Here’s one LED light that didn’t make it to 50,000-hrs

@BlairH - you can't backpedal yourself out of an idiotic comment (trust me, I know this here more than anyone...). LEDs are USUALLY current controlled (we have a patent filed on voltage-controlled-intensity LED bulb at Zelentek, actually, which makes dimming super easy) and even with your 24VDC centralized supply (dumb if you know how a house is wired) you still need current control, aka a POWER SUPPLY, in the bulb. You could have a centralized current source, but then you run into Christmas Tree Lamp syndrome in trying to find a failed bulb, not to mention the applied voltage going to high heaven if an open circuit occurs. As far as solder joint failures go, three words: MADE IN CHINA.

With the typical arrogant CEO and CFO driven mentality to soak the customer for all that he has and to provide the cheapest & fastest to the trash can items of today, I will not use CFL's nor LED screw in replacements for fear that the cheesy electronics will burn my fricking house down to the ground.
My experience is that all claimed advantages are outweighed by the horrible waste of premature failures of these items never mind the fire hazards that some of them present.

Just had a 50 W Sylvania par type led lamp give out after maybe 10 hours of service. The MYTH of led lamp longevity is just that!

Why don't we bite the bullet and go with low voltage lighting. It is a small matter to diconnect the A.C. at the breaker and fit a low voltage high current power supply between the breaker and the lighting circuit. Then the LED lights will burn almost forever with no electrolytic capacitors to kill them.

My thought: Current screw in LED replacement bulbs are short term in the eventual path to better LED specific fixtures with seperate power supplies away from the LED for better heat control. Table lamps, emergency lights and recess fixtures might be the fist candidates.

Dear Muna
As you said your lamp damage in one years , it is not wonder in my idea .
I will say how to make a good LED lamp in this mail .
From your idea , it is damaged by power supply , it is right of your idea .
LED lamp is different of other lamp , it need power supply inside the LED lamps . because it works by electron . and other lamps work by Element loss . so LED is environmental protection .
1 . We need design a hull of LED heat disspation . it is help of life only to LED chips . a good design of hull to make LED heat disspation will help LED chips work more then 50000 hours . please understand , it is not to the lamp , only to LED chips . if we want LED lamp also had 50000 hours life , we also need a good power supply can work 50000 hours .
2 . Power supply it is very chaos in the market . lots of LED lamps damaged in one years because of power supply . so we need a good power supply to work inside LED lamps . the price of power supplys had large differents . only blub (small lamps) , the price of power supplys Change a times more differences . so how to choose factory you work with is very important .
As I said ahead . the design of hull is base , we need a good hull to head disspation , but not only , we also need a good power supply .
Which hull is gooder , it is Physical knowledge . as I know , hull is better of AL , high purity of aluminum is better .
Which power supply better can we choose , the best is had UL , but it is too expensive , so we need get CE , really get CE power supply , don't listen to other people said , you need inquiry .
your's kaison zhang , kx@kaixin-china.com

I would not be at all surprised if it turns out that LED lamps like these don't live up to the lifetimes claimed by their manufacturers. I know that (in my experience anyway) compact fluorescents rarely come anywhere close to their rated lifetimes. In fact, more often than not, incandescents seem to outlast CFLs at our house.

Cold solder joint. How many cell phones, stereos, MP3 players, TV's and appliances have been repaired with this failure? Or worse, how many been trashed due to simple lack quality. Just to save the MFG a few pennies? (yuan?) We HAVE to look at the larger picture.

I installed 4 watt mr-16 LEDs running on 12 vac from a transformer. From China, the cost was the same as Halogen. No failures yet. Low tech drive circuit well protected from power line funnies.

Brad Wood commented:
"In the long run it would seem that distributing low voltage d.c. and providing a simple current regulator (even just resistors without too-great voltage drops) at each LED ... "
There are a couple of problems with that:
1) Lower voltage means higher current and therefore larger wires and more transmission losses (voltage drop)
2) Just using resistors to regulate LED current won't work. Tried it. Several times. The result is much shortened LED life. If you are runnig the LED near it's maximum rated output it must have an accurate current regulated source or life will be diminished.
This is not to say that there is no place for a low voltage distribution system. Over fairly short distances (one or two rooms) the losses are not so bad if the wiring is sized properly. And if your power source is solar cells producing 12V and charging a 12V storage system then 12V distribution makes sense in that it eliminates the losses inherent in converting from 12V to AC and then back again to low voltage at the light fixture.

Last week, I installed a new track light in the living room. It has 5 GU10 50w bulbs. They look like MR16, but they sadly run on 110. But lo and behold, I bought a bulb with 3 CREE 3 watt LEDs in the same package and powered by 110. It works! We don't use it much but the LEDs are CREE and the lamp comes from China. The design is so small, it's no bigger than the 50W Halogen bulb.

I hope you're using unleaded solder as you could be breaking the law in most countries :-)

In the long run it would seem that distributing low voltage d.c. and providing a simple current regulator (even just resistors without too-great voltage drops) at each LED cluster would work fairly well. Obviously this is not a retrofittable approach, and is additionally fraught with regulatory hurdles that will bog things down. But the idea of having a complex and noisy switching regulator at each "bulb" seems to be inviting failure.

A common thread in these articles about LED and CFL bulbs is quality and the tendency of mfrs to cut corners and sell junk that quickly ends up in landfills to unsuspecting customers. This is a huge environmental and economic problem if one looks at quality problems in bulbs as just a microcosm of consumer-product quality in general. These are usually items that are too low-cost to be reviewed in the likes of Consumer Reports. What's a poor consumer--or even a busy engineer with too little time to research, say, a $25 toaster oven--to do? (we recently had a new Sylvania toaster oven fill our kitchen with smoke)
Also, note that outsourcing is not necessarily the problem: a company with good quality controls can manufacture great-quality products overseas or at home.

Great comments everyone, nice article. This reminds me of the usual struggles a new industry goes through. I make it a policy to test my products before selling them. Of course I push the limits of application. That said my shop has buckets full of junk that doesn't work. For the most part they come from offshore with poor components bad designs, and poor manufacturing techniques. I have found some really great products here in North America and Europe that are really engineered and assembled brilliantly. As usual though it is a tough find, after 3 years of research my conclusion, "the end user is getting burned and will continue to get burned by the inferior junk on the market. Everyone wants to enter the consumer market as cheap as possible with the most profit, lacking engineering, durability, and projecting the LED market as more junk that the consumer has to throw away in landfill. It is so unfortunate that the NA corporate world has chosen outsourcing instead of home built on home ground by the people that buy their products...

I agree the problem is because of the bad soldering. It can easily be solved by using connectors. I also discover the bad capacitor can reduce the life of the LED bulb. As the LED light claimed the life is 50k hours, it means 5.7 years if using 24 hours a day. How can you check the LED make it to 50k hours.

cgqnw commented:
...
"CFL almost burnt my house down. ... I have never heard of an incondescent 60 watt bulb burning a house down." >> Ask anyone who has done a lot of fire investigations - all that is required tio get a fire from an incandescent lamp is to leave it in contact with flammable paper or cloth while operating. > I believe museum curators would disagree with you on this.

Nice teardown bits, thanks!
Just stock the early-failure CFLs away from the early-failure LEDs, so the kid engineers we get stand a chance of good RoHS-compliant materiel receiving audits. Additionally; how far are we from re-factoring these into either: all low-oxygen copper, welded, (bis-azo?)passivated heatsink-shell liner circuits; or remote surround-speaker or mic. units, with lampage? At $70 a 120W-equivalent bulb, I'd better see '+6 Dwarven Warm Refectory LED' in the packaging.

As counter point. I have had, after about 20 years with standard filament bulbs, light fixtures that physically disentegrated from the effects of the heat. Replaced with CFL. Noticed that some take a lot longer to fire up than others, but I haven't had a fixture fail?

CFL almost burnt my house down. I have a table lamp that we leave on all the time. Base down in a heated room. No moisture. We came home from a short trip to find the fixture melted and black. The lamp shade was a pile of carbon dust except for a few shreds on the wire bracket. We were just real lucky that the fire did not get to the drapes 4 feet away. I have never heard of an incondescent 60 watt bulb burning a house down. Also, very little polution or energy used in manufacturing and they do not fade the furniture or pictures. I am sure that with all of the manufacturing energy and polution in all of the parts assembled for LED or CFL lighting that you will never come close to becoming green....Stay green and buy incondescent bulbs. I have several hundred that I have put away, and they were all made in the USA.

Funny, that although glass fbre optics (GFO) functional architectural lighting (still the most sustainable, energy-efficient technology now known) has been successfully used abroad for decades, it is under used in the USA.
Terminal inertia to change procedures, lack of practical education in the subject, and the impression that LEDs are now the sole answer to illumination (one size/type lighting does not fit all), place the USA at least ten years behind our foreign colleagues in Lighting.

Innovative as they are, current LEDs are not as sustainable or energy-efficient ss the expensive, massive but fanciful promotion they are given. Although the actual LED chip may survive, because the entire system is composed, like an automobile, with many parts from varied sources, any one of the companents can fail.
If the current LED lighting loses 30% of its light level in only five years, it is not the best choice for permanent architectural lighting, especially for a client without deep pockets.
Indeed, some LED installations are obsolete by the time they are completed, because there is such constant and rapid change.
In addition to needing additional energy to dissipate the heat created by present day drivers, LEDs are vulnerable to moisture, vibration, temperature and voltage change, points not often mentioned in the glowing ads.
A more exciting version is the Organic LED, produced by roll-to-roll, like newspaper. Its features give imagination few limits.

The issue is well stated above. The LED is a solid state device that will last a long time, (light output degradation aside) based on operating temperature. The rest of the assembly is a power supply. I designed and built many power supplies over 20 years. The reliability is usually spec'd in MTBF (mean time between failures) it is specified and proven with design verification testing and accelerated life testing. There are various methodologies to some based on MIL STD 217 to base the designed MTBF. The largest influence is operating temperature and number of connections. The issue raised is one of manufacturing quality control. If the manufacturer cares about this, they either manufacture it themselves, or more likely contract it to China and then spend additional money doing an additional set of quality control inspections. Based on what I have heard recently on NPR (national public radio) had a show about a visit one reporter made and his observations of the massive sweat shop conditions, only the top tier companies can have any influence on the contract manufacturer with regard to work conditions or product quality. As is typical in complex electronic assemblies the life of the product will follow a bathtub curve;

Interesting comments. Of course poor solder joints from offshore manufacturers go back at least 50 years, so that is nothing new. Of course the whisker problem is much worse with the leadfree solders. In addition, the cheap ones have a higher melting point and poorer wetting ability, so the probability of poor solder joints is much greater. As for what effect would poor solder joints have on connectors? The answer is that in most electronic assemblies at least one side of the connector is soldered to the board, also, it could have been a reference to connections, not connectors.

I've been working with LEDs for 8 years. The LEDs themselves are quite reliable and individually, they will likely provide the thousands of hours they claim. That said, the non-binned cheap LEDs in bulk from unknown manufacturing sources in China tend to lose color and brightness more quickly than reputable manufacturers (China or not).
The electrical wiring device industry (Hubbell, P&S, Cooper, etc.) have been making wiring devices in CHina for many years. THEY HAVE STRINGENT Quality Control and testing requirements, along with supplier contracts to enforce them. The China manufacturer is required to take back product that does not meet the quality stipulated in the CONTRACT.
So let's hope the Sylvania's Phillips' and other name brand manufacturers do the same. Only then will LED lighting devices become reliable.
I've had meetings with China mfr's reps hawking their LED products like ceiling fixtures, high-hat replacements, etc. BEWARE - they do their own QA and QA is not really a China Mfr's strong point yet. Personally, I'll use individual LEDs in my own designs, but I won't buy the complex LED bulbs that are showing up in Home Depot and Lowe's, etc.
CFL's were and still are a big disappointment.

Wow - Big suprise that the actual part didn't work as well as the "Glossy Marketing Brochure" said it would.

It actually frustrates me as a technologist, when things like solder joint reliability (someting we KNOW how to solve) impedes transitions to efficient solutions. interesting to note that the non-engineers set a policy to ban the use of lead in ALL products regardless of availability of alternatives. These are the same idiotic policy makers who will shrug their shoulders and deny responsibility and accountability for costs and impact. Can we make politicians accountable for the decisons they make?

The LED bulb I bought for the kitchen lasted ~3 months before half the LEDs went out. Another 2 or 3 months later the rest went dead. I didn't bother to open it up to check. It is far worst than the CFL that it replaced (which lasted ~1 yr) and if recall correctly cost >3X more. So far, the regular incand. light bulb lasts the longest and cost the least.

How did the company respond to this to you and
was there a corrective action?

I have 8 Cree LR6's in ceiling that run about 8 hrs/day for the last three years. Never a problem. I installed 6 Sylvania LED lights 4 months ago and have already replaced one. I've given up on cheap LED lights as they radiate a lot of noise on the line. (X10 doesn't work when cheap LEDs are on.) I have also installed 8 of the HD version of the cree can lights and hope they last as they are 20' up.

THANKS to all of you who are buying the new technology, you are helping to bring the price down and the reliability up.

So the LED bulbs have manufacturing problems? That is fairly common for a new product--incandescents have been around and the manufacturing enhanced for many years.
But I see no excuse for some nanny state rules that force me to abandon incandescents. If LEDs (or curlies) are so wonderful let them compete in the marketplace.

Your failure could have been a "cold" solder joint. If your picture of the red wire is "as you discovered it"; it has all the appearance of a "cold" solder joint. A "cold" solder joint is a manufactruing (human error) and quality control problem, rather than a LED light problem.

Bad solder joints are all too common. Even a simple device like a shoe dryer failed for me. The pair of dryers contained a total of four 10W sandblock resistors. Every lead had a bad solder joint (8 out of 8!) because both the resistor leads and the PCB were badly tarnished when soldered.

I've been using LED bulbs for almost 2 years. Almost all bought from China on Ebay. About HALF of them were junk and failed withing the first 100 hours of use. Some are still working. Some I was able to repair. This is the price I pay for being an "early adopter".

Just finished reading an "advercial" (and advertising commercial information) article in the latest (print) edition of DESIGN NEWS magazine. Gary Eskow of AVNET lays out some very informative analyses of WHY HB-LEDS fail. In summary, it all boils down to the chemical composition of the "white" light, being the result of blue emitters, the incompatibility of silastic covering materials,etc. One goes away with the feeling that ALL HB-LED devices are doomed to failure BEFORE they're rated life expectancy elapses. I would suggest that anyone w/ a inquisitive interest in this arena, to get a copy of the January 2012 edition.

What do you mean by 'pc board flow-soldered joints, just because there are so many of them'. How does that compare with solder joint count on even simplest of the PC boards? It is solderability of components; most likely, components were warehoused in humid environmentand and leads corroded to the point of not flowing the solder. The flux in wave soldermachine was probably half burned out and not replenished to save few pennies.

Lessee, when our 50k hour LED light fails at, say, 10k hours we are supposed to repair it or take it to a repair shop? How will we bill the manufacturer for this repair? How about damages for fraud if they don't pay?

I have several LED lights in our house that lost 50% brightness in 1 year and 4 that failed. CFL also have a high failure rate regardless of brand at less than 15% of rated life. Lately, Tungsten filament lamps from China have a high failure rate at less than 50% the rated life. One common denominator for all lamps. China. No quality control of any kind whatsoever. Is there any answer? USA made bulbs cost much less over their life than China junk.

In reply to Forest Mims:
I've heard reports that some of the early LED cases have interconnect reliability problems. A thermally-sensitive open could easily cause motorboating oscillation. In my bench design work, I've seen motorboating from a whole slew of mishaps, open/short servo loop components, open bypass caps, oversensitive voltage/temperature protection circuits, etc., etc.
I also notice streetlights where one string is out and another glowing abnormally bright. I think this is an open substring that is paralleled with another one, on a single constant-current drive. The good substring takes the entire current.
Doug W.

In addition to my eralier post, one of the drawbacks to LED lighting is that brightness is typically dependant on current rather than voltage. Incandescent bulbs are power devices and the manufacturers have been able to develop processes which produce consistent product. LEDs have a wide variety of current consumption and light output performance if used in a voltage controlled configuration. a 24V distribution to 7 bulb modules could result in a wide variety of light output. Some manner of current control helps in maintaining consistency.

In response to questions about LED traffic lights: We have seen a very high failure rate here in New Jersey on the multiple-small-LED types. One engineer thinks they are being over-driven. Possibly lower efficiency bulbs are being driven to meet brightness standards, or just very cheap Chinese LEDs are being used to get lowest bidder status with government agencies. The newest LED globes seem to have a single high-power LED behind a conventional lens, and so far I have not seen any of these going out.

Perhaps we should re-think that LED bulbs are used in basically a retrofit application as opposed to having a power system designed specifically for LED lighting. Put a 24VDC converter in a central location and run that to all of your LED lights, each 12W bulb (60W equivalent) consumes as much current at 24VDC as a 60W AC bulb consumes at 120V. Eliminate the power conversion at the bulb. You can also include battery backup to keep the lights lit when power is lost. Granted, this requires splitting the facility wiring and possibly adds additional wiring, and complicates matters in an existing facility, but should improve the reliability of the lighting system as a whole.

To those who have posted comments regarding leadless solder vs. legacy alloy solder, I would refer you to the most recent findings regarding the TOYOTA "acceleration" problem. It seems that in the past week or so, it has come to light by close analysis of the p.c. board of a sensor in the circuit that microscopic "hairs" have formed from one copper trace to an adjacent trace. One has to know that TOYOTA (and other auto manufacturers!) sure put MORE effort into designing critical engine control modules than has been put into these new-fangled lighting products. Silver migration, dendrites, etc. ARE a very real problem. As one poster stated, the U.S. military completely DISQUALIFIES leadless solders for use in mission-critical applications. Need one say more?

I tend to agree with Don Menning regarding CFLs, which are still around, though they maybe replaced by LEDs. Unfortunately, the herd instinct prevails here as well and lawmakers rarely understand what they make laws about. First of all, we need to look at a person's total energy consumption, including, transport, AC, cooking, heating etc. with the workplace and public energy consumption, such as shops, streetlighting etc. added to it. To travel one possibly unneccary meeting overseas pays for 20 years of lighting load in one's house. I dismantled many CFLs which failed prematurely and found that in over 60% of the cases the starting circuit went faulty, often due to some cheap capacitor. Manufacturers should replace FoC any CFL or LED light which became faulty through an electronic failure. What about the disposal of a faulty CFL containing 0,5g of mercury. How nmany people will travel 10km to an official disposal site? The world is rapidly running out of landfills. The most important feature of a CFL is rarely used, that one can lower the voltage by about 30% without increasing the current. How much harmonics are generated by electronic circuits? And please do not hold the CFL by its glass tube when you change it as it easily breaks.

Also, avoid LED's with micro fans for cooling. I have a very nice bulb that was being sold with lumens camparable to a 100W incandescent. The bulb still works great but after a little more than a year the fan is now very noisy. Eventually I'll try to break in the case to see if I can calm or replace the fan. The Philps
"Ambien" (sp?) 60W eq. are great so far. CFL's be damned!

I remember the 70s. Cars where overly complicated, and unreliable due the the government smog and CAFE rules. After a few years things inproved and now cars are better than ever. Does the repeal of the lightbulb efficiency law have us stuck in the 70s?

When you buy swag, it's junk, it looks like junk the way the components re on teh board, Philips, GE, Osram, stick with real lighitng companys. Not some swag garbage. The high cost of low prices. Philips has teh best LED out there. color, life, longetivity.

If the Shelby Electric Company's 4W carbon-filament bulb can glow for 110 years, just take your Twist and shove it.

Nobody expected massive failures over 10 years of use? I think the Cost Vs Life question is still out for bid!
These bulbs last long enough for grain-boundary migration of PbSn alloy to open up cracks and Kirkendall voids in the solder joints...If the Mfg produces a poor joint, or uses Solder with reclaimed lead, they will accelerate the failure...Every time they cut the cost per 100K units, the ice on the pond gets thinner...
Stamp them with a date like car batteries, and refund the unused portion when it fails.
$40 is alot to ask when you need 100's in commercial applications like a hospital or office building. Part of the initial Hype was reduced labor costs to replace the old bulbs!
I don't see these bulbs living up to the press releases any time soon...

as i recall whiskers (dendrites) often form in pure metals like silver or zinc. not so much in alloys. i would be surprised if a silver alloyed solder became a problem with dendrites unless a poor combination was chosen. anything hand done by humans not well trained will have a high failure rate. that will go away as processes improve. i have been burning CF lamps since their inception. out of a couple hundred lamps, i have maybe 6 that lasted the expected time. many outdoor devices had cracked glass likely due to tiny spatters of rain water through fixtures. almost all CFLs that were base up were short lived. until recently many were really bad color. i love the LED lamps except that the collimated (as opposed to defused) nature of the light emitted tends to make harsh shadows (one cost of high efficiency). that requires a little different thinking. several overlapping LED lamps can be used to simulate the defused lighting system we are used to.. if i had to choose between a $40 LED bulb that would last for 10 years .. or a $5 LED bulb that would last only 2 years .. i think i would choose the low cost alternative. yes manufacturing will choose the lowest cost way to do things and that is not always bad. seems like CFLs tend to unscrew for some reason. i suspect standard bulbs tend to micro-weld with inrush current or high temperatures reducing that problem. Now with good color LED and CFL lamps, incandescent are looking like poor color and terrible efficiency. Its certainly time to switch. making the manufacturers honor a specific warranty would tend to improve quality and life times.

Traffic light bulbs require a backup in case the main lamp fails. There are minimum brightness specs for both. Incandescents use a primary and backup filament. The LED traffic lights use series/parallel configurations. If one LED fails the series string fails which explains the visible patterns. There's 5 or 6 series strings...the number of LEDs in series is determined by the voltage across the assembly.

The problem is the voltage conversion. I switched to a 12 VDC ( actually 14v) world of the RV and marine world. three LEDs in series, requires only a small ballast resistor. I have a high efficiency switching power supply keeping a gel cell floating. Very low failure rate and lower cost lamps. Also no hum or stroboscopic effects. It even runs when the power goes out.

Incandescent bulbs, in spite of their simplicity fail rather rapidly when the filament becomes disconnected from the supporting electrode - unfortunately, this wiring problem occurs in each and every one of them not just isolated cases. Incandescent bulbs, especially those sold in North America, contain a substantial amount of hazardous materials particularly (but not exclusive to) the frits used to coat the inside of the bulb (banned in Europe) and form the seals, the 'whitners' particularly chromium compounds, tungsten itself, various materials used to extend filament life and so on. The total volume of waste as well as the total quantity of hazardous material per lumen-hour is higher for incandescent bulbs.
In any electronic product, the failure rate is proportional to maximum circuit current and volumetric power dissipation. Lowering device power consumption is always a good thing.
My LED worklight has never let me down in tight dark crawl spaces, my CFL one only occasionally, my incandescent one countless times. Only the last one has ever burnt me.
I have 15 LED lamps in my house in everyday operation - none has failed yet.

Ah yes, I've had a similar experience with some of those 'value-rated' LED lamps - very poor quality. That doesn't mean that the technology is suspect, it's really more of an issue of shoddy manufacturing.

About a 18 months ago I replaced 10 incandescent ceiling spot lights with new fluorescent substitutes due to the need for constant replacement in a somewhat inaccessible staircase. 2 units failed immediately, 2 others have become rather dim and sometimes illuminate with a red light instead of white, and 1 failed after about a year. The rest are still functioning. All of these lamps take around 2 minutes to reach a reasonable lighting level so we tend to leave them running for a longer period in the evenings.
I recently tried a LED replacement for a 220 VAC 50W halogen bulb. This failed after about three months of occasional use.
I am now trying an 8.5 watt replacement for the above-mentioned fluorescent lamps. The light is excellent and of course starts up immediately. I am hoping this lamp will last at least a couple of years because it is 10 to 15 times as expensive as the old incandescent lamps.
My experience to date indicates that the money saved on energy is spent on initial purchases of very expensive modern lamps, which on average do not last very much longer than the incandescent types that they replace. This is a relatively new industry which is still developing and so I would expect the prices will go down and the reliability will improve over the next decade. For the time being, it is very frustrating to keep replacing extremely expensive lamps.

There is a lawsuit against Lights of America by the FTC for false claims.
I had two of their CFLs catch fire.
Run away.

Eric, the problem with lead-free solder is that it will spontaneously sprout tin wiskers. These wiskers can start growing within days and can bridge across solder pads or fall off and short other circuits. It's a big enough problem that the military bans lead-free solder in all critical electronics.

I really like the transformer idea, and somebody might find it's even economical relative to the AC-DC converter scheme (which always has its own inductor anyway, plus, plus, plus). But there will have to be an artificial efficiency bar imposed that prevents simple solutions that don't self-destruct from winning in the (corrupt) "marketplace" - or government-company-store. Bet you won't see legislated lifetime performance, though - only legislated lack of options.
Meanwhile the Edison bulb enjoys the shortest, simplest, most environmentally benign supply chain and nobody has to think about it - and it works just fine inside your kitchen oven and your freezer. And no up-front cost to speak of, for you the consumer to amortize while waiting for the premature fail.

Seems the complex systems are more failure-prone than the simple systems. Who woulda thought?

You might want to consider that the full court press banning incandescent filament lamps may well be flawed from several perspectives.

First, there is the extreme simplicity of incandescent lamps relative to anything else. They have only two failure modes: filament burnout or a broken/cracked glass envelope (which leads to almost instantaneous filament burnout). And they operate properly and dependably at all ambient temperatures amenable to human activity.

However, any type of discharge lamp has multiple internal elements plus some kind of power supply. I contend that there is a small but very real failure rate associated with every such component (igniter, anode/cathode, gas fill, phosphor, inductor, capacitor, solid-state device, and each interconnection/wire bond) deployed to create what one might call a lighting system rather than simply a lamp. Lets say there are ten components per CFL, each a potential source of failure and each exhibiting 98% reliability (a made-up number). Multiply 0.98 by itself ten times and you get about 82% reliability for the lamp itself. (If you turn the CFL on and off frequently, the numbers could get much worse. Quite possibly some better mathematical analysis should be used here.)

Now, when you consider that practical, commercially-available CFLs must be mass produced at the lowest possible cost, the opportunity for increased component-level failure rates could become significant. If this happens, and should the average CFL lifetime ever drop anywhere near the all-too-short life of the incandescent, a real political problem may develop. Consumers will not be happy paying a premium for something that once set them back only 50 cents, especially if Government edicts give them no choice. (I expect they then won’t be much interested in a lifetime energy analysis discussion.)

There is also the issue of light quality. Some seeing activities may not adapt well if deprived of incandescent light. I have no figures or studies to quote you on this, but that’s the impression I get from the ongoing commentary I have read on general lighting and the importance of color rendering.

I would assume that LEDs should also be considered a lighting system with their inherent reliability and, ultimately, usefulness tied to component count and the operating environment. Cost and heat dissipation issues seem to be the limiting factors for LED deployment thus far, although this may well be transitory.

I have spent 62 in electronics and I use small transformers for my LEDs and they work fine. Switchng power supplies not only are unreliable they way they are manufactured but also produce RF pollution. Try to put AM radio next to the bulb and you will hear it .

Hi,
I have had similar results with LED & florescent screw-in bulbs. Widely varying outputs, long warm ups (florescent)and high failure rates. They are extremly susceptable to line noise/spikes. In fact, when placed in a circuit with light activated solid state switches they fail regularly.Seems as if "...the Juice ain't worth the squeeze". They need to be "snubbed", if you know what I mean.

How would the required removal of lead from solder (another environmentalist solution to a non-problem) be effecting connector failure rates.

Ha, Ha, Ha! I could have told you so! Surprised that you're surprised, really. Sounds like the same thing that's been happening with CFLs, flat panel TVs, etc. Modern corporate quality culture is 'make only good enough to sell', then foist it upon the masses by corrupt legislation. It's true now more than ever: they just don't make 'em like they use to! 'Maximum margin' is the new biz goal!

This is a pet peeve of mine with CF bulbs. They publish very long life spans, and as long as they are mounted base down and consistantly powered that are OK, If you cycle power, mount then in a horizontal fixture or base up, the power supplies overheat and fail. Cheap caps are probabily the culprit in most cases, though most consumer will just throw them in the trash.

The failure of LEDs in traffic signals has become a huge problem, Green LEDs are especially prone to failure. Often, clusters of green LEDs in traffic signals are flashing for reasons I don't yet understand. Any ideas out there? (Having written some of the first books on LEDs, I'm very much a pro-LED guy. But I'm not pro-defective LEDs.) Forrest Mims (www.forrestmims.org)

Connectors are from my experience the weakest point. A solder joint (with good leaded solder) will outlast all of us. It just must be done right. Dry contact in connectors will eventually corrode and your circuit will die. Try to use a connector or a mechanical switch on a boat. You will be lucky if it works well for five years. Get the politically correct solder joints out of the circuits and they will last.

Your experience with the poorly soldered wired in LED driver board only confirms the basics of reliability.
Incandescent bulb had only two soldered joints. Fluorescent lights needed four, plus a ballast choke; less reliability. LED looks promising but needs MANY soldered joints.

I have had a number of MR16 LED fixtures as samples from a few different manufacturers. They all seem to be copies of the same reference design, and more than one of them did not last beyond a few hours. In every case, I dissassembled and tested the LED chips and those were still good. All of them had reasonable solder joints, but the failed ones had faulty DC-DC converters. After messing about with one of them I discovered it would operate until it got warm and then shut off. So from my very limited experience the LED chips are the least of our worries, but manufacturers need to concentrate on better power supplies. And if they *claim* 50,000hrs life they should back it up with a 50,000hr warranty. If they aren't willing to do that then you should not buy them, or they should be much lower cost.

I too have converted much of my home lighting to LED's and have had 3 failures now. One was a bulb made by the manufacturer "lights of america".. the bulb was placed in an outside fixture. the leds became dimmer and dimmer until they barely glowed. this fixture is on every night but less than one year later needed replacement. 2nd was an LED rope light.. and much the same problem.. dimmer and dimmer until it's too weak to really be of any benefit. Finally last week, a "corn cob" type bulb with SMD's on it also in an outdoor fixture failed. that was only 1yr old as well. I'm beginning to question the benefit of these very expensive, dim, and faulty bulbs.

50,000-hrs is probably just a calculated MTBF based on components count and an assumption of nominal input voltage. Most likely the LED bulb will fail much sooner due to voltage spikes in the line. It is unlikely its $40-50 cost will ever be recovered.

But the average consumer doesn't have the skills or tools to repair lightbulbs. A dead LED illumination device is dead whether the LED's died, the electronics, or simply a bad connection. Dead is dead. Along those lines, I have often wondered about all the dead "pixels" seen in traffic signals (stop lights). Margery, thanks for blazing a path on the bleeding edge. Saves the rest of us the pain. As an aerospace engineer, I am disheartened to see cheap electrolytics in very expensive lightbulbs purported to last "decades". Makes on wonder whether we should have standards to make the LED drivers and LED arrays interchangeable so either could be replaced separately if needed.

Now for the birth of a lighting repair business!

Nice detective work, but most that purchase a LED bulb see the unit as a failure regardless of what aspect of the technology fails.
I have similar experience with florescent vs filement bulbs is that I would purchase the filiment bulb as a result of it's reliability for brightness independent of manufacturer. It seems every compact florescent is different for brightness after power on.
While I would expect LED bulbs to be more consistent in lumen output at turn-on, these bulbs have lower output at much higher cost. Failure due to quality of assembly process should not even be an issue that a consumer should have to deal with, but yet here it is...

I had two LED bulbs on two staircase lights that are left on 24 hours a day. One failed after about 1 and 1/2 years. The other is still burning. I should have taken the failed bulb apart to see what may have caused the failure. That being said a loose wire sounds like a potential fire hazard. I am not sold on LED bulbs as the cost, power usage and actual life are not a good product.