Cree's Gen-2 facts - More efficient but shorter life

[clemence]

I just read the Cree's published document about the lumen maintenance and found that ALMOST all their newer generation LED performs better efficacy with shorter life span.
For example: the popular XP-G2 average to have 20% better efficiency in brightness and power consumption (according to Cree's Product Characterization Tool) but also have 33,75% lower expected L90 lifetime compared to XP-G
You might want to check it for yourself:
http://cree.com/~/media/Files/Cree/...Application Notes/XLamp_lumen_maintenance.pdf

This maybe the answer why they still sell the older generation emitters for those user demand very long life maintenance free light engines.

 

[Yoda4561]

It's been a while since I looked that stuff up, but didn't Cree change their testing to use higher current/temperatures a few years ago to represent more real world conditions?

 

[petrochemicals]

Yep cree started using the industry standard 85•C rating, so I suppose there lifespan will shorten. On a similar note if they could make leds with a 10,000 hour lifespan at 25•C but more efficient I would be happy anyway, the cost of an led (2 dollars) versus its lifespan is extrodinary. Most will be changed long before they reach there 100,000+ hours, at a cost of 500hours a cent.

 

[sidecross]

Flashlight's along with LED's and battery technology seems to be out pacing the length of Cree's LED life expectancy. I have a Nitecore TM11 I bought when it first came out and now this light while still in use, is not very efficient.

For example the specifications for TM11 are 2000 lumens 1 hour 15 minutes using 4 18650 batteries while the new Fenix TK35UE (2015) has the same output and length of time using 2 18650 batteries and a Cree XHP 50 LED.

From my reading of the new Cree emitters they have an expected life of 50,000 hours. If innovation remains at this pace 50,000 hours of service might be all that would be useful before a replacement light with better technology exists.

 

[SemiMan]

-----

 

[clemence]

Flashlight is just a tiny fraction of Cree's LED use. Road, industrial, residential, public space light makers account for most of their customers. Yes, even 10.000 lifetime hours probably enough for flashlight since the new emitters always announced faster before the old one even finished the lumen maintenace test. But how about for example: public road lights? It's not something you can just replace yearly without troubles, let alone the time the replacement proposal has to spend in government office desks

 

[petrochemicals]

Cree always used to give the ratings at 25•c, lumens spectrum etc. You have a point about that, changing bulbs in traffic lights street lights etc is a factor. Its hightly unlikely although that the leds used in flashlights will be the same as used in streetlights. Either the cost, light output, they will probably go for something more tried and tested an specifically designed for it. They could always design different types.

 

[SemiMan]

-----

 

[petrochemicals]

Wrong on all counts.

The fact they are binned at 85C has 0 impact on LM80 test results.

Exactly same LEDs in flashlights and streetlights at least if taking Cree.

Semiman

Sort of notmy point and sort of my point.
By led I mean xpg2 as different to something a little less temperamental with lower lumens mm2. I always takes technology a little while to filter down and iron out the creases. They will be robust installments sereouslyover engineered. If your a company like cree you do not want to be putting out technology into the infra structure that isn't 1000% (thousand)proven, as things like airbags or car control units can be abother when they go wrong. It would be unlikely that they would use original xpgs in a large fixture. . Emmitter price is a factor in there calculations, and even thoughSodium lamps arnt cheap the price still factor. What are they using then these days in the industrial lighting xml originals? Or are they using something else? My initial thought they would use an emmitter without a lense on, and probably with alower starting effincy but a lower degridation curve.

As for the temerature cree definatley used to quote from 25•c, but I am not sure of the effect that would have on theLM80, dou you know the technicalities of it

 

[Woods Walker]

I think this issue is more an internet hypothetical concern than real world gear issue. For example I have a Sawyer mini water filter that has a life expectancy for the filter component of only 100,000 gallons. But I know the other plastic components won't last that long. I also will never actually filter 100,000 gallons through it. I haven't had too many flashlight failures but the few that did weren't because the LED was past it's life expectancy. No other things killed them a long long long time before that future date with entropy. Still more were made obsolete by advances in technology and changing requirements not that dark has gotten any more dark.

 

[SemiMan]

-----

 

[SemiMan]

-----

 

[petrochemicals]

Lense being the silicone dome, the reason being many auxiliary lenses are made specifilally for a certain type of domed led, where as auxilliary lenses made for undomed emmitters are going to be a lot more flexible I regards to the emmitters fitted. I have seen streetlights with 150+ throughhole emitters on test, not very bright back street lighting, it didnt stop the council changing the entire fixture on 30ft mainroad lampost lights to newer sodiums at a later date.

 

[SemiMan]

-----

 

[clemence]

Price, build cost, minimum lumen requirements and maintenance, and reliability DO matters! Especially for ultra long life installations. In my country it's common to find those orange mercury vapor lamps built during 70's-80's, they're already very dim but still used until today (dunno where do our taxes go....). Below is the screen captures for those don't have time to download the whole file. Enjoy,





L90(6K) means the samples have been tested/ran for 6000 hours, the rest of the data extrapolated
L80(14K) means the samples have been tested/ran for 14000 hours, the rest of the data extrapolated
As you can see, the XPG is tried and true, it's data samples aren't just extrapolated. 14000 hours of testing