It's worth noting that the often quoted "50,000 hour" LED bulb lifetime largely comes from the data sheets for the LEDs themselves, it's the time it takes for the light output to drop by 50% - how long the rest of the lightbulb lasts (or LEDs if the enclosure allows them to run too hot) may or may not be as easily quantifyable
I need to replace some LEDs twice a year in our apartment, whereas formerly bulbs lasted at least several years. I have tested different vendors, including expensive ones, but all last equally short times.
Light fixtures were designed for incandescent bulbs. Even with advent of fluorescent bulbs and now LED based bulbs little has changed with how light fixtures are made. Old style tungsten bulbs had no problem with high temperature, fixtures trapped heat - hot air rises up and is kept inside. Components of LED bulbs like electrolytic capacitors have known life expectancy based on ambient temperature, even high temperature ones (105degC series) degrade fast when in such conditions.
Another problem is that manufacturers overdrive components. To make bulbs cheaper they just use few more mA of current, that makes LEDs run hotter, more smoothing is needed and caps get hammered by switch mode power supplies.
Without changing light fixtures to be open, allowing circulation of air nothing really can be done for standard e27/bayonet bulbs. I have personally experienced this and had clearly seen the huge difference in LED bulb lifetime between light fixture - glass globe with hole in the bottom (no air vents at the top) and another one that was just a bowl with open top. Never had to change a bulb in second one versus 5-6 changes in globe one.
Heat is not the only problem, I had to replace once a year even 5 Wt LED lamps which are barely warm. I guess LED drivers (one cannot connect LED directly to 230V) are unreliable and fail long before the LED itself.
I totally agree that if run them cool enough they last forever.
but I had some bridgelux 10W that I just threw in some copper tubes on an aluminum slug with a minor affordance for cooling (some slots cut in the Al to promote convection). These ran at 80C and lasted for >5 yr at at least 80% brightness. and still the primary failure mode was the little meanwell dc/dc converters playing thermal runaway. and that whole assembly cost $10
it just has to be penny pinching at the end of the day.
Yes, it’s because they drive the heck out of the LEDs to get their maximum rated light output. This produces excessive heat, shortening the lifespan of everything inside the case, including the very cheap electronics they use to rectify mains AC.
In Dubai they actually have efficient and long-lasting LED lights [1]. These exist in order to satisfy strict government regulation there. They’re not sold anywhere else because they’re more expensive to make (more LEDs per bulb running at lower output).
I would gladly pay more for longer lasting but environmentally sustainable ones! Another case where government intervention seems necessary and there is no real "corrective" force with only capital oriented companies.
Capital oriented companies can lead to the correct outcome if the consumer has enough information to distinguish the superior product. One solution to this problem that capitalism has found is UL Listing (Underwriters Laboratories.) Of course you could say they are just outsourcing what should be done by the government. It comes down to who do you trust to not fall into corruption.
So just import them from Dubai? Or you could start a company and start selling long lasting lightbulbs yourself. Its not like you need a team of super scientists to design a LED board with a bigger heat sink. Its weird how so many peoples' first thought is always to have the government micro manage everything through legislation.
“Just rely on another government’s strict regulations and illegally create a business around it so you can get a few durable items oh and stop complaining about government regulation.”
It's not only heat from LEDs but also from power unit that LEDs uses. They transform electricity from 220V sine wave (AC) here to 5V DC.
In 90% of cases where LED need replacement problem happens in this transformer unit. In rest 10% it's on LED fault - but it's also problem because they are series connected and when one is gone whole circuit stop. If they are parallel connected such issue won't happen.
The LEDs are not driven at 5V - that would be one of the worst voltages to be driven. The LED bulbs are driven approx at 3.2V x number of LEDs. (some LED chips may have 3-4LEDs in series as well)
You can buy them in Australia, or at least the bulbs sold here by philips as their "UltraEfficient" range look the same and match the specs of their Dubai bulbs marketing page.
They don't appear to be available in the US though, that's true.
I think I'll get a couple to try for those places where I think "60w incandescent equivalent" is sufficient, although cynical experience tells me to expect 40w at most - but for low single digit wattage power draw I'll accept that in some places.
Mostly I like to see though, so I'll keep buying their 150w equivalent bulbs.
I find it completely ridiculous you can’t buy these anywhere else or it is very difficult. Lends credence to the theory that we are in a new led bulb light conspiracy. I don’t think I’m the only one that would like to buy an over engineered LED bulb that will actually last as long as it should.
This has been discussed multiple times here, a sibling comment is right
- LEDs light bulbs are overdriven to show better numbers on the box
- they are designed for universal voltage range: AC 90-264V
- which leads to extra overheating in the US/Canada and passive component failure, due to high current on the primary
- in 230V range, the common failures are the LEDs, themselves (black dot of death), due to lack of decent heat dissipation (and generally overdriven)
- as a general guidance - buy the heaviest bulbs at a specific lumens, they may have Al heat sinks, better heat dissipation, better passives
Overall the LEDs cannot retrofit well in the existing sockets, e.g. E27, esp E14, for incandescent light bulb. The LEDs need to have enough room to dissipate heat and run AC/DC converter with decent passives. OTOH, near custom ceiling LEDs driven at decent currents (with no terrible drivers) would last many years - but they require some electronics background to repair.
> they are designed for universal voltage range: AC 90-264V
I have never thought of that, since it's not a problem on its own. But yeah, this makes the first point way worse than it by itself. It's enough to explain all the geographic difference in quality I have noticed.
> as a general guidance - buy the heaviest bulbs at a specific lumens
But this does not work very well. You can easily get a 20 times difference on the power supply waste between bulbs. If the one that is 20 times worse is 10 times heavier, it will still have a much shorter life time.
The most noticeable difference is still the metal heatsink. LED drivers would be non-isolated ones, so only a single coil (not a transformer). The other part of the weight difference would be capacitors (hopefully larger ones), and (hopefully) input filtering. Some LED bulbs have terrible input filtering (or not at all) and can easily die due to transients (incl. inductive motor noise/kickback) - input filtering normally would be a capacitor + coil.
One thing we noticed with other brands was that not only many fail after a year or so, the ones that still work have significantly redeuced efficiency.
5 LED Downlights in the kitchen, 3 in the upstairs bathroom, 2 in the downstairs bathroom all installed by an electrician 11 years ago and still going fine.
LED bayonet fitting bulbs in the regular light fittings in the rest of the house seem to only get a few years though.
Exactly the same experience here. I think the downlights are designed with cooling in mind, whereas the bayonet style are just shoehorned into that form factor
And of course the enclosures have entered a race to the bottom, such that they're just as disposable as they always were. Which sucks as the marketing remains about the longevity. So now ppl complain about the technology, not the implementation.
We put 18 LED flood lamp bulbs in as part of a renovation 12 years ago. They get probably 3-4000 hours of use a year, so We've replaced maybe... four or so? So ~190k average hours per failure or thereabouts.
I think it's reasonable to be skeptical about numbers. But to imply that modern LED bulbs are unreliable is IMHO absolutely ridiculous.
I think the 12 years ago variable is important here as well as what I’m assuming are hardwired drivers for the LEDs since they’d be better designed for heat.
Good news is we can solve the problem if we ban incandescents and mandate light fixture replacement renovations.
So... you think the only LEDs that are unreliable are the ones so new that we don't have data showing they're unreliable? That seems sorta laughably circular to me. How about we just get some data showing reliability problems instead?
The Phoebus cartel lowered the average life of bulbs from 2500hrs to 1000hrs. In 1949 a US court found that GE violated the Sherman Antitrust Act in part due to it's actions as part of that cartel. And from the article on the topic on Wikipedia:
The cartel tested their bulbs and fined manufacturers for bulbs that lasted more than 1,000 hours. A 1929 table listed the amount of Swiss francs paid that depended on the exceeding hours of lifetime.[11] Anton Philips, head of Philips, said to another cartel executive, "After the very strenuous efforts we made to emerge from a period of long life lamps, it is of the greatest importance that we do not sink back into the same mire by paying no attention to voltages and supplying lamps that will have a very prolonged life."[6]
So, yeah that looks exactly like the evil actions of a cartel unless maybe you're claiming this isn't factual. If so, I'd be very interested in your source of information.
One data point from the above-linked Technology Connections video: decades after the end of the Phoebus cartel, the lifespan of most lightbulbs remained at 1,000 hours. You could get longer-life bulbs, but they consumed more power for a given amount of light, which made for a more expensive total cost to the consumer.
If the goal of the fine was to make sure power efficiency was good, they should have made a fine for having bad power efficiency, not a fine for lasting too long.
At the time there was one way to make the usual commercial lightbulb and the only one the group was regulating. I don't think it would make any difference even if the threshold was lazy / technically wrong.
I would think there would be a lot of variables that could be tweaked, such as shape, composition of the filament, gas, glass, filament support parts, thickness of the filament, shape of the filament, temperature that the filament was treated at.
The fine was in industry speak because it was only for industry. This is a chronic problem where commoners read stuff they don't actually understand, but think they do, and then completely freak out about it.
"Industry speak" sounds like it's just a debate about terminology. But I'm not simply suggesting changing the terminology, I'm suggesting actually measuring something else.
The bulbs emit more lumens per watt when the filament runs hotter, but it lowers its expected life span. Using a bulb with a longer life span wouldn't necessarily save you money, as you would need to use more electricity to get the same amount of light.
Efficiency is lower in a colder filament because more of the radiation is in the infrared for a given wattage. So you get more heat for an equal amount of visible light. Here is an example with plausible (but not real) numbers.
All of the electricity turns to heat in any case. The problem is the nature of incandescence. Incandescent filaments are black-body radiators. The spectrum of black body radiation changes with temperature[1], so filaments that run hotter emit a larger proportion of their energy in the visible spectrum.
I would guess both through soft power from the heads of its members and their network as well as with the threat of exclusion from the cartel and being made an enemy of it. Cartels exist to crush competition after all and anyone who isn’t in the cartel is competition.
>I would guess both through soft power from the heads of its members and their network as well as with the threat of exclusion from the cartel and being made an enemy of it.
What does that translate into in practical terms for a lightbulb manufacturer?
Pick any anti-competitive tactic. For lightbulbs I’d say dumping cheap product priced well below cost in their sales territory would be an obvious option.
The thing is the the lifespan of bulbs is a tradeoff between brightness and lifespan. You make a 2000 hr bulb and it's half as a dim as a 1000 hr bulb, and so you're going to need 2 bulbs instead. Even today, long after the cartel is gone, there is no magic technology to make longer-lasting incandescent bulbs than they conspired to enforce back then.
The cartel was more of a marketing self-regulation conspiracy so that one manufacturer can't make claims that their bulbs are more cost-efficient when they're just plain dimmer.
> You make a 2000 hr bulb and it's half as a dim as a 1000 hr bulb, and so you're going to need 2 bulbs instead.
Dubious. Most people would not rewire a room in their house to install a new ceiling light because a new bulb was half as bright as the old, because the new bulb would probably still be adequately bright and the new light level is something they could adjust to (much easier than rewiring their ceiling.)
In other words, there isn't a linear relationship between bulb brightness and the number of bulbs people use. The number of bulbs people use is relatively unresponsive to bulb brightness because bulbs need fixtures and changing fixtures is a pain in the ass.
The problem is that nobody checked lumen/watt, or had the knowledge to make an educated choice based on it.
Imagine you sold a 2000 hr bulb that has two filaments in it, in parallel. It would fit the same fixture and make the same light as the equivalent 1000 hr bulb, but it would consume more because it runs at lower current and temperature.
You would make more money: almost all of the cost is the same as for the 1-filament version, but you could sell it for a higher price because it lasts twice as long. The buyer would pay for it slowly in the electricity bill, and ultimately be screwed.
There might be no connection to how many bulbs people use, but how powerful they are. If you make bulbs less efficient, people will uses higher-power bulbs. Like 100 instead of 60W. And if the dimmer bulb is good enough, why not use a 40W bulb in the first place?
It was a conspiracy made between evil people, that got the support of some non-evil ones and general acceptance because it happened to improve the life of most people.
If that matches the meaning of "evil conspiracy" isn't something that people will settle on.
There is an excellent video about Lightbulb Cartel by Technology Connections, highly recommended if you want an engineers perspective, with no 'great conspiracy' hype.
I replaced some overhead halogens recessed lights with some retrofit LEDs a decade back. Bought it on Amazon from some Chinese company. They have huge metal case/heat sinks so it seems to dissipate heat well and nothing has failed so far. Can't find them on Amazon anymore but I bought a few extra just is case.
Incandescent bulbs used in microwave ovens easily outlast the ovens themselves. When a 20 year old microwave oven dies for one reason or another, it's not unusual for the light to still work.
We need to decide on a new design for "LED bulbs" that maximizes cooling, instead of insisting on sticking to a design that was optimized for a completely different bulb type (vacuum bulbs) where the electronics have to be stuck in an unreasonably small and difficult to cool base which is then mounted inside a cavity with no airflow.
Here in Japan, ceiling lights have a standardized socket[0] so ceiling lamp fittings are easily user-replaceable without people being stuck with old bulb fittings, and so ceiling LED lamps are not fitted as bulbs but typically a large flat metal plate with the LEDs mounted to it[1], which results in fantastic cooling and good lifespan.
Edit: Our apartment was built in 1998, at the height of "CCFLs are the future", and the hallway lights are all recessed "G24q" bulbs, which I had never heard of before.
There were supposed to be the CCFL-native replacement for Edison screw bulb sockets, where instead of having to build the CCFL ballast into each bulb, it was external in the ceiling socket, increasing longevity and decreasing the cost of the bulbs themselves.
We have yet to need to replace any of them, a much longer lifespan than I've ever had from one of those CCFLs with the ballast embedded in the base of the Edison bulb. With a less-compromised ballast they also work much better (light quicker, less affected by temperature)
I've seen a few failures, most of which seem to involve electronics.
A set of reflector spots (replacing 45W halogen bulbs) all in recessed ceiling fixutres in one specific location within the house failed within a few months of one another.
Another spot in the kitchen, in an exposed can, failed within a year or so. It was located directly over the sink.
I suspect moisture / condensation as issues in both, as the ceiling cans have minimal insulation between them and the roof, and the kitchen fitting would be exposed to steam / vapour coming off the sink.
Other than that, a couple of stick-style closet bulbs have failed, also apparently electronics. And one fixture in the garage which seems to have been physically struck.
That's over roughly a decade's experience and roughly 30 bulbs / fixtures. Far better than incandescent or even CF bulbs. Though still not quite meeting expectations.
I have a bulb in my bathroom that I have had to replace every year or two. Meanwhile, in other rooms that I have renovated and I put good quality high-CRI LED downlights in, they're all still going strong after seven years.
I think the biggest difference is that these downlights have separate driver modules and a short wire to the lamp itself, which must make the thermal aspect a lot easier to manage. This also puts the electronics up out of the condensation.
The bulb in a regular socket in the bathroom on the other hand has the driver electronics there in the bulb, being heated by the lamps, so they're getting a lot hotter, and also it's there in the steam from the shower etc.
This is the downside of the race to the bottom caused by globalization. Commodity electronics are almost exclusively made in China by mystery brands that aren't around long enough to care about reputational damage from unreliable products.
Adding more LEDs to the design allows them to be underdriven, increasing efficiency and longevity. The driver board and LEDs run cooler that way. Less heat equals less stress on electrolytic capacitors, a common failure point.
There’s only a few bulbs I’m aware of that follow this pattern. The Philips Dubai Lamp takes this to an extreme. A 60W equivalent uses only 3W. Other 60W eq. LEDs tend to be ~3x that. All that heat has to go somewhere.
I just picked up a 40W eq. bulb from IKEA that has double the normal amount of LED strands- they’re unusually long, too. It also uses about half the power of the LED bulb it replaced. I expect it to last quite a long time.
Someone thinks your comment is worthy of down voting, but you're absolutely correct.
Heat is a primary killer of many electronic things, including LED bulbs and their constituent parts.
And LEDs -- the diodes themselves -- do tend to become more efficient (in terms of lumens produced per Watt of input) when not pushed to their extreme operational limits.
And it is definitely possible to create a longer-lasting, more-efficient LED bulb.
It's absolutely trivial to do this, even: To start, just add more LEDs, reduce their individual RMS current, improve heat sinking and dissipation, and use better capacitors.
But it does cost more to do these things, and regular consumer products are all built down to a price.
> And LEDs -- the diodes themselves -- do tend to become more efficient (in terms of lumens produced per Watt of input) when not pushed to their extreme operational limits.
This isn’t some weird surprising tendency. The forward voltage (the amount of energy needed to shove each electron through the LED) increases with current, approximately according to the Shockley equation. The external quantum efficiency (photons out per electron in) is no more than 1, and tends to get lower at higher current [0]. The wavelength of the emitted photons is pretty much constant. So you get the best efficiency when the current, and hence voltage, is low.
Incandescent bulbs, in contrast, don’t have a meaningful quantum efficiency, and they produce a more useful spectrum as they get hotter.
These aren’t the same as the Ultra Definition series they mentioned, unfortunately. They lack the dim to warm feature that makes those desirable. RAB is the only manufacturer I’m aware of that makes 3000k dim to warm A19 bulbs for the US market.
The RAB bulbs have a weird stair step warm-then-cool when turning on and off. They also have a huge non-linearity when dimming. That makes the Philips version more desirable, but they’re 2700k.
> It also uses about half the power of the LED bulb it replaced.
This may or may not be the only thing in play for that much of a difference, but my absolute favorite thing about LEDs is that their power consumption vs brightness is not 1:1, so taking an LED to 50% brightness will use <50% power. (So then doubling the number of LEDs means you will always get more brightness or less power or I suppose both if you split the difference.)
I would love to know why this is apparently being downvoted, as far as I know it's true. A lot of cheap LED bulbs use few LEDs being grilled at relatively high current and definitely get hot enough to impact the lifespan of the LEDs. Am I missing something?
Early ones for sure, but I've noticed steady improvements. And I've only ever had problems with fairly inexpensive bulbs. I have a house full of Cree retrofit cans which have been 100% reliable for years now. My guess is that is partly because Cree is higher quality, and partly because they are not bulbs, and so the heat management is way better.
Some of the early ones were great too. I have a bunch of L Prize bulbs [1], some of which have been on for significant duty cycles since their production in 2011, and have had zero failures.
Is Cree considered higher quality? I had one of their CFLs try to burn my house down about 10 years ago. I smelled burning plastic from the adjacent room and found a bulb in an open fixture with a completely blackened base.
I haven't bought a Cree product since then no matter how cheap Costco marks them.
They were, for sure. However, in 2019 they were sold to another company and I do not know what may have changed since. All of my Cree lights pre-date that.
> I had one of their CFLs
Cree was founded entirely on the premise of creating LED lighting, I am pretty sure they have never done CFLs.
When the incandescent ban went into effect last summer I bought a Phillips Hue system, since it was going to be harder to get regular bulbs. I could tell the difference, but was willing to just deal with it. Then I got 4 ocular migraines in 2, maybe 3 days. I shipped them back, bought a bunch of regular bulbs from a local hardware store with stock, and really don't know what I'm going to do when my stash of incandescents is done. It seems like Waveform Lighting is recommended for high R9 values and flicker free, but they're certainly not cheap.
Don't get me started on blinding LED headlights. It's so hard to see on the road at night these days.
You're not alone with these LEDs causing migraines.
See https://FlickerSense.org for a PhD scientists personal account, and, https://LEDStrain.org forum full of people that get migraines, dizziness, eyestrain and headaches from LED lights and other forms of flicker like temporal dithering.
Had similar issues with CFLs, when those were A Thing. “Oh they last sooooo long! Don’t worry, they’re expensive but they’re actually cheaper than incandescent because they last so long and use so little power!”
Shitty light, and they consistently averaged a shorter lifespan than cheap incandescents across two houses. Nope. Done with that tech forever.
I still have CFLs in my bathroom. The plastic in the plug has faded from white to yellow. Yet they still work fine, although it takes them 3-5 seconds to get to full brightness.
Yep, same. I replace CFLs and now LEDs about as often as I used to replace incandescents. On the upside they put out less heat. On the downside they cost 10X or more what an incandescent did, making the payoff in electrical energy savings suspect at best, and the quality of the light is worse (subjective).
As a consumer, I don't give a shit - it's false advertising. I have taken to writing the installation date on the bulb body in Sharpie; I suspect I won't be able to force manufacturers to honor their claimed lifetime, but perhaps I'll get a free box of bulbs after a blog post about it goes viral.
[...] You've heard of the Kirghiz Light? well that's the ass end of a firefly compared to what we're gonna—oh, you haven't heard of the—oh, well, too bad. [...]
