2855 hrs. Now steady state. For 39ma, this LED has held up well.
5-18-10 update: 3551 hrs. Remains steady state. Amazing LED to handle 39ma so well.
6-18-10 update: 4295 hrs. Seems to be starting to descend.
8-19-10: 5783hrs slight gain in output. So, after falling a bit, it has stabilized.
10-10-10: 7031hrs. Slight drop. Still at 86% initial intensity.
10-26-11: 16175hrs. LED fell to 80% of initial brightness. At 39ma for nearly 2 years and well above 70% of initial brightness, the LED is made for the long haul. At this point, I consider the test over.

The smaller Nichia (not charted yet) is just a bit less bright and cool white. It is testing at 30ma and after nearly 600hrs has not changed
3-18-10 update: 984 hrs. Now added on chart. starting to decline.
4-19-10 update: 1752 hrs. After starting to drop, it has regained steady state output.
5-18-10 update: 2448 hrs. Starting to decline. Now at 80% intensity.
6-18-10 update: 3192 hrs. Dropped below 70% of initial intensity. Not bad for such a tiny LED. Test completed.

This is a reminder of how pathetic those cheap ebay/Chinese LEDs are. This one fell below 70% in about 3 whole days at 30ma

I decided to make it easier on the cheapo ebay LEDs. I tested them at 12.5 ma. They lasted about 500 hrs. Hey! that is 3 weeks of use!

I point out this junk because they end up in many products from cheap flashlights, to "shower head" PAR styles lights, landscape lighting and under cabinet lights. A new Menards store opened nearby and some LED lights on display were faded in 1 month!

Coming up: I have some Fox Group 360nm UV LEDs on the way and I started testing some Cree low dome 5mm 110 Deg LEDs.

 

[Erasmus]

Thanks for posting your observations

 

[Evilsizer]

good stuff, done testing on any other LEDS?

 

[JohnR66]

good stuff, done testing on any other LEDS?

If you check this LED subforum over the last year and a half, I have tested many LEDs, from 3mm to 4 pin flux style. Most from ebay. Compared to the LEDs above lasting several thousands of hours, they all are junk. Only the flux style LEDs gave a few hundred hours of useful life.

The 1/2 watt multi chip LEDs from ebay were the worst junk I have ever seen. Tested at only 70% of max current, one type nearly went dark after only one week.

To put it another way, nothing really interesting to see in those old tests.

I am self employed and business sucks now, so I'm not making any money. I rely on people here to send samples to test. When I do get some extra money, I buy LEDs, like the Crees and the Fox Group UV LEDs.

 

[Phaserburn]

I was in Wally World last night, and saw the Coleman 2AA 5mm leds light. What I hadn't noticed before was that they are using 5 5mm Cree leds (stated on the packaging). I would be interested in getting one of these (around 15 bucks) if I could tell if it was using the old or new Crees. Any way of knowing? I'm hoping for a light reminiscent of the Inova X5 but using 2AAs.

 

[JohnR66]

I was in Wally World last night, and saw the Coleman 2AA 5mm leds light. What I hadn't noticed before was that they are using 5 5mm Cree leds (stated on the packaging). I would be interested in getting one of these (around 15 bucks) if I could tell if it was using the old or new Crees. Any way of knowing? I'm hoping for a light reminiscent of the Inova X5 but using 2AAs.

You can tell the type of LED if you can look directly down into it while it is on. Underdrive it so the LED is very dim and safe to look at. All the 15 Deg 503C LEDs I have (110 pcs) have a rectangular blue die that shows through the phosphor. I can say this is true only for the bin I have. Here is what the die looks like. I photographed it through a blue filter to help it stand out. On the OLDER LEDs, the phosphor appears to be too thick to see through.

 

[Phaserburn]

Thanks, I may have to try that out. I wonder what Coleman is driving those 5mm leds at.

 

[Dave_H]

Good of you to do this service JohnR, makes sense that people
who want specific LEDs tested drop a few in an envelope to you,
assuming you can take them all Would be neat to see a shot of
your test area with all other lights turned off, if indeed you need
any other lighting in that area.

I don't have immediate application of your results but it sure is
an eye-opener. Anyone doing a retrofit like jtr's fluorescent night-light
would be better not using self-fading LEDs (sort of like self-splitting
pants...).

Any comment on Ledshoppe LEDs...are they in the same category
as eBay?

Incidentally, these are fade tests but do you see any other early
failure modes besides outright dying? Some cheap 5mm devices go
into a rapid flicker mode, before they die, although some temporarily
recover; thermally related, could it be intermittent bond wire
heating/cooling?

Dave

 

[JohnR66]

Thanks Dave! I have not tested Ledshoppe LEDs. If they are cheap, beware.

All the LEDs I have tested just faded. I didn't see any flicker or go out. I did notice that cheap LEDs tend to be more sensitive to static electricity (ESD). I tested some by zapping them on purpose. Some would flicker or not light after the ESD damage.

 

[Evilsizer]

Thanks Dave! I have not tested Ledshoppe LEDs. If they are cheap, beware.

All the LEDs I have tested just faded. I didn't see any flicker or go out. I did notice that cheap LEDs tend to be more sensitive to static electricity (ESD). I tested some by zapping them on purpose. Some would flicker or not light after the ESD damage.

how cheap would be to cheap? i bought a few led's from electricgoldmine site and some where 1.49 and one i got was .99c, the .99c one was a 5mm led, the others cost up where 8mm+.

 

[JohnR66]

how cheap would be to cheap? i bought a few led's from electricgoldmine site and some where 1.49 and one i got was .99c, the .99c one was a 5mm led, the others cost up where 8mm+.

$10 for 100 or ~$18 for 50 is what I paid on ebay for 5mm white LEDs. The 50 I paid $18 for were a bit better, but still faded much too quickly.

 

[JohnR66]

Testing Two Cree C535A-WJN-U0-Wf-29 5mm low dome 110 Deg white LEDs. Continuous drive is 30ma and 39ma. Light meter tested at 20ma (typical for all my tests).

These are neat little LEDs. The short epoxy dome makes them fit into tighter spaces and the wide angle beam makes them great for low power wide angle use. Max rated continuous drive current is 25ma as stated by Cree's datasheet. Since I got so many of them, I tested one at an abusive 39ma. It is fading quite a bit, but you should not take this as a lifetime result. Follow the 30ma curve which shown pretty much steady state.

These LEDs seem to be twice as bright as the Nichia and Osram surface mount LED tested above. This could be due to the true beam angle of these LEDs.

If you were to force me to make a negative comment, the tint bin of "Wf" specifies a really cool white light. At least it is free of blue, violet or green tints of some cheap LEDs.

3-18-10 update: 384 hrs. LED at 30ma is steady state (slight increase), while the LED driven at 39ma has fallen to 1/2 light output and has dropped out of this test. This shows how overdriving a LED can shorten its life dramatically. The overdriven LED started to fade after only a few days. Actually both LEDs are overdriven since the spec sheet give 25ma as max continuous current. I got some of these in a little less cool white Wm bin. Quite nice!
4-19-10 update: 1152 hrs. Slight increase in output. I was concerned of how long the LED was going to last with the other fading at 39ma. At 30ma, which is 5ma over the rated max, these are quality LEDs.
5-18-10 update: 1848 hrs. Very slight decline but still above initial output.
6-18-10 update: 2592 hrs. Gentle decline continues. At 95% of initial output.
8-19-10: 4080hrs. It has fallen to 59.6% of the initial intensity. I estimate that it dropped below 70% at 3,500hrs. At half current, the LED will last at least 14,000hrs. Since the same LED would last only 1/10th as long at 39ma as it did at 30ma, I'd say it probably could exceed 20K hrs at 15ma. Definitely a good LED where you need a low power wide angle light. Test completed.

I made a little board using 8 LEDs. Two in series for 4 circuits paralleled and limited by one resistor. Vf seemed close enough and I don't want to waste resistors. 9 volt Vs. Makes a nice little nightlight, if not a bit bright.

I'm not sure if the CPF member wants to be named, but he said that he appreciated my tests and sent me thousands of these LEDs for free. I couldn't believe it! I had to check to be sure he hadn't made a shipping error. I told him I really appreciate what he did for me. I don't have much money, but when I get some extra, I go blow it on LEDs. I guess I'm a LEDaholic of sorts, so this was a great gift from someone I didn't even know.

I have sent many of these LEDs to another member who is going handle giving them away + the cost of shipping to interested CPF members. He will make it known when the LEDs will be available when he is ready. I would become quite overwhelmed if I tried to handle this myself.

 

[JohnR66]

Nichia GS vs. Cree

Has the Nichia GS met its match?

A CPF member (who wishes to remain anonymous) sent me some Cree 5mm white LEDs: C503C-WAN-Cb-Wp-29 which are bighter and are warmer in tint than the Crees I was raving about a few months ago. Here is my review of them vs. the Nichia GS C0w bin.

Relative intensity at .5 meter:
GS: 23.2
Cree: 22.1
The bullet shaped lens of the GS focusses a bit thighter for an edge intensity. This is so close that you would not be able to tell the difference.

Since I don't have an integrating sphere, I filed down the dome of the LEDS so it was flat just above the bond wires. I polished them to be clear again. This normalizes the output so I can compare the light output more accurately. I then tested the LEDs at different currents to see how they compare.

At 20ma they are essentially the same brightness. The GS starts to open a gap at 30 ma and increases from there. So at higher currents, the GS outperforms the Cree.

Here are the beam shots. Cree on left. Same current on both.

Nichia GS advantage:

• tight focus with some spill
• high brightness
• excellent lumen increase at overdrive currents
• long life (has not dropped out of test after 3,000hr at 30ma
• overall close to neutral white tint of C0w bin
• silver plated copper lead frame for best heat dissipation at overdrive currents

Nichia GS disadvantage:

• currently not available to some markets (US) by normal means
• weird oblong "angry blue" center with yellowish halo
• Rather expensive

Cree C503C advantage: (of bin listed above)

• brightness on par with the GS
• much better beam than GS (slight yellow halo, but no "angry blue" center)
• low price for a quality LED and good availability
• long life (perhaps will outlive the GS)

Cree C503 disadvantage:

• brightness does not increase as much as it does with the GS with overdrive currents
• steel lead frame not as good at dissipating heat under overdrive conditions

Conclusion
There is no clear winner here. The LEDs are both high brightness (for 5mm) and have a long life and are not very cool in tint.:thumbsup: The GS has the advantage of being brighter at overdrive currents and that is fine for flashlight use. Unfortunately, Nichia chooses to make them hard to get at reasonable prices in some markets. Some are put off by the blue beam center.

The Cree's pice/performance, availability and cleaner beam put it a notch above the GS for me. It may be different for you depending on your use. Cree's C503D LED should be even better, but I have not tested any yet.

 

[JohnR66]

All LEDs under test have been updated with latest chart data. I added any comments after "3-18-10 update" entry.

Some interesting things are brewing in these tests.

 

[JohnR66]

Nichia GS Vs Cree: Part II

I now have the brightest bin of the Cree C503C series white LEDs
C503C-WAN-Da-Wn-29 (Da bin is the brightest)
Vs.
GS C0w (brightest bin of GS)

At 20ma the Cree is a hair more intense. The GS overtakes as current increases. In the 20 to 40ma range, they will appear to be the same brightness visually although they have quite different beam profiles. Given the price, availability and better beam characteristics of the Cree, it is the clear winner here. If you are going to push the current above 40ma, the GS is still the LED to beat.

Axis: X=ma, Y=relative intensity

If you're not happy with the angry blue beam of your light with the GS LED, this is the upgrade LED. Tint bins of Wn or better will not be too cool white.

Cree C503D Db bin is even a brighter LED, but I have yet to get my hands on any for testing.

 

[jtr1962]

Re: Nichia GS Vs Cree: Part II

Cree C503D Db bin is even a brighter LED, but I have yet to get my hands on any for testing.

Db is out? The data sheet I have only goes to Da. If so, then that would be one amazing LED!

One of these days I'll lumens test the Nichia GS you sent me, as well as all the Crees. Just going by your results, it's entirely possible the Cree is 100+ lm/W at 20 mA. I remember the Nichia GS's I tested a while back nearly broke 100 lm/W, but I think the ones you sent are even brighter.

 

[JohnR66]

Re: Nichia GS Vs Cree: Part II

Db is out? The data sheet I have only goes to Da. If so, then that would be one amazing LED!

One of these days I'll lumens test the Nichia GS you sent me, as well as all the Crees. Just going by your results, it's entirely possible the Cree is 100+ lm/W at 20 mA. I remember the Nichia GS's I tested a while back nearly broke 100 lm/W, but I think the ones you sent are even brighter.

jtr, According to the 503D datasheet the Db bin is 39,500 to 46,100 mcd
http://www.cree.com/products/pdf/LEDlamps/C503D-WAN.pdf
It should outperform the GS at 30ma and under and probably tie it a 40ma. Given the Cree's better beam, flashlight manufacturers using the GS should take a hard look at these newer Crees (Wm bins or warmer).

I'll send you some of these Da bins I got from you know who. They are quite nice. As for the top bin of the 503D series, I don't know if they are in the wild yet. Not sure why Cree has both the 503C and 503D. Binning seems the same except that the D has the one higher brightness bin.

 

[jtr1962]

Re: Nichia GS Vs Cree: Part II

jtr, According to the 503D datasheet the Db bin is 39,500 to 46,100 mcd
http://www.cree.com/products/pdf/LEDlamps/C503D-WAN.pdf
It should outperform the GS at 30ma and under and probably tie it a 40ma.

I see now-it's 503D, not 503C as I thought. Also, the Db bin might well outperform the GS at all currents, not just lower ones. I noticed a trend looking at your charts where it appears the higher-binned Crees have less droop at higher currents. This isn't unexpected. A more efficient LED will make less heat. Droop is partially related to die temperature. A cooler die obviously means less droop.

I'll send you some of these Da bins I got from you know who. They are quite nice. As for the top bin of the 503D series, I don't know if they are in the wild yet. Not sure why Cree has both the 503C and 503D. Binning seems the same except that the D has the one higher brightness bin.

Thanks! As I said, when I get a free day I'll be lumens testing everything you sent me. I wouldn't be surprised if the Da bin finally breaks the 100 lm/W barrier for small LEDs. Power LEDs surpassed that a while back, but I"ve yet to test indicator-type LEDs which manage 100 lm/W or more at rated current.

 

[coors]

I just became interested in the 5mm LEDs, recently. Found this thread today and I must say thanks a whole lot, for all of your time in sharing this information. I suspect that you're also the guy on Youtube that has some 5mm comparison reviews. Just yesterday I purchased the RadioShack 2-pack of #276-0017 5mm LEDs, that were so highly touted in the Youtube review. I put one in a Chinese clone of the PT EOS headlamp, replacing the red 5mm LED, which was driven at 20ma:

This Radioshack LED's neutral tint is really nice, in my opinion. For reading a book, it's 7000mcd and 30degree angle of output is just "perfect". However, walking around outdoors last night, I found myself wanting a little more output/flux. I could see fine to navigate but there was a notable lack of detail in the topography.
I've searched for the C503D-WAN and found them listed on Cutter's site: http://www.cutter.com.au/proddetail.php?prod=cut926 . However, it looks to me as if you can end up with a tint ranging anywhere from 4600k-15000k. Is there no way to buy these Cree 5mm LEDs at a more specific temperture bin than this? I did order some of the Cree C503C-WANs, from Digi-Key, but again there was no way to order a specific tint. Is this why some folks order large amounts of the 5mm LEDs, so that they can sort through them to find the desirable tint?

 

[JohnR66]

I just became interested in the 5mm LEDs, recently. Found this thread today and I must say thanks a whole lot, for all of your time in sharing this information. I suspect that you're also the guy on Youtube that has some 5mm comparison reviews. Just yesterday I purchased the RadioShack 2-pack of #276-0017 5mm LEDs, that were so highly touted in the Youtube review. I put one in a Chinese clone of the PT EOS headlamp, replacing the red 5mm LED, which was driven at 20ma:

This Radioshack LED's neutral tint is really nice, in my opinion. For reading a book, it's 7000mcd and 30degree angle of output is just "perfect". However, walking around outdoors last night, I found myself wanting a little more output/flux. I could see fine to navigate but there was a notable lack of detail in the topography.
I've searched for the C503D-WAN and found them listed on Cutter's site: http://www.cutter.com.au/proddetail.php?prod=cut926 . However, it looks to me as if you can end up with a tint ranging anywhere from 4600k-15000k. Is there no way to buy these Cree 5mm LEDs at a more specific temperture bin than this? I did order some of the Cree C503C-WANs, from Digi-Key, but again there was no way to order a specific tint. Is this why some folks order large amounts of the 5mm LEDs, so that they can sort through them to find the desirable tint?

Sorry to be slow. I have not been around lately... When replacing the red with white, remember the forward voltage (Vf) of the white is much higher and can result in very low drive current depending on the driver of that light. The RS led has a lovely color temp, but it does not match Cree and Nichia GS led in overall output, plus it's wider beam means low intensity.

The highest bins will probably be cooler as warmer tints tend to be less bright. I looked at cutter's C503D and the best bins range from 28 to 46 cd so it is kind of a crap shoot of what you will end up with, although these are quite bright.

I have C503C-WAN-Cb-Wp-29 (28.2-32.9 cd) warmer bin of cool white.
Also, C503C-WAN-Da-Wn-29 (32.9-39.5 cd) cooler but brighter.
Either way, your within Cutter's brightness range offering.
Also, Nichia GS C0w bin brightest and warmest of cool white bins

I can send some to you some for free (+$2 for shipping). PM me.