How long do LED shop/ceiling lights really last at full output anyway?

How long do LED shop/ceiling lights really last at full output anyway?
In another thread, the topic came up that LED lights may not last as long nor as bright as claimed on the package:
Lights of America exaggerated LED light bulb performance http://archive.jsonline.com/blogs/news/246355581.html http://green.blogs.nytimes.com/2010/09/08/federal-agency-sues-led-bulb-maker/
Since the failure mechanism is electronic as much as it's the fact that LEDs diminish in light output over time, I wondered how long the LED lights (the entire unit, including electronics) really last, and, what "rules" were in place for the claims on the package (since I never get the life that the fluorescent or incandescent bulbs claim either).
As just one (admittedly egregious) example, the bulbs referenced above:
CLAIM: Replaces light output from a 40 Watt bulb (aka 400 lumens) TESTS: The F.T.C. found the bulbs produced only 74 lumens of light
CLAIM: The LED bulbs in question would last 30,000 hours TESTS: They lost 80 percent of their light output after only 1,000 hours
That didn't even cover sudden failure from the electronics.
So, I wonder aloud ... How long do LED shop/ceiling lights really last at full output anyway?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 31 Oct 2016 16:08:58 +0000 (UTC), Algeria Horan wrote:

Doing some due diligence, I find that there *are* rules in place. https://www.ftc.gov/tips-advice/business-center/guidance/ftc-lighting-facts-label-questions-answers-manufacturers
The Lighting Facts label contains: 1. a bulb’s brightness, 2. energy cost, 3. life, 4. light appearance, 5. and wattage
Specifically: 1. The brightness in average *initial* lumens rounded to the nearest 5 2. Annual *initial* energy cost at a low 3h/day at an unrealistic 11¢/kWH 3. Life in years, rounded to 1/10th based on a low usage of only 3h/day 4. Color in Kelvin ranging from 2,600K on the left to 6,600K on the right 5. Wattage in average *initial* wattage
Note that, for LEDs, the "initial" specs are almost certainly going to be vastly different than the actual specs over time, so, already, the label has to be taken with a grain of salt.
Also note the highly unrealistic 11¢/kWH (which is essentially unattainable, where I live).
But at least we know what the LEDs are supposed to deliver, initially.
So now we need to figure out what typically happens to these LED light bulbs over time, mostly in terms of light output & when the lifetime brightness cutoff point has been reached.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

http://www.digikey.com/en/articles/techzone/2012/feb/understanding-the-cause-of-fading-in-high-brightness-leds
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 31 Oct 2016 18:08:23 GMT, Scott Lurndal wrote:

Thanks for that url. Here's a relevant set of snippets from:
Understanding the Cause of Fading in High-Brightness LEDs By Steven Keeping Contributed By Electronic Products, 2012-02-21
LED failure” is most likely to be the result of light output falling below an acceptable threshold (typically 70 percent of the initial output.
The primary cause of that fading (or “lumen failure”) is triggered (for the most part) by the minute threading dislocations introduced to the chip during wafer manufacture.
Threading dislocations are a major problem...where threading dislocations are vertical micro-cracks caused by strain generated by the mismatch in InGaN and Sapphire or SiC crystal lattices .. and where ... things get worse over time, as the rate of degradation is directly related to the initial density of threading dislocations and the heat to which the LED is subjected ... all of which gets worse ... due to heating during operation, thermal expansion and shrinkage when the LED is turned on and off, and mechanical stress such as vibration.
Worse yet ... as the chip ages, it will run hotter and hotter ... due to an increased number of phonons, accelerating the formation of dislocations and the device’s eventual demise.
------------------- So now we know what kills LEDs to the 70% level (which is the formal definition of dead), which is that inherent cracks between crystals form over time, just as mud cracks as it dries at the bottom of a pond.
The fundamental problem is cracks between crystals only gets worse. b. Heat makes things worse even faster c. On/off cycles makes things worse even faster d. Vibration makes things worse even faster
So, given they don't heat/vibrate/cycle the LEDS (other than the 4 hours per day), you'll likely never get the advertised L70 lifetime in the real world.
But what do you get for a 70%-illumination lifetime in the real world?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 31 Oct 2016 19:08:08 +0000 (UTC), Algeria Horan wrote:

I forgot to mention that none of what kills the LED over time discusses the frail electronics, which also fail.
So, we need to know how long the "LED drivers" last too.
The life of the fixture is the shorter of the two failure modes: a. Inherent cracks between crystals get exponentially worse over time b. The drivers suddenly fail at any time
So, we still don't have any decent grasp on how long LED fixtures last, in the real world.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 31 Oct 2016 19:08:08 +0000 (UTC), Algeria Horan
Dunno. Your real world LED application is not the same as my real world LED application and the science fiction world of marketing tech products.
Perhaps you might be interested in "lumen depreciation", L90, L80, L70, TM-21, LM-80, LM-40, etc. <http://www.p-2.com/blog/lighting-basics-lifespan-lumen-depreciation-l70-tm-21-lm-80-lm-40-and-other-confusing-yet-useful-terms/ There are some interesting methods of calculating LED life such as: "Reported TM-21 values have an upper limit of 6-times the number of LM-80 test hours. So if an LED chip is tested for 6,000 hours, its max reported TM-21 lifetime would be 36,000 hours. If the chip was tested for 10,000 hours, its max reported TM-21 would be 60,000 hours." Magic is a reallly nice way to produce bigger numbers.
One can also be devious: "It’s worth noting that there are two different types of TM-21 ratings, "reported" and "calculated" ratings." Sigh...
Few run 30,000 hr tests. At 8,760 hrs/year, it would take over three years to run the test, by which time the product is obsolete and replaced by something new and improved. Instead, they run a HALT (Highly Accelerated Life Test), which is faster, and presumably produces the necessary inflated figures: <https://www.google.com/search?q=led+accelerated+life+testing The basic idea is to test the LED at various elevated temperatures, connect the dots on a graph, and extrapolate to how long it might last at some specified operating temperature. At 25C it will last nearly forever, which assumes that the LED has some form of cooling system. One can generate amazingly large and impressive numbers this way. Despite my derisive comments, it does work quite well when performed honestly and where the test parameters are sane.
For example, IEC 62717 and IEC 62722 LED life testing standards both demand 6,000 hrs of test time, with total output in lumens recorded every 1,000 hrs. Measuring lumens accurately requires an integrating sphere: <https://www.google.com/search?q=integrating+sphere&tbm=isch which might explain why they don't just use a common light meter and why there are so few measurement points. Some detail on how Luxo specifies its lifetimes: <http://glamox.com/gsx/led-lifetime-and-the-factors-that-affect-it I doubt if we will ever see detailed life test results from Costco LED's as you might from higher priced LED luminaires.
--
Jeff Liebermann snipped-for-privacy@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 31 Oct 2016 15:13:01 -0700, Jeff Liebermann wrote:

Yikes Jeff!
You would bring up *more* complex LED-lifetime terms to figure out!
And, you would also find MARKETING BULLSHIT in the mix!
LM-80 = a standard for measuring LED lumen maintenance & depreciation LM-40 = time for 50% of the lamps in a large group to burn out L70 = time to degrade to 70% of its original lumens L80 = time to degrade to 80% of its original lumens L90 = time to degrade to 90% of its original lumens Reported TM-21 = predicts lifetime using LM80 + optimistic magic math Calculated TM-21 = predicts lifetime using LM80 + more optimistic magic math

Hence the "magic math" on the lifetime figures...

Seems to me that the "LED lifetime" figure everyone is quoting in this thread and in other threads is total bullshit, so far...

Well, I was sick of replacing very expensive non-standard Philips fluorescent bulbs, so I bought the LED light fixture from Costco just to get rid of the non-standard wacko shaped bulbs that kept burning out anyway. http://uk.rs-online.com/web/p/non-integrated-compact-fluorescent-lamps/4075597/
I asked Costco and the manufacturer to provide the information on the lifetime of the LED light that I did buy.
I didn't get anything more than "hey, the warranty is 5 years so that's how long it lasts".
What irks me is that they seem to never have run into someone who doesn't accept that bs as an "answer" to the question of how long the light fixture is expected to last.
I'm guessing the LED light fixture I bought lasts no longer than a couple sets of incandescent bulbs would have.
Time will tell.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 1 Nov 2016 02:38:46 +0000 (UTC), Algeria Horan
I should have guessed it was you. I like to explain how things work, without offering a judgment or opinion. This type of question really belongs in Candlepower Forums.

It happens. I have some marketeering experience somewhere on my resume. Speaking of bullshit: "How LED Lighting May Compromise Your Health" <http://articles.mercola.com/sites/articles/archive/2016/10/23/near-infrared-led-lighting.aspx >LM-80 = a standard for measuring LED lumen maintenance & depreciation

Nice summary. Sounds about right. I believe there are a few other standards that I missed. Standards are a good thing. Every company should have one.

Nope, because it's all we have to work with. Like I ranted, nobody does 30,000 hr life tests. Therefore, nobody knows the "real world" lifetime of an LED light. The best we can do is parametric testing, accelerated life tests, and the usual guesswork. The first two are quite valid and result in numbers that usually come fairly close to reality. The guesswork, you can guess what I think.
It's much like MTBF (mean time between failure) which attempts to estimate the life of a device based on historical tests and operating conditions. These component estimates are conglomerated into a figure for the device. However, the intent is not to estimate the lifetime, but rather the number of expected failures in a population of LED's. "What Every LED Engineer Needs to Know About MTBF" <https://www.fairchildsemi.com.cn/Assets/zSystem/documents/collateral/whitepapers/LED-Lighting-MTBF-White-Paper.pdf (Note: I haven't read through this yet)

600lumens / 9watts = 67 lumens/watt. Barf. Philips claims 200 lumens/watt and Cree claims 300 lumens/watt: <http://www.philips.com/consumerfiles/newscenter/main/design/resources/pdf/Inside-Innovation-Backgrounder-Lumens-per-Watt.pdf <http://www.cree.com/News-and-Events/Cree-News/Press-Releases/2014/March/300LPW-LED-barrier You may not see that at Costco for a while, but maybe if Philips and others get back into the LED biz. <http://www.memoori.com/samsung-joins-philips-siemens-in-led-lighting-exodus/ It's not too obvious, but both claims assume that the LED is cooled to approximately room temperature.

For good reason. From the point of view of the manufacturer and vendor, the ideal product blows up 1 day after the warranty expires. I've ranted on the topic before, where simulation and modeling tools are used to insure that multiple parts all fail just after some preset time limit. My favorite example are GE(?) water heaters with 6, 9, and 12 year warrantees, and roughly proportional pricing, but where the only difference is the type and size of the anode rod. Details if anyone wants them.

Talk to me in 30,000 hrs and we'll compare notes.

You have a talking clock?
--
Jeff Liebermann snipped-for-privacy@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

One more. ARL: <http://www.bulbs.com/learning/arl.aspx "Average Rated Life (ARL) is how long it takes for half the light bulbs in a test batch to fail" I seem to recall others, but I'm too lazy to Google.
--
Jeff Liebermann snipped-for-privacy@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 01 Nov 2016 09:08:01 -0700, Jeff Liebermann wrote:

I remember reading in your prior reference that the time for half to fail isn't useful for LEDs though, since they fail differently.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 01 Nov 2016 09:02:59 -0700, Jeff Liebermann wrote:

Yeah, it's me. My friends joke I'm half the Internet alone.
I'm always solving problems, asking questions, delving deeper, etc., as are you (but you don't like to admit it).

You have always been balanced, ever since I learned from you how to set up my WiFi rooftop antenna on a.i.w years ago, when you still frequented that forum (before you absconded to s.e.r that is).

Me too, truth be told (but I try to hide my curricum vitae far more so than you do).
When I was in Marketing, we made hay with any advantage we could, and we swept under the rug all the disadvantages. Plus we said things like "better" and "new" and "more" since they couldn't be easily disputed.
Basically, we took whatever it was that the engineers gave us, and we marketed the shit out of it, so that it *looked* like gold in the literature.
But it was no different than anything else was. Every good thing had a bad downside to it.
Like everything on this planet does.

OMG. I'm losing out on all that healthy infrared radiation! And that EMF is the "leading cause of blindness" in the USA! Quick. Gimme one of those famous infrared saunas in Santa Cruz hippytown!
:)

Thanks for noticing. I generally read all your references. If you are gonna go to the trouble to reference them in a thread I authored, I'm gonna go to the trouble to at least skim them (I read fast, very very very fast, faster than most people can talk, and I type fast too, so it's easy for me. When I was a kid, I was in a special reading program for the gifted, where they had a machine that forced me to read faster and faster and faster - dunno why my parents subjected me to that - but they did.)

I think the important point is that we each can pick the standard that makes the most sense to us, but also, that information has to be readily available to us.
I'm not sure yet, which is the readily available standard, but I'd prefer the L70 myself, to be the standard that I get the information on.

At the moment, I'm guessing the one LED lamp I have will last no more than 4 or 5 years. (Call me up in 5 years and I'll let you know how it turned out.)

Except that every once in a while, there will be failures in the drivers that I don't think are being tested here. Are they?

Understood.

Makes sense.

The abstract mentions MTTF, which is essentially what I'm asking in this thread, I believe, whereas MTBF is for repaired items (according to the abstract).
It implies that we should use MTTF since we're gonna throw out the LED fixture once it fails us.
That Fairchild paper goes into details (e.g., how to accelerate and what happens if the failure rate is 0), but that's the net I take out of it by a quick skim.

The funny thing is that there are so many stupids out there who talk about "warranties" as if they're NOT purely marketing bullshit!
On the car forums, I hear all the time people comparing batteries by their warrantee, as if the warrantee conferred some magical quality on the electrical and lifetime properties of the battery!
They even compare *tires* by warrantee! Geesuz. It's sad how stupid people are, in general. Very very sad. Sigh.

Interesting. Very interesting. I just had a water heater go, in fact, and, um, I shouldn't say this, but I had never replaced the anode. All that was left was some whitish stuff and the inner steel wire. The heater corroded in 7 years, but that was my fault for not replacing the anode (although it was almost impossible to twist off, so, if you're gonna replace anodes, at least crack the top hexnut every six months or so).
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

root,
Believe it or not, the tests are designed to reflect real world usage.
You keep implying that there is something wrong with all advertised LED bulb lifetimes. Then you tell us you need to Google search to find out if that's true.

...

Jeff,
Actually, 3 years wouldn't be useful for anything but the decidedly "not realworld" test case of continuous use.
It would take considerably more than 3 years to test the common on at night, off during the day test case.

You could have left the word "inflated" out of that sentence. It's an insult to the rather clever testing that you described.
... snipped test description.
--
Dan Espen

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 01 Nov 2016 00:14:43 -0400, Dan Espen wrote:

Real world seems to *always* be less than advertised lifetime. At least for me they seem to be.
I'd be pleasantly surprised if LED bulbs last 5 years.
What fails? a. The electronics! b. The bulbs.

We know the manufacturers lie, or, more accurately, the MARKETING people, who make the labels, say only what they want you to think.
For example, they'll tell you an LED itself lasts 50,000 hours, but they won't tell you the "driver" lasts anywhere from zero to a couple of years.

The one thing about Jeff is that he's shown himself over the past decade to be a well-balanced person who is not swayed, like most people are, by exaggerated claims, whether they be any claim by Apple for their WiFi reception, or a bogus claim by LED manufacturers (such as the ones I received on the phone yesterday) as to lifetime.
He'll deny this though, as he doesn't like accolades, but he is always on the money, and, he almost always provides *proof*, something which you need to provide also in order for us to believe your claims (I'm not saying you didn't or don't provide proof - I'm just saying that Jeff almost always does - so what he says carries weight).
Also, Jeff runs his own *tests* of tons of things, which are tests that most of us have never run, they're that detailed (ask him about router claims versus the real world some day!).
Jeff is like I am. We prove what we say, and we provide references, and photos and other reliable statistics. You'd need to do the same to hold water with us.
Anyway, fact is, I have a bulb, in my very hand, incandescent, that failed in two days.
https://s15.postimg.org/92aoq3xej/burned_out_in_two_days.jpg
Notice the package says "1.4 year life".
https://s18.postimg.org/602tefda1/ge_bulb_burned_out.jpg
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

See Appendix C as I stated in my other post. Then tell me how your methods are so superior to those used by the experts.

Do you honestly think that proves anything?
I have 3 CFLs in a lamp post since 2000. Just this year, one of them gave up the ghost. They are activated when it gets dark. So, they're on at least 12 hours a day, every day. Construct some statistics out of that.
--
Dan Espen

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 01 Nov 2016 10:59:44 -0400, Dan Espen wrote:

I understand your point, which is that I can say it lasted 10,000 years. If I could prove it completely, and if it was worth the effort, I would, just like Jeff *proved* that WiFi reception in routers was NOT what the manufacturers claimed.
But you are like those people who say "prove it" to everything, which is fine, but *you* have to provide some semblance of a reason to go to the effort to prove things that we just have to accept on faith.
I was backing up your unproven claim that Jeff was not being balanced, in effect, when I know, from the last decade on s.e.r and a.i.w that Jeff "is" well balanced, and he proves what is worth proving.

You entirely and completely missed the point. Did you buy too many arguments this week?
All I was saying is that your claim against Jeff's veracity are completely unfounded. You're entitled to your opinion, but if I asked you to prove that you had sex with your wife five times this week, do I really expect you to prove that?
What I'm saying is simply that your criticism of Jeff was unfounded, if you look at the entire record. And, I'm saying that 11 cents per kilowatt hour is a magical number entirely unachievable by me, in California.
If you claim otherwise, I'm only asking you to attempt to back up your very own claims with fact, as I did with Jeff, and as I did with the price of electricity in California.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 11/1/2016 10:59 AM, Dan Espen wrote:

That is the reason they last so long. What people don't realize is the life of fluorescent, whether tubes or CFL, is shortened when they are subject to constant on and off. They are not made for constant on/off action unless the fixture contains a "program start" ballast. Due to the cost of that ballast, they will not be within the common hardware store fixture. When fluorescent lamps remain on all day, chances are you can see 10 years on them. The average home will not see that life cycle since it's not common to leave them on all day everyday.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 1 Nov 2016 11:23:09 -0400, Meanie wrote:

I don't run statistics, but I appreciate what you wrote because my fluorescent lamps don't last more than a year or two, it seems.
I used to mark the bulbs with a Sharpie, but I stopped doing that long ago. I don't think I *ever* got anywhere near the claimed life.
But we turn them on and off a few times each day.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 11/1/2016 10:11 AM, Algeria Horan wrote:

I can assure you, we replaced incandescent floodlights back in 2010 with LEDs in one of our buildings. These lights are on 8 to 10 hours a day, if not longer and we've yet to replace any of them (over 50).
I understand your skepticism as a general consumer, but I've been experiencing the products first hand. I've been dealing with LEDs for many years and I agree, they do diminish in brightness after their manufactured rating, the quality lamps last up to their claim if not longer. Of course, we don't purchase the cheap ones you buy at Home Depot.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 1 Nov 2016 11:18:11 -0400, Meanie wrote:

This is good to know because the whole point of this thread is to nail down the actual life of the lamps.
Of course, you can't expect me to NOT buy at Lowes or Ace or Home Depot, for quantities such as we buy for a home as replacements, so the word 'quality lamps' is to be taken with a grain of salt.
But at least it's good to know that you *understand* that an LED is never as bright as it was on its first day, and that cycles, and heat, and vibration exacerbate the existing cracks between crystals, such that LEDs drop off exponentially in light output over time.
As stated in the standards that Jeff kindly referenced early on in this thread, the diminished light output is very difficult to detect, since it happens over time, and since there may be other bulbs compensating for the lack of output, such that an LED bulb that has actually reached it's L70 lifetime may not be easily observed by you.
Nonetheless, if the driver failed, which I think can be the weakest point (that premise needs to be explored), you'd know that. But you might not know when any particular bulb has reached its L70 point without isolating the bulb and actually measuring the output (since the gradual decline in output isn't going to be suddenly noticeable, according to that report Jeff referenced).
My point is that things failed, perhaps, and you don't realize it. But that needs to be explored since you'd know of some failures (but not all, unless the bulbs are isolated, and if you have a keen eye for such things).
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Which is impossible to do based on unreliable anecdotal evidence in an usenet newsgroup read by perhaps a hundred people.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

HomeOwnersHub.com is a website for homeowners and building and maintenance pros. It is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.