Seoul P4 initial Evaluation- Production LEDs

[Curious_character]

No real benefit to C_C's method except being able to reuse the cheap copper spacer unless you weren't using one.

Mike

The benefit I saw over EngrPaul's method of epoxy on both sides of the copper slug was that I could position the emitter after the epoxy cured. But the alternate method I just proposed in another posting (expoxy the copper slug down and use compound between the emitter and slug) has the same advantage. You could also put epoxy on either side of the copper slug and sand it down flat as you suggest, then use compound on both sides and retain this advantage.

I doubt there's really any major benefit to one method over another, but I do like the ability to position the emitter after the epoxy has cured.

c_c

[EngrPaul]

Another factor is the difference in expansion rates of the emitter and the copper slug. The emitter will heat up much quicker than the copper slug. This puts an in-plane stress on the adhesive joint. After a while, I suppose this could lead to failure of the joint to some degree (no pun intended). By having compound there instead, there will be no joint to degrade.

[Kryosphinx]

Has anyone tried a P4 with an IMS20?

[Gryloc]

Well I didn't think about external forces on the emitter and the epoxy bond. I guess that one of the biggest downfalls of the multiple emitter Maglite mods is the use of separate reflectors. You have to tighten the top lens holder down, which puts pressure on the reflectors and the emitters. It seems as though the IMS reflectors that I used were not meant for bare emitters (like it claims) because the legs on each reflector is not long enough. Therefore, the force is directed onto the emitter and not bypassed to the heatsink by the legs.

Oh well. This is okay for me. I am still happy with the mod. I may replace the emitters in a few months anyway with the SSC P4, so I wont have to worry about having to break the emitters off now. Maybe I can create a spacer rings that will fill that little gap. I will stick with the AA Alumina epoxy (its all I got). I will figure out a way to ensure that the heatsink and the slugs of the LED is isolated.

Anyone think of using the engine paint that features small bits of ceramic? I seen it at my local advance auto shop. Its commonly used on manifolds and headers. If you get a thin enough layer of this paint, it may insulate the LED while still transferring heat. Just an idea. I am sure it has been tried before. Well, good luck all with your projects...


-Tony

[Mike abcd]

The benefit I saw over EngrPaul's method of epoxy on both sides of the copper slug was that I could position the emitter after the epoxy cured. But the alternate method I just proposed in another posting (expoxy the copper slug down and use compound between the emitter and slug) has the same advantage. You could also put epoxy on either side of the copper slug and sand it down flat as you suggest, then use compound on both sides and retain this advantage.

I doubt there's really any major benefit to one method over another, but I do like the ability to position the emitter after the epoxy has cured.

c_c

Sorry, I just put that badly. I definitely see the benefit in your method. I was questioning the limited benefit in my suggestion of only using the epoxy to create a thin insulating layer and not gluing "anything together" but using AS as you suggested between the parts used.

Mike

[Jon_L]

Anyone know of good electrically conductive cement?

I'd like to ground my emitters and reverse the batteries.

How about epoxy mixed with aluminum or zink powder?

[Gryloc]

I am sorry, but I think dealing with more layers will not be worth the trouble. I think you are worrying too much about a small and rare problem. Go ahead and epoxy the emitter to the original heatsink. There should be plenty of time (a few minutes) to position the LED during the cure time. I have had some good luck with electrical isolation of my LEDs with AA epoxy. My emitters broke loose (but stayed in place against the heatsink) because of the odd stresses on the emitter from the reflectors. The emitter by itself shouldn't have that much momentum in a drop to do harm, just make sure both surfaces are prepped and you have the exact ratio of adhesive and thermal parts. If you add too much of the thin part of the adhesive, the bond will we weaker.

This is just my thought.

-Tony

[WeLight]

I am a Cree distributor so let me state that to ensure my bias is established.

I find it interesting that both Cree and SSC are using the EZ1000 die platform and I find it equally interesting that Cree has only just announced the XLamp is now qualified for 1Amp usage. Clearly the thermal constraints related to qualification required some significant work to ensure the XLamp house was in order to support 1 Amp. Some of this thread content would suggest that SSC may not have been quite as rigourous in their development

Marketing driving Engineering again perhaps
Cheers
Wl

[NewBie]

Remember, just how thin of a layer 0.001" actually is. A lot of paper is in the order of 0.003" thick.

Besides the stresses of metals at different temperatures, also consider that the CTE of epoxy is quite different than copper and aluminum, and it, itself also expands and contracts. Just food for thought. As I said in the epoxy thread, proper surface prep and mixing is everything...

[NewBie]

Update.

I have a sample of just one Seoul P4 that I've ran for three weeks now, on that rather extreme heatsink shown earlier in the thread. Remember this is just a sample of *ONLY* one.

Anyhow, I've been planning on setting up a batch of these, as well as other LEDs from other manufacturers, for a long term brightness drop over time test. This ONE LED though has dropped in output in about 500 hours now, by about 10%.

I'd planned on waiting until I got batches all set up, before I mentioned this, but I was reading an article today, which toggled something in the brain housing group. An excerpt from the article:
"Chromaticity shift is not limited to warm white LEDs. Seoul's new P4 power LED package has a nominal CCT of 6500 K, and a typical output of 85 lm at 350 mA, equivalent to 76 lm/W. Elsewhere on the same P4 datasheet, there is a footnote to say that "the chromaticity coordinate of the LEDs can shift approximately x = 0.02, y = 0.03 in the direction of blue 1000 hours later." So during that time period the CCT could shift to approximately 8200 K. These figures suggest that Seoul's phosphor is very efficient, hence the excellent luminous flux and efficacy, but is not particularly stable over time."
http://ledsmagazine.com/articles/features/4/2/1/1

I'd remembered that on the datasheet, but it did not register as 1,000 hours when I read it. I looked it up, and they had it correct, it is at the bottom, in the orange box, as Caution number 2.

So, I hurried home from work, and took a look and realized the part has shifted from the very nice white it used to be, to a definite purplish looking white. This might also explain the drop in light output that I am seeing.

The amount of shift they are talking about is demonstrated as an example on this chart:



As far as flashlights, thats a lot of flashlight time, imho.

Has anyone been running these things and have 1000 hours on them yet, and could talk about their experience with them?


.

[tebore]

^^ Was this the LED on the super big copper heatsink? What was the drive level on this LED?

[Billson]

Newbie,

What current are you running your test setup at?

[easilyled]

Even after a drop in output of 10%, after 500 hours, this is probably
still similar to the output of a new Cree P4 bin, since the SSC-P4-U
is at least 10% more to start with, if not more.

This needs to be born in mind.

Likewise in 500 hours, to go from slightly warm to slightly cool doesn't
strike me as a disaster.

However, with this knowledge its probably judicicous to select WO or
warmer tints from the start.

[Brum]

Even after a drop in output of 10%, after 500 hours, this is probably
still similar to the output of a new Cree P4 bin, since the SSC-P4-U
is at least 10% more to start with, if not more.

This needs to be born in mind.

Likewise in 500 hours, to go from slightly warm to slightly cool doesn't
strike me as a disaster.

However, with this knowledge its probably judicicous to select WO or
warmer tints from the start.

But ARE the SSC's brighter compared to a Cree? It's nice that they state that in their specsheets, but they dont have any verification (unlike Cree's NIST testing). And the graphs we've seen from Newbie indicate that theres very little difference, and if you include the power draw (the XR-E has a lower Vf, so less power consumed at the same current) the difference is even smaller.
About the tint shift: it might not bother us as flashlight freaks, but where the LEDs are used continiously (ie fixed lighting) the tint shift is unacceptable, since after a month of use the nice warm white has dissappeared. And fixed lighting is the main market, us flashlight freaks just make a few percent (-points) of the total LED market.

[bullinchinashop]

So it MIGHT have a ring or two. BFD
What is the real point of these led's? Are they brighter? More efficient?
I started to get a Seoul light from DealExtreme but I can't see any real advantage to getting this vs another Q3.
Is everyone getting excited about these things because they're new or are they actually that much better than a LUX 3 or LUX 5?

[easilyled]

But ARE the SSC's brighter compared to a Cree? It's nice that they state that in their specsheets, but they dont have any verification (unlike Cree's NIST testing). And the graphs we've seen from Newbie indicate that theres very little difference, and if you include the power draw (the XR-E has a lower Vf, so less power consumed at the same current) the difference is even smaller.
About the tint shift: it might not bother us as flashlight freaks, but where the LEDs are used continiously (ie fixed lighting) the tint shift is unacceptable, since after a month of use the nice warm white has dissappeared. And fixed lighting is the main market, us flashlight freaks just make a few percent (-points) of the total LED market.

Yes - they definitely seem brighter to me.

Some of Newbie's graphs do back this up.

My Seoul-P4-U on a Nexgen750 is considerably brighter than my
Cree-XRE on GD825 (even though the cree is driven harder)

I was commenting on the tint shift from a flashlight point of view, but
even in fixed lighting, will it really make a big difference if it shifts from
white (on the slightly yellow side) to white (on the slightly blue side)?

To be honest I don't really care about its use in fixed lighting at the moment
though.

[easilyled]

So it MIGHT have a ring or two. BFD
What is the real point of these led's? Are they brighter? More efficient?
I started to get a Seoul light from DealExtreme but I can't see any real advantage to getting this vs another Q3.
Is everyone getting excited about these things because they're new or are they actually that much better than a LUX 3 or LUX 5?

The Seoul and Cree-XRE's are just about twice as bright as lux3's
for the same power output. They're still brighter than lux 5's at about
half the power output.

That is a BFD to flashlight enthusiasts seeing the biggest leap in
led efficiency in led's short history.

[bullinchinashop]

The Seoul and Cree-XRE's are just about twice as bright as lux3's
for the same power output. They're still brighter than lux 5's at about
half the power output.

That is a BFD to flashlight enthusiasts seeing the biggest leap in
led efficiency in led's short history.

Thanks. I haven't been around as much as I used to be & I'm a bit behind on my information on Seouls. I think I just might have to get that light fromDealExtreme.

[BPH]

I think NewBie’s previous testing does show the Seoul has significantly more lumen output (lumen/watt).


His excellent and thorough tests appear to measure “intensity”, which I assume is on axis, with units like Lux; lumens per square meter (If this is not the case, please disregard the remainder of this email).


The Cree has a much narrower beam pattern than the Seoul, so even if it measured the same intensity on axis, the Cree would be putting out less lumens due to the narrow beam pattern. I did some very quick calculations, which conclude if the on axis beam intensity is measure to be equal for a Cree and Seoul; the Seoul will be putting out much more total lumens (based on radiation patterns provide in the relevant PDF documents).


Also of critical importance is optics, so your really need to test a prototypes of each for the intended application to determine the best solution.

[NewBie]

Okay, well, I tested a second CREE, which is #2 on the overkill heatsink.

CREE #1 is already mounted on a 2" by 3" by 0.165" piece of copper sheetmetal only.

The device used for "brightness" testing was an Extech 407026 mounted to a white pail, such that the light from the LED would not have a direct path to the sensor. LED was inserted in a hole in the jug.

Meter here:
http://www.extech.com/instrument/pro...50/407026.html


Caveats:
-Like all light meters, it is not a spectroradiometer, and will have some error depending on the spectral output of the LED. Expect any errors due to spectral differences between LEDs to cast doubt on accuracy of measurements.
-CREE #2 was not mounted on the overkill heatsink, and it's thermal solution is different than the rest of the LEDs under measurement.
-CREE LEDs were both directly soldered to copper, NOT a MCPCB.
-The CREE LEDs were both from the "low output" P3 bin (73.9 - 80.6lm), NOT the Q3 bin (93.9-100.4 lm) which would be in the Seoul P4 U bin range (91.0-118.5 lm).
-CREE #1's bin appears to be higher than what the vendor that supplied said it was. Possible mixup at the vendor.

For the testing to be fair, I'd need to obtain parts from the same lumen bins from Seoul and CREE. This comparision here is a bit biased, as I do not have any Q3 bins from CREE yet. Some folks have been running around saying the Seoul P4 is brighter than the CREE XR-E, which is dicey at best to say, since we are not comparing the same lumen bins. Of course a part from a higher lumen bin will be brighter!!! It may be that companies are paying premium for the top end CREE bins, like Surefire, and thus the lack of availability- *conjecture on my part*.


Anyhow, the new graphs:




The BIG version:
http://www.molalla.net/~leeper/sp4int3.png


.

Seoul has nice projected reliability and life time graphs, one really nice thing about it, is later in the presentation, they show the effects of what a junction temperature of 90C does to the lifetime of their parts:
http://www.essc.co.kr/_HOMEPAGE/hom...Reliability.pdf

Has anyone seen any published information on the P4 like this? Since I'm seening a relatively dramatic drop in light output (which could in part be caused by the significant spectral shift in white tint) I'd be really interested in seeing similiar tests on the P4. Also, any information which includes the tint shift over time on a graph would be handy. Anyone know specificially, which phosphor this new phosphor is that they are using in the Seoul P4, which shifts around in color over such a relatively short time, actually is?

^^ Was this the LED on the super big copper heatsink? What was the drive level on this LED?

See below:

I think NewBie’s previous testing does show the Seoul has significantly more lumen output (lumen/watt).


His excellent and thorough tests appear to measure “intensity”, which I assume is on axis, with units like Lux; lumens per square meter (If this is not the case, please disregard the remainder of this email).


The Cree has a much narrower beam pattern than the Seoul, so even if it measured the same intensity on axis, the Cree would be putting out less lumens due to the narrow beam pattern. I did some very quick calculations, which conclude if the on axis beam intensity is measure to be equal for a Cree and Seoul; the Seoul will be putting out much more total lumens (based on radiation patterns provide in the relevant PDF documents).


Also of critical importance is optics, so your really need to test a prototypes of each for the intended application to determine the best solution.

One *really* *really* needs to go back and read a thread from the beginning, instead of jumping into the middle of things with assumptions, it will make life easier for both of us. See the first thread post I quoted.

No, intensity does not infer that I used a light meter at a given distance. Nor did I use a true integrating sphere. I used a white plastic paint bucket, to help gather the light. I used intensity, since it is not really open lux into space, nor is it a absolute total lumens result from an integrating sphere. I've been doing some testing with different coatings inside the bucket and other shapes, like boxes, buckets, spheres, and it is quite interesting.
-Shinny white paint/plastic gloss coatings can give certain beam distributions much higher readings than others.
-Flat white paints like you typically get for bright white house paints ~85% reflectivity, have almost the opposite effect.
-The amount of light a paint absorbs also has an effect.
-Sensor placement in non-spherical integrating solutions has a decent effect depending on light distribution.
-For more accurate measurements, there should be no direct light path from the source to the sensor. Professional integrating spheres have a baffle that blocks any direct light path.
-Different paints have different reflectivity curves, especially consumer/commercial grade paints which have cheap filler materials, and the reflectivity can aid/hurt different color bins differently.
-A single coating of paint is not sufficient. It takes multiple layers. I did a test with a clear glass and painted it. When you blast it with the light from one of these higher output flashlights, it is pretty amazing, how thick the paint needs to be, before you stop seeing the light pass thru.
-You really want a sphere...

I've got some paint that cost 250 dollars for a jar, that I'd like to use for a home-built integrating sphere, but I haven't found a decent 24" sphere to start it all off with. I have had it for a long time, the search continues.

Of particular note, light meters only approximate the human eye. Earlier I mentioned that this is one of the problems or caveats with this testing presented here. And different light meters respond differently, to different white tint bins, and will favor some bins over others, depending on their actual sensor response. So, when comparing LEDs for absolutes, not only do you need to have two LEDs that are close in tint to the human eye, they also need to have an actual spectral output that matches, otherwise you get further errors out of a light meter. For this reason, actual integrating spheres use a sensor that is responsive to the spectrum, where you can see relative output at each wavelength. Anything that uses a simple light meter is just a ballpark- not absolute.

Also, in comparing two LEDs for absolutes, you need to have them mounted similarly. You can't use thermal paste on one to the heatsink, then take one that is mounted on cheap China MCPCB, and then thermal paste that to the same heatsink. You need to offer the same thermal scenario for each.


Anyhow, my tests didn't use a simple lux meter, nor a perfect integrating sphere, more of a hybrid. Depending on how I collect the light, I can make one look better than the other (Seoul or CREE). Without a true integrating sphere and a sensor that can indicate output at each wavelength, trying to make absolute comparisions is quite futile.

With the Seoul P4 shifting around in color and lm/W so quickly in time (as indicated on it's datasheet and now what I see), due to this new type of specific phosphor they are using, we add a whole new dimension of things to worry about. For typical folks, who don't go thru a couple of batteries every week, this is less of a concern. For power users, or for fixed lighting/automotive headlights and such, this is something one really needs to look into very carefully.

In the past, McGizmo has shown how different tints have a surprisingly important effect on color rendering outdoors. Others have done similar tests. The difference between a XO and a X1 bin, is quite surprising, even though they are just one bin apart. In the case of the Seoul P4 over time, we are changing not just left and right (ratio of red to blue), but also moving a large distance vertically (ratio of red+blue to green). At one point, the preference that a number of folks had for outdoor flashlights was the green tinted X1, since it over-emphasized greens. For others, they liked XO or even a cooler white. Yet others preferred a warm VO. To each his own. The human eye changes as we age, due to the lens actually turning brown, and yes, the human eye lens actually looks brown when removed. There is a thread I posted this information on cpf. Something that looks like a icy cold white to a kid at 16, can look warm white to someone that is past retirement age. So, each can argue over what "perfect white" appears to them, and they'd both be right. Such is the nature of the human condition...

One could even argue for various tints, depending on what they wanted to see more. If you want to see green vegetation, something with more green would be better. If one wanted to seen browns more, than something warmer would be better. If one wanted something white to stand out in the forest, bluish tints would be better in many forests. It is interesting what you can do with light, when you have control over the light spectrum, and how you can make certain things stand out from the rest. You can also make things stand out by making them black against the background. Essentially, the name of the game is to increase the contrast between what you want to see and what you don't. The human eye is really good at picking up contrasts.

Anyhow, the LED under test has now dropped from a reading of 310 on that chart to 270 now. The current is measured the same in both, to within 0.1% on the meter reading. The tint has visibly shifted very considerably, and is easily apparent- no need to compare against anything, you can look at it and tell it is purple white now (not cold white or even blue white, purple white). It was a rather warm white or even yellowish white when compared to normal while LEDs. I suspect that with more of the light output shifting towards blue and red (together they make purple), that since the measurement sensor and the human eye are more sensitive to green, this is the real reason for the actual light output loss- spectral shift as the phosphor degrades.

[NewBie]

Newbie,

What current are you running your test setup at?

1000mA on that big monster heatsink above.

[BPH]

NewBie,
I have done my best to follow the thread, but I have missed a few comments, and didn't see your note on your hybrid integrating sphere.

My main point/question was the Other people are interpreting your data (again, thanks for providing it) to suggest that the Cree is producing nearly the same output as the Seoul, and I was commenting on the validity of this based on the different beam patterns (and test set-up).

So, my question is; from your data, can you make any reasonable guess as to the relative total lumen output between the Cree and Seoul chips you tested?

Thanks,
BPH

[chimo]

Seoul definitely has a different phosphor mix.

Here is a comparison with some Cree and Luxeons. The top photo is just taken in daylight. The bottom one was taken at the same time but has been illuminated by a 1W CreeUV light. I had to move the UV source a fair ways back so the Seoul phosphor would not saturate the image.


Paul

[Kinnza]

Anyhow, the LED under test has now dropped from a reading of 310 on that chart to 270 now. The current is measured the same in both, to within 0.1% on the meter reading. The tint has visibly shifted very considerably, and is easily apparent- no need to compare against anything, you can look at it and tell it is purple white now (not cold white or even blue white, purple white)

If the tint shift is so noticiable, likely the output drop in lux reading is mostly due to it, meaning that about 1watt of radiometric power is emited, but with a luminous efficacy of 310 lm/w with the original tint, and 270 lm/w when it shifted.

If the shift is both to the blue and the red, the luminous efficacy may be easily higher than those 40lm/w. In contrast, the CRI should improve as consecuence. This trade off is very usual, but not in the same product. Glad you aware us, Newbie, really is something that SSC had to note. IMO, isnt a bad trade off, but the fact is P4's photometric efficiency is lower than announced when it reach a given (short) time use.

So there is a photometric output drop together with CRI improvement, but probably the radiometric output is somewhat stable. Impossible to say without a spectrometer, but if the tint shift is so noticiable, i believe that this analysis is reasonable. Of course, its impossible to say how much of the output drop is due to tint shift and how much of radiometric output degradation.

Personally, i like that kynd of tint (purplish), so im interested in this phospor for home use. Definitively, id like to know more about it. For me the main problem with the P4 is the gummy dome. IMHO, it make it just a good choice for enclosed lights.

So, my question is; from your data, can you make any reasonable guess as to the relative total lumen output between the Cree and Seoul chips you tested?

As far as i can see it, it was already done along the thread. Newbie has explained in the last post why he cant get more accuracy than he had already showed. With this limitation in mind, the graph in the last Newbie's post show the relative emission (refered to watts consumed instead of current drawed, wich is more accurate than anything).

I've got some paint that cost 250 dollars for a jar, that I'd like to use for a home-built integrating sphere, but I haven't found a decent 24" sphere to start it all off with. I have had it for a long time, the search continues.

We are in the same boat . If you find it, please tell me . For the coating, i suggest you a 50/50 mix (by weight) of latex flat white paint and BaSO4 (Barium sulphate) (adding a bit of water make the mix more manegeable, its very dense). It had a pure lambertian reflection and excelent reflectance between 430 and 800nm. In most of the visible range, its just about 2% less than spectralon. The white paint only degrades reflectivity at 430nm or lower, and current blue/white chips emits very little in that range. From 430 to 700nm, differential reflectivity is very small, ranging from 96 to 98%.

[BPH]

I'll try again (and then stop bothering you'll). I don't think I was able to make myself clear.

Can a reasonable conclusion be drawn from NewBie's testing as to if the Cree or Seoul has more Total lumen output? (that would be either yes or no)

I assume the answer is not really, too close to call. It could go either way depending on how NewBie's hybrid integrating sphere favors the particular radiation pattern of the emitter and other testing details.
-BPH

[Calina]

Seoul definitely has a different phosphor mix.

Here is a comparison with some Cree and Luxeons. The top photo is just taken in daylight. The bottom one was taken at the same time but has been illuminated by a 1W CreeUV light. I had to move the UV source a fair ways back so the Seoul phosphor would not saturate the image.

Paul

Wow, it glows ! Does this means that GITD buttons and O'rings are unnecessary with these lights?

[chimo]

Wow, it glows ! Does this means that GID buttons and O'rings are unnecessary with these lights?

It doesn't stay glowing. The phosphor is being excited by a UV light I was shining on them. It glows a lot brighter than the other phosphors.

[NewBie]

I'll try again (and then stop bothering you'll). I don't think I was able to make myself clear.

Can a reasonable conclusion be drawn from NewBie's testing as to if the Cree or Seoul has more Total lumen output? (that would be either yes or no)

I assume the answer is not really, too close to call. It could go either way depending on how NewBie's hybrid integrating sphere favors the particular radiation pattern of the emitter and other testing details.
-BPH

With the parts I've managed to get my hands on, I'd tend to believe the Seoul P4 is a tad more efficient, once you look at Watts consumed, at least when it is new.

One of the falacies is that I've seen, folks pick a bottom binned P2 CREE, and then snag a nice U bin Seoul, and claim smackdown. Thats like picking an R binned Luxeon III and a U bin Seoul P4- of course, they are from different bins, quite obviously one will be a lot more efficient.

Some folks are attempting to measure accurate lumens with a typical light meter. Differences in color bins can cause considerable errors, and lead to erronous results. Even differences in spectral output between different phosphors can add error.

This would be considered a very high end photodiode, that has a much better color filter on it than one finds in many light meters:
http://www.vishay.com/docs/81519/81519.pdf

At the bottom of page 3, figure 5, you will see Vlambda eye-this is the human eye response. The black line is the actual photodiode, and it's response curve to light. Notice the large margin of error vs. wavelength for this part, when compared to the human eye. This where large errors can creep in.

I've seen folks compare an LED mounted on a MCPCB to one direct mounted on a copper plate. When you compare in this fashion, obviously the MCPCB will measure a tad lower, especially with the higher profit margin MCPCBs. To be fair, it is best to give them an equal thermal solution.

Power is what the LED is going to use in a flashlight, so you really want to look at Power, which is Volts * Amps. Current only can be rather misleading.

Besides shifting color in a relatively short time, and the loss of lumens, the Seoul P4 LEDs I've tested, also shift more in color with changes temperature than most white LEDs I've tested.

Some folks like to claim that this LED is worst, best, or whatever. Nothing wrong with the LED, it is just different, and different trade-offs were made in it's design as compared to other LEDs. Some of the trade-offs will be a positive thing, and some will be a negative thing. It all depends upon your goals, application, and such. It is nice to have a variety of choices for power LEDs these days.

[wojtek_pl]

1000mA on that big monster heatsink above.

I wonder if tint change would happen so fast with less amperage... Say 750 mA...
In general... this doesn't look good as reliability goes...

[NewBie]

The tint shift should be less with less current over time, but this is not something I've tested.

One of the large factors for tint shift over time is the die temperature, and it's affect on the phosphor. As I recall, the LED slug is held below 30C, the exact number should be mentioned earlier in the thread. This part has been held rather cool, due to the excessive heatsinking, and the thermal resistance to the LED die is very low in this part, so I doubt that is much of a contributor for this LED.

The datasheet cautions about the shift, but does not give an actual specific cause for it.