How much has LED efficiency improved over the past 10 years

[tom66]

biggest issue is lifetime, you find that most common LED units are not as long lasting over the CFL units, I still find the old CFL and PL lamps will last longer than LED, at least for the quality brands. the only LED I have gotten that has good lifetime in LED was a batch of Osram LED retrofits, where they  survived for 5 years running 24/7, only failing when the neutral was stolen, and they blew the internal fuses, as the built in MOV units safely shorted out. Terminal to open them, they were really well stuck together, though likely they did not have much longer left on them, as the internal capacitors had heated up to the point the orange sleeve had split and disintegrated off the capacitors, though they had not yet bulged, just run really hot.

It's strange because I've got something like 40 LED bulbs in my house and many more spares in drawers... I have replaced maybe two bulbs in total, and those were really cheap bulbs that had died because, well, £2 of LED bulb is what you get.  If you spend even a small amount more than the absolute cheapest then LED bulbs seem to last practically forever. My garage has four Philips LED tubes in it and they are going on for 8 years old, all still fine.  Meanwhile the fluros at the other end of the garage are starting to show signs of mercury exhaustion and probably want to get replaced with LEDs sooner or later.    The kitchen has 8 x GU10 spot lamps and these run pretty hot but are over 6 years old and still fine.

[Someone]

If you spend even a small amount more than the absolute cheapest then LED bulbs seem to last practically forever.

Similar experience here. Replaced an entire houses lamps with moderately priced brand name LEDs (still only $5-6 each) and they have had zero failures over more than 10 years. Another house that had them upgraded for "free" at a similar time by local carbon offset scam, numerous failing lamps within months.

It is possible that quality of supplies/drivers have gone down with increasing volume/competition, but my guess is the usual problem of consumers seeking "value"

Maximum lumen output for a given fixture size, and minimum price. Destined for failure.

Domestic lighting has been so excessive the Australian regulators introduced minimum standards (maximum W/m2: https://build.com.au/bca-lighting-restrictions). But like water "saving" shower heads there is some portion of the population who just ignores it and retrofits their wasteful alternative. When that happens for lighting fixtures the heat really becomes a problem.

[Siwastaja]

LED vs. CFL/incandescent lifetime is fundamentally different.

On incandescent and CFL, there is fundamental limit to lifetime and it's not that long; it sets maximum possible lifetime. Poor quality control / manufacturing or design errors can only reduce it, but good design cannot much increase it. And lifetime is short enough to hide manufacturing/design errors.

On LED bulbs this fundamental limit is so long that it's practically "infinite", meaning comparable to human lifetime, so the actual bulb lifetime is completely defined by small but important choices manufacturers do in design, plus manufacturing quality control.

Some CFLs for example failed from cooked capacitors somewhere close to 6000-8000 hour life but that was already close to fundamental lifetime of the technology itself so didn't matter much. But if a LED bulb cooks its capacitors at the same 6000-8000 hour mark it's worse because all other parts of the bulb could have easily lasted 50 000 hours.

Therefore LED bulb lifetime is kind of random. There probably is correlation by more expensive brands doing better bulbs on average, but sometimes even cheap bulbs can last for a long time, and even expensive ones can fail. If you replace all 30 bulbs in a house at once using same brand, same batch for every bulb, then you probably either have near zero failures, or a lot of failures if unlucky.

[Zero999]

biggest issue is lifetime, you find that most common LED units are not as long lasting over the CFL units, I still find the old CFL and PL lamps will last longer than LED, at least for the quality brands. the only LED I have gotten that has good lifetime in LED was a batch of Osram LED retrofits, where they  survived for 5 years running 24/7, only failing when the neutral was stolen, and they blew the internal fuses, as the built in MOV units safely shorted out. Terminal to open them, they were really well stuck together, though likely they did not have much longer left on them, as the internal capacitors had heated up to the point the orange sleeve had split and disintegrated off the capacitors, though they had not yet bulged, just run really hot.

It's strange because I've got something like 40 LED bulbs in my house and many more spares in drawers... I have replaced maybe two bulbs in total, and those were really cheap bulbs that had died because, well, £2 of LED bulb is what you get.  If you spend even a small amount more than the absolute cheapest then LED bulbs seem to last practically forever. My garage has four Philips LED tubes in it and they are going on for 8 years old, all still fine.  Meanwhile the fluros at the other end of the garage are starting to show signs of mercury exhaustion and probably want to get replaced with LEDs sooner or later.    The kitchen has 8 x GU10 spot lamps and these run pretty hot but are over 6 years old and still fine.

Don't forget the person you're responing to is in South Africa which is notorious for having unstable mains power. It also has a hotter climate. Both will contribute to a much shorter life for LED lamps.

[Siwastaja]

Don't forget the person you're responing to is in South Africa which is notorious for having unstable mains power. It also has a hotter climate. Both will contribute to a much shorter life for LED lamps.

Yeah, this is a shitty situation for designer because it would be entirely possible to design a more tolerable bulb which still costs only a few % more to manufacture. But it makes little sense to develop a special product just for markets like these, it ruins the economy of scales, and it also doesn't make much sense to overengineer the products for all other markets where this overengineering is not needed.

I'm not surprised bulbs are designed to survive just the specified grid voltage tolerance and no more, that is what you have even decent companies doing for cheap mass market products: products that last when used as specified but not much extra margin because that is money down the drain.

[nctnico]

biggest issue is lifetime, you find that most common LED units are not as long lasting over the CFL units, I still find the old CFL and PL lamps will last longer than LED, at least for the quality brands. the only LED I have gotten that has good lifetime in LED was a batch of Osram LED retrofits, where they  survived for 5 years running 24/7, only failing when the neutral was stolen, and they blew the internal fuses, as the built in MOV units safely shorted out. Terminal to open them, they were really well stuck together, though likely they did not have much longer left on them, as the internal capacitors had heated up to the point the orange sleeve had split and disintegrated off the capacitors, though they had not yet bulged, just run really hot.

It's strange because I've got something like 40 LED bulbs in my house and many more spares in drawers... I have replaced maybe two bulbs in total, and those were really cheap bulbs that had died because, well, £2 of LED bulb is what you get.  If you spend even a small amount more than the absolute cheapest then LED bulbs seem to last practically forever. My garage has four Philips LED tubes in it and they are going on for 8 years old, all still fine.  Meanwhile the fluros at the other end of the garage are starting to show signs of mercury exhaustion and probably want to get replaced with LEDs sooner or later.

That depends highly on the ballast used and buying good quality tubes. A good HF ballast can get 20+ years out of a Philips tube with daily use of approx. 6 hours. While cheap (inductive) ballast with old style starters make a tube break within a few years (with the typical black bands at the ends of the tube).

[coppice]

That depends highly on the ballast used and buying good quality tubes. A good HF ballast can get 20+ years out of a Philips tube with daily use of approx. 6 hours. While cheap (inductive) ballast with old style starters make a tube break within a few years (with the typical black bands at the ends of the tube).

The absurdity with most HF ballast lights is they have replaced regular replacement of the tube with frequent replacement of the poorly designed and built ballast.

[nctnico]

That depends highly on the ballast used and buying good quality tubes. A good HF ballast can get 20+ years out of a Philips tube with daily use of approx. 6 hours. While cheap (inductive) ballast with old style starters make a tube break within a few years (with the typical black bands at the ends of the tube).

The absurdity with most HF ballast lights is they have replaced regular replacement of the tube with frequent replacement of the poorly designed and built ballast.

Buying quality stuff always helps. The Philips HF ballasts for the CFLs in my office are over 25 years old and they are doing their work every day.

[David Hess]

If you spend even a small amount more than the absolute cheapest then LED bulbs seem to last practically forever.

I suspect that depends on power quality.  My experience in Missouri where we had regular lightning storms, mostly in spring, is that LED bulbs last a year or less.  LED bulbs were an order of magnitude more expensive to replace making their lower energy costs irrelevant.

The power quality might also explain why the EPA mandated ECM (electronically commutated motor) in my refrigerator evaporator regularly failed as well.  That by itself was more than doubling the operating cost of the refrigerator.

California finally acknowledged that their LED bulb mandate ended up costing consumers several times more because replacement costs were orders of magnitude higher from high failure rates.  Of course California still bans incandescent bulbs.

[coppice]

Buying quality stuff always helps. The Philips HF ballasts for the CFLs in my office are over 25 years old and they are doing their work every day.

The problem is knowing up front what will turn out to be good. You now know a 25 year old Philips seems pretty good. You know nothing about other makes, of a more recent Philips products. Price is only weakly correlated with quality, and I have yet to see a product like this with a long warranty.

[SeanB]

Yes quality matters, cheap Ya Ming lamps I changed pretty much in under a year, but the Osram and Phillips that were made in various parts of the EU you pretty much got 5 years out of them, even on 24/7/365 in some cases for PL9 lamps running base down with CFL adaptors that are a magnetic ballast. I have the remains of a box of Philips Tornado 25W lamps, bought in around 2000, that still consistently do 4 years plus in operation as night lighting, despite the warranty on them being close to a quarter century expired. Some of the 1970's era Thorn blackends that were on instant start ballasts you would do the 5 year ballast replacement, clean the tar out of the fitting trough, and put the new Tridonic ballast in, and put those old lamps back in service, for 10 hours a day 5 days a week. after the instant ballasts went up in price to more than a new fitting I simply replaced 2 fittings, and took the old ones off, replace the instant ballast with 2 40W magnetic ones, and put in 2 starters, and next time they went in as replacements. had to buy Osram tubes by the box, always making sure i got them from EU, the US made GE stuff just did not last at all, only the Bucyrus plant 3ft tubes did last, the 4ft Mexico ones were all junk in short order. but not much call for the 3ft pink tubes elsewhere, so i still have one or two left. Last tube i changed the tube was made in France. Got about a dozen left.

If you are doing a retrofit to LED best is to leave the ballast in place, it makes a really good mains noise choke.

[johansen]

LED's for general lighting, how do newer products compare to a decade ago in terms of brightness and power requirements, lumens per what?

In practical terms, about double.

Its all diminishing returns at this point. For 40$ you can buy an 8000 lumen overhead lamp and you can feel the heat radiating out of it as if its a 300 watt halogen lamp from the same distance away

How much of that heat is from the light hitting your hand, being absorbed and warming it up?

That comes out to be a 300 watt halogen lamp generating 7500 lumens and an 80 watt LED bulb generating 8000 lumens, so hopefully most of it.

Considering the led housing probably blocks IR and you feel the heat instantly..

[nctnico]

Buying quality stuff always helps. The Philips HF ballasts for the CFLs in my office are over 25 years old and they are doing their work every day.

The problem is knowing up front what will turn out to be good.

It is not a problem. Buy gear which is to be used in commercial buildings. People are not going to be happy when ballasts need to be replaced in a warehouse with an 8 meter high ceiling after 4 years. You likely won't find that in a shop around the corner where they only sell consumer crap but online it is easy to find.

[tooki]

every white LED is really a blue LED plus phosphors to convert some of the blue light to other wavelengths.

Don't the better ones use violet LEDs? (405 nm.)

Not as far as I can tell.

There have been persistent claims of UV-based white LEDs for years, but despite fairly extensive searches to try and find such a thing in the wild, I’ve only ever found blue-based ones. 405nm LEDs are much more expensive and less efficient than 450nm, so the entire white LED industry (especially the phosphors for it) have been designed around 450nm.

[tooki]

   Interesting realization:
   That term, 'LED Light' has migrated from being, strictly a single LED component, to a bit of slang reference, to a full household lamp type of image,  there next to your reading chair.

I mean… sorta?

I would never have called a non-lighting LED product an “LED lamp”, as I’d call it an indicator or similar.

   But,  for components,  I know that blue color components were a challenge,  needing slightly higher operating voltage.  Many BLUE LEDs are built by having an imbedded LED emitting onto some chemical or phosphorus light re-emitter.

The development of the blue LED is a fascinating story.

But as others have said, your memory/understanding of it is inaccurate: there have never been blue LEDs that converted red, yellow, green, or infrared light into blue light using phosphors. It took a while to develop practical, affordable, and efficient blue LEDs. But once we did, it was realized that the blue could be made more efficient than the others, and the higher-energy photons of the shorter wavelength makes them more practical to use as the light source for exciting phosphors. That, combined with the fact that white light made by RGB LEDs has spectral peaks that make it terrible for lighting, led to the development of the broad-spectrum phosphors we use today. And the efficiency and sheer volume of blue LEDs made today means that it’s even becoming sensible to make phosphor-converted LEDs of colors that can be produced by a bare LED — like red and green.

[tom66]

It is indeed impractical to convert red or green light into blue light.  Almost every type of phosphor does downconversion, so blue light becomes yellow, for instance, or UV light stimulates red, green and blue phosphors (plasma displays, fluroscent lamps).  I believe such "anti-phosphors" do exist but they are considerably less efficient and rely on less direct principles, since they need to consume e.g. three photons of a lower energy to emit two photons of a higher energy.  (Anti-Stokes shift.)

[RJSV]

   I meant that the term of a 'lamp' being a common word loosely associated with the common house lamp, often standing next to an easy chair, or couch.
Ironically,  that doesn't include some typical ceiling or dining room 'light'...(See how that word shifted).
   Most of this is expressed as 'common' language,  perhaps not exact,  especially as it may seem arbitrary,  even.

   For example of the inconsistency,  I would term a car headlight 'bulb' as a 'lamp' for the car.  Although, again, the car example seems inconsistent,...until you realize it's casual, not directly scientific,  like many other human used words.

   Thanks for providing those details,  on the phosphorus and helping understand some of the physics.  I really want, now, to go read about the efforts in developing the BLUE led.

[tooki]

   Thanks for providing those details,  on the phosphorus and helping understand some of the physics.  I really want, now, to go read about the efforts in developing the BLUE led.

Here’s some links to get you started:
https://www.nobelprize.org/uploads/2018/06/popular-physicsprize2014-1.pdf


https://spectrum.ieee.org/rcas-forgotten-work-on-the-blue-led
https://spectrum.ieee.org/blue-leds-changed-our-world

And wikipedia has a good overview: https://en.wikipedia.org/wiki/Light-emitting_diode#Blue_LED (and the following section about white LEDs).

[Marco]

There have been persistent claims of UV-based white LEDs for years, but despite fairly extensive searches to try and find such a thing in the wild, I’ve only ever found blue-based ones. 405nm LEDs are much more expensive and less efficient than 450nm, so the entire white LED industry (especially the phosphors for it) have been designed around 450nm.

I guess I should have said "the ridiculously high end niche LEDs" rather than better.

Yujileds have violet emitter LEDs, their VTC range. Seoul Semiconductor has Sunlike ... ~1 buck per Watt for LED strips and COBs.

[tooki]

There have been persistent claims of UV-based white LEDs for years, but despite fairly extensive searches to try and find such a thing in the wild, I’ve only ever found blue-based ones. 405nm LEDs are much more expensive and less efficient than 450nm, so the entire white LED industry (especially the phosphors for it) have been designed around 450nm.

I guess I should have said "the ridiculously high end niche LEDs" rather than better.

Yujileds have violet emitter LEDs, their VTC range. Seoul Semiconductor has Sunlike ... ~1 buck per Watt for LED strips and COBs.

Oh awesome, thank you!

Their site also has this interesting page: https://store.yujiintl.com/blogs/all-about-led-technology/led-phosphors


What is interesting is that the claims of UV-based white LEDs go back a long time — long before we began to use LEDs for lighting. So Yuji using them for extra-high-CRI LEDs makes total sense now, but wouldn’t have been on anyone’s radar back when the first white LEDs came out.

[ejeffrey]

every white LED is really a blue LED plus phosphors to convert some of the blue light to other wavelengths.

Don't the better ones use violet LEDs? (405 nm.)

Not as far as I can tell.

There have been persistent claims of UV-based white LEDs for years, but despite fairly extensive searches to try and find such a thing in the wild, I’ve only ever found blue-based ones. 405nm LEDs are much more expensive and less efficient than 450nm, so the entire white LED industry (especially the phosphors for it) have been designed around 450nm.

It's also easily verified.  Better lamps will have a data sheet with a special response plot.

I checked several manufacturers of high CRI bulbs and most of them showed a blue peak from the emitters.  However Soraa appears to have both violet and blue emitters:

https://www.soraa.com/assets/cloud/product_specs/soraa-vivid-warm-dim-mr16-gu5-3-12v/SS-SM16-7W-NA2_WARM_DIM.pdf

That's a color temperature changing bulb but their fixed temperature bulbs also have a violet peak.

[Marco]

Googling a bit more this seems the manufacturers with violet emitters : Soraa, Nichia Optisolis, Seoul Semiconductor Sunlike and Yuji VTC/AP/APS. Lumen/Watt seems to top out at 140 lm/W for the bare LEDs. The Soraa integrated lamp efficiency are much lower, but I'm not sure that's because of the integration into spotlights or the bare LEDs (they don't seem to sell bare LEDs any more).