Banning 5mm LEDs from general illumination and an interesting comment from Nichia

Sorry about the long thread title here but there are two sides to this coin, both of which make for interesting reading. Many here have said 5mm LEDs are unsuitable for general lighting applications as they tend to degrade and color shift too quickly, sometimes within weeks. Here is an article supporting that point of view, calling for a outright ban on using small, indicator-type LEDs for consumer general lighting. The most interesting part is this paragraph:

If we don't pull together as an industry in combating the junk, consumers and small commercial users are in for a rude awaking. Not all LEDs are created equal and that message needs to get out into the consumer and commercial markets ASAP. Before we ever see LED light bulbs on a store shelf, the merchandise buyers should demand that any "LED light bulb" be US DOE Energy Star compliant. If not, they should insist on seeing the independent test data that verifies compliance. They should run all new LED lamp suppliers through a continuous 1,000 hours (6 weeks) "life test" to see if it maintains the same color, or if there is any color-shift compared to one fresh out of a box. Yes, they should even ask for the data on the drive current each LED will see, and compare that to the device ratings. Of course, they aren't going to do those things... so the next best alternative is to take the "easy" approach and just say "NO" to those 5-mm LED "cluster" lamps and fixtures and "YES" to buying only US-DOE Energy Star compliant luminaires and modules! Caveat emptor!

Now along comes Nichia in response to this article, as discussed here. More to the point, Dr. Daniel Doxsee, General Manager, Nichia America, said:

It turns out that recognizing that there were plenty of non-capable 3mm and 5mm packaged devices out there, Nichia set out a while ago to build lighting-capable versions, and succeeded with their DS series. I had missed this, Bob had missed this, and it looks like a good chunk of the industry missed this. Since mid-2007 Nichia has offered a line of 3mm and 5mm packaged white and color LEDs that incorporate proprietary resin and other design features that get the heat out, stand up to the heat that's left, and last 40,000 hours at rated drive currents (or longer if you drive them more gently).

Note: There appeared to be a typo in the article where 40,000 was written as 40,0000. I assumed they meant 40,000 rather than 400,000 as a 400,000 hour life isn't common even with power LEDs.

Interestingly, these comments seem to be borne out by a recent test here of Nichia 5mm's. So it does indeed seem that some 5mm LEDs are perfectly suitable for general lighting applications. While I love power LEDs and they would be my first choice for most fixed lighting uses, I can see the advantages to using 5mm types in certain applications. Now at least we know the Nichias are up to the task. And in my mind this justifies the generally higher selling price of these versus cheap eBay LEDs. Also of note are that Cree's 5mms seem to hold up well in lifetime tests. So all indicator LEDs are not created equal after all!

I agree with this 100%.

Most people who are not REALLY informed cant tell the difference and end up dissing LEDs in general because they were sold crap.
An acquaintance of mine ended up buying 5 or so of those showerheads, and they died during the first week. He didnt know he was buying crap, and he as a habilitation in physics.

But about that link testing the 5mm leds:
a) From the beginning, the test seemed very suspect to me. The poster claimed that it was done with professional testing equipment, but still: If one of my experiments would yield such a contraintuitive result, i would start to double and tripplecheck, and not post it online first (with undefined sample size)
b) A showerhead is a completely different situation than running a single led

My friend works at H&R Block helping out the accountants, and one day while I was eating lunch, my phone rang. It was my friend, calling from work, to ask me a question about LEDs! I was stunned that he called me from work, but also pleased that he "knew" to call the right guy.

A guy from a retail store was going around liquidating a flashlight with a ton of 5mm LEDs for $10. My friend called me asking if it was worth the money. I said, "give me ten minutes, and I'll let you know." Though I can't recall the name of the product, I did find it on Craig's site, and after seeing his review, decided it was certainly *not* worth the money. His intended use was as a blackout light, and I pointed out that for a bit more money, he could get a lantern popular on these forums for under $30. When the traveling salesman returned, my friend declined to buy the item, citing my advice.

The main reason I advised him against the purchase is because many 5mms LEDs are of suspect origin, and they generally do not last/look pretty/give off a lot of light.

Should they be banned from general use? I do not know. It's an interesting idea, thougn.

I think banning unenlightened consumers would be the better approach. There is junk and quality in every product line. It's up to the consumer to research his/her purchase before buying. It only takes a few minutes on the internet to know what you need to know about a certain product to make a wise choice.

but customers are right. Misinformed, lazy and unknowing. Expecting them to understand what they're buying is overestimating the intelligence and - more to the point - the desire to know of Joe Average.

So, apply a standard. We have them here for fridges and stuff: Class A to G. It's pretty simple... It's even on lightbulbs. Why not there?

Bret

I think banning unenlightened consumers would be the better approach

Click to expand...

With due respect to Nichia Corp, I consider non-power LEDs as per the category this thread is directed at to be so dominated by 'jewelery class' LEDs it's often not worth the effort finding the good ones.

However, I've had enough bad luck with so called 'power' LEDs suddenly dying for no reason I'm not cutting them a break either. When a 10 cent LED pukes for no reason nobody bats an eyelash, but when a $8.00 LED dies for no reason it's naturally assumed the reason is anything but the fault of the LED.

blasterman said:

However, I've had enough bad luck with so called 'power' LEDs suddenly dying for no reason I'm not cutting them a break either. When a 10 cent LED pukes for no reason nobody bats an eyelash, but when a $8.00 LED dies for no reason it's naturally assumed the reason is anything but the fault of the LED.

Click to expand...

I have dealt with a bunch of LEDs and can't think of one instance where one failed for no reason. There is always a reason.

Then what's the 'reason'?

I mean, if it's made by Cree, then there's absolutely no way it's a Q/C issue. (/cynical off)

blasterman said:

Then what's the 'reason'?

I mean, if it's made by Cree, then there's absolutely no way it's a Q/C issue. (/cynical off)

Click to expand...

There's a bunch of ways that they can be killed. They all come down to one thing though. Human error. Be that during the modding/installing or due to faulty design of the device it is going into including mechanical and electrical stress. LEDs don't just stop working for no reason regardless of the brand. Around here most die from people who do not know what they are doing modding their lights.

I have to say I have a bulb showerhead (E27, 220V) in a desk lamp. It is rated at 1W and have 15 white leds.

Got it a year and a half ago to check if LEDS were bright enough to start to replace low power incan bulbs (I didnt know almost anything about power-leds then), but I was a bit dissapointed about brightness and it is too focused to use as general illumination. In a desk lamp it is not too bad and it has been used (and abused) a lot (up to 10 hours day, sometimes I go bed and forget to switch it off) since purchase with no change in tint nor bright so I think it survived the "6 week test".

I suppose there are many out there who would push the 5mm LEDs to color shifting or failure but I'm not one of them. My very oldest LED's are still working as well today as when I purchased them and I've never had a single failure of an LED on any piece of electronic equipment including the types on appliances and other 3mm types. It seems kind of strange that there is so much concern but I'm sure they're seeing stats and figures that I'll never know about. Interesting nonetheless.

The problem with white 5 mm LEDs is the same as the problem with red 5 mm LEDs and green 5 mm LEDs. It is the junction temperature of the LEDs. Higher the junction temperature shorter the life span (70% of light output point). More than 99% of all 5 mm LEDs only have a life span of 1,000 hours or less at 25 degree C and driven at rated current of 30mA-50mA. If over driven the life expectancy will be much shorter than 1,000 hours. This is because for 5 mm LEDs the only way the heat can escape is through the cathode lead of the LED. Most of the lamp designer completely or partially ignore the thermal management issue. Candle Power Forum had a lot of discussions of LED traffic lights failing everwhere especially in the tropical area. A lot of those LEDs also uses steel lead frame instead of copper lead frame. Since the thermal conductivity of copper is 8.7 times of the thermal conductivity of the steel, steel lead frame 5 mm LEDs will have even shorter life than copper lead frame LEDs. All you need is a small magnet to tell whether the lead frame is steel or copper. Another reason that thermal management issue is ignored is the difficulty to measure LED junction temperature directly and accurately. Using thermocouple measuring a point close to the LED cathode does not measure the junction temperature. And 5 mm LEDs are all enclosed in optical epoxy so to put thermocouple directly on the junction will destroy the LED. The only instrument that can measure LED junction temperature directly while it is being driven with a current source can be found at Forge Europa website. Until the junction temperature problem of 5 mm LEDs is solved there will be more failed LED traffic lights and PAR lamps and other light engines.

I have too a 3mm led flickering in a ubercheap direct drive 3xAAA powered... thing (I refuse to call it a flashlight). It was a gift in a party and used it maybe 5 non-consecutive minutes before it started to flick.

Leds can fail, but I think only the cheapest of them.

I can believe how many crappy things you can end having even if you dont buy them. I dont give them away because they are too cheap. Maybe when I get short of space... :shakehead

auburn said:

The problem with white 5 mm LEDs is the same as the problem with red 5 mm LEDs and green 5 mm LEDs. It is the junction temperature of the LEDs. Higher the junction temperature shorter the life span (70% of light output point). More than 99% of all 5 mm LEDs only have a life span of 1,000 hours or less at 25 degree C and driven at rated current of 30mA-50mA. If over driven the life expectancy will be much shorter than 1,000 hours. This is because for 5 mm LEDs the only way the heat can escape is through the cathode lead of the LED. Most of the lamp designer completely or partially ignore the thermal management issue.

Click to expand...

Sorry, but that is wrong.

5mm LEDS that are NOT WHITE and are run at their rated drive currents have life spans WAY OVER 1,000 hours. Your statement is not based on fact or knowledge. For one, most 5mm LEDS have a rated current of 20mA-30mA. Very very few are rated high. That is the rated operating current, not the "Maximum" which they are not intended to operate at for long periods.

A mounted 5mm LED with a copper lead frame will have a thermal conductivity, die to ambient, properly mounted of say, 400C/watt. I could argue lower, but lets go with this. At 25C, a 20mA RED led with a 2.5V forward voltage will only be 20C above ambient. Even a 3.5V blue at 30mA is still quite tolerable. Now if you put 1,000s of them on a small board, yes the thing overall is going to be too hot.

Now when you look at streetlights, you have several issues. There are a lot of LEDs on a board where it is tough to get the heat out. Not impossible, but odds are not done right. Two, 5mm LEDs are more susceptible to mechanical failure due to thermal cycling due to the rigid LED/bond wire mounting. Three, you have a ton of LEDs, so the chance of a point failure is significantly higher.

All of this is much different from the primary reason why WHITE 5MM epoxy encapsulated LEDs degrade so quick. It is not the LED die that is degrading, it is the epoxy that is yellowing at the interface of the phosphor and the epoxy by a combination of light intensity and heat. High Powered LEDs use silicone as an encapsulant for exactly this reason. They also have the aforementioned mechanical issues, but that contributes to point failures, not degradation. Yes, they are not as good at getting heat out, but last far more than 1,000 hours if treated as they are intended to be.

Semiman

saabluster said:

I have dealt with a bunch of LEDs and can't think of one instance where one failed for no reason. There is always a reason.

Click to expand...

I tend to agree. I've been working with LEDs for over 20 years and when one fails, there is always a reason. Even the cheap ones from overseas are orders of magnitude more reliable than what they replace (i.e. cheap incandescent lamps) if used properly (i.e. think of them as 5mA LEDs, not 20 mA ones). Case in point-yesterday I finally took down my LED Christmas lights. They were outdoors since mid-November in all kinds of weather, and turned on by timer each night for 7 hours (that's about 700 hours to date). Didn't notice any fading of either the white ones are the colored ones. More importantly, out of a total of about 2100 lamps there were only 20 failures. They all occurred on the six 100 light strings where the LEDs weren't adequate waterproofed. Of these 20 failures, 17 were directly attributable to corrosion of the leads (in some cases they corroded in half, open circuiting that portion of the string. So only 3 "real" failures out of 2100 lamps and close to 1.5 million lamp-hours. And I'm not even 100% sure those failures weren't also caused by water damage, perhaps penetration into the LED which wasn't readily visible. But even if they were "real", MTBF of half a million hours doesn't sound too bad, and this with cheap LEDs used in low-cost products.

I have been running LED Christmas lights for about five or six years now. I must say that the newer ones are much more reliable even within the same brand. On some of my earliest strings, virtually every one of the green and many of the blue LEDs in the string have died and had to be replaced. Since they were sealed, this involved cutting out the bad lamp, testing polarity, and soldering in a new lamp cut from a 'spares' string. Maybe water penetrates the sealed strings, maybe the LEDs overheat in the solid plastic housing, maybe both. Or maybe the LEDs were just of very poor quality back then.
I have some early blue strings whose LEDs have so high a Vf (around 4V at 20 mA) that the string was given virtually no ballast resistance in order to get the current high enough, since they had 35 in series. The slightest surge comes along and the current goes way up and poof! Failures right and left.
Even today these strings use cheap LEDs and there is a much higher random failure rate than I would expect were they using Nichia or some other reputable brand. (of course that would make them completely unaffordable). I have made a number of deep red strings using Lumex LEDs with not one failure ever.

I have used about 170 cheap 3mm Warm White LEDs to retrofit incans in step lights used in a theatre. They run about 12 hours a day, and so are around 5000 hrs. The LEDs are spaced 2 inches, leads cut short and pushed into little brass sockets set into a wire, and this laid into aluminum channels. 10 to 16 run in parallel on one resistor pack, the LEDs having been Vf-matched to within 0.01V. They seemed to be working fine. After 25 months only four have completely failed, two of those in the last two weeks. When I replaced the first two of those four, I discovered that all the LEDs had uniformly faded and dimmed dramatically, so that they look about 50 to 60% as bright as new and have lost a lot of the green/yellow (phosphor degradation?) Oh, and they are running at 10 mA per LED on regulated 5V supplies.
So if a showerhead lightbulb has seemingly worked fine for a great many hours with no degradation, I invite you to compare it to a lightly-used one of the same type, from the same purchase, that looked the same at the time of purchase and see if they still look the same. Those 3/5mm clusters tend to trap a lot of heat, with no good way to heatsink it, and they use steel leadframes, so the degradation is likely pretty severe.

I've run numerous tests a running 5mm LEDs at 30ma for 192 hours and I find that all but one I got from various ebay sellers faded. The one that didn't fade was not very bright to start with and had a very bluish cast. ebay has not been a good source for 5mm LEDs - not even JELEDs.

Warm whites I tested faded severely and some did okay.

After a month at 15ma, many of these LEDs faded as well.

The best LEDs I've put my paws on are from Nichia, Cree and those heaven sent Rat Shack 276-0017 medium white ones.

If you can deal with a wide angle beam, try the low current (20-30ma) flux LEDs. None that I have tested faded. They have much better thermal management and seem better for general lighting.

JohnR66, which wide angle low current flux 20-30mA LEDs?

Bill

Bullzeyebill said:

JohnR66, which wide angle low current flux 20-30mA LEDs?

Bill

Click to expand...

superbrightleds.com were good but expensive.

Ebay seller "Light of victory" warm whites were good (very warm color though). The white ones are very bright, but tint is a lottery (mostly purplish white). I have found them to be very static sensitive as well.

Have not tried JELEDs flux LEDs yet.