Clive has also made a few more videos since then, showing how reducing the power at which LEDs are run at can dramatically increase their efficiency - they put out less light, but at more lumens/watt.
> Overall, it is interesting that a nice design came from what amounts to government regulation.
I guess the author isn't familiar with what caused dishwashers to become more efficient[0]. In general I don't see how anything except government regulation can force these improvements whenever consumers find themselves in an "any color as long as it's black" kind of position regarding market choices, and this looks like it is one of those situations.
EDIT: Also, before anyone thinks I'm blindly pro-government regulation, Philips got rich by exploiting the fact that the Netherlands did not enforce patent laws for a large part of the 19th century. This let them combine innovations that other competitors didn't dare to use due to fear of being sued by Edison, resulting in them having the best light bulbs available at the time (the fact that many competitors that Edison sued out of business happily joined Philips probably helped too). So it's all context-dependent.
>EDIT: Also, before anyone thinks I'm blindly pro-government regulation
The fact that you feel the need to add this edit makes me think about how the conservative framing of almost every issue is assumed to be true by both sides. Nobody would make an edit like this about being blindly anti-government regulation, because that's the default position. Government regulation is not bad by default and your overall point is excellent.
The edit was simply because I know on what kind of discussion forum I am right now and in which direction it leans. While I agree that I shouldn't have had to make it, I just wanted to be ahead of the likely fallacies that would be thrown against me.
You sincerely think that government can drive efficiency? Really? Don't you see that such regulations harm other players that might offer better thing in the market in the future?
This is lobby, and it's bad for economy and the environment.
What do you mean "think"? We have two real-world data points right here showing cases where government regulation forced manufacturers to improve the quality of their products, and where if left to their own devices their own conflicts of interest would have gotten in the way of doing so. Namely, the inefficiency of dishwashers being a fully externalized cost, and planned obsolescence of light-bulbs ensuring that customers come back.
Do you have any concrete arguments for why these two cases of successful government intervention do not hold up? Do you have any argument to suggest that the situations they describe are exceptional instead of common and all over the marketplace?
That "might" is doing a lot of work. We're stuck with many mediocre options in many markets because they're locally maximal outcomes but not maximal outcomes for society, which is exactly why we need a strong government.
The power factor is 0.5 which is appalling - can't really get lower than 0.5 with capacitive load. Industrial clients do pay for the VA (apparent) than W (actual) power. But even then most urban EU power network is dominated by capacitive loads as the inductive ones tend to use 3-phase 400v (which is balanced)
Most of the constant current driven LED would have similar characteristics when they are not overdriven. 25KHz+ PMW driven LED would have similar features as well. It's not hard to make LEDs that actually last and have high lumen/watt ratio with off the shelf components.
Overall the main culprit is the LED installation mechanism, e.g. E27 - there is just not enough room to make efficient light out of that. Near ceiling lights do look better (at least to me) and they can have both not overdriven LEDs (say 90mA per 5730), high (over 87%) power factor, decent passive elements... and they are repairable with some electronics background.
My living room is lit through 30(3W 240V) LEDs. I installed them about 2 years ago and nearly half of them are gone. The lifetime of these is supposed to be 20K hours. Obviously the lights are not switched on all the time, more like 4 hours on average. That's less than 3K hours of usage with about %50 fatality rate.
Out of curiosity, I teared down a few failed ones and they were all gone because one of the LEDs on the LED array blew up. Here is a picture of a failed one: https://imgur.com/aOg7D7S
I know someone who retired from working in the LED industry in Taiwan. She said the manufacturers were never going to make enough money from lights which last for ten or twenty years. They are 100% degrading the product so that they need to be replaced much more often. They are all colluding on this. You basically can’t buy the true LED light bulb. It’s a shame.
We have this, another example is Luxotica (https://youtu.be/yvTWjWVY9Vo). I'm wondering if someone keeps track of all industries that need to be disrupted.
I read an interview with the Luxottica founder - he's an absolute bastard, and proud of it, like a Bond villain. We'll have to wait until he dies to break that monopoly up.
When you visit an eyeglass store in the US, it's likely all Luxottica frames, especially the chains they own:
I bought my house about 6 years ago, and since then I've converted the whole house (about 100 bulbs) to LED, using bulbs from maybe 5 different manufactuers.
One of the mfg bulbs go bad. I think I'm at about a 50% failure rate with them. Something like 10 out of 20 bulbs. None of the other mfg bulbs have gone bad. Basically I bought bulbs that were on sale. The "bad batch", IIRC is when average 60W led bulbs were in the USD$2.5 range and I got them for about $1.0, so about 6 years ago. I bought some other bulbs at a similar price point/date, and they are still all fine.
So, it could just be you got unlucky with a crappy mfg. Try some different bulbs.
wow 3W, with just 6x 2835 (so each is 0.5W). The thing has been driven through the roof. Even at 250mW I'd consider them overdriven - it's surprising they survived that long.
Also only the PCB is aluminium, not extra heat sinking. Is this GU10 socket?
Yes, my experience with LED bulbs from a swath of manufacturers is that they have only slightly better longevity than the old incandescent bulbs in reality, no matter what the theoretical lifetime claims are.
This is at least better than the CFL situation. At least in my house they had an absolutely horrific failure rate. The whole shift to CFLs had to have been an environmental disaster.
I still buy LEDs because I like the light from them, but man, I'm basically paying 10x the price (I like the 100-watt equivalent bulbs) for the same quality bulb. It's depressing.
That would be the case only if the driver is exceptionally bad, heat kills them pretty much (which would be continuous use). Any usable driver should limit the inrush current.
@mrtksn, pretty much all LEDs you buy in the shop are heavily overdriven to get 1) high power, 2) shorten their life span, so you have to replace them, i.e. the planned obsolesce.
It's exceptionally rare to see e27/e14 designs that would be half-decent. The "dubai ones" are an example of not-overdriven leds with okayish electronics but they still are subpar.
---
(Edit) A simple solution for DYI, are LED strips - they are not super efficient as they have current limiting resistors (i.e. heat) but they can be dimmed easily (PWM) to last for long long time. Also they can be used in a lot of creative ways.
Indeed - can you keep all the receipts? I have had LEDs replaced in warranty - but it's a massive hassle. They fail one by one, so less with than 2 quid each, it is mostly driving for a single lamp (and finding the correct receipt).
Replacing led bulbs under warranty is sometimes too complex and costly versus buying new ones: you need the receipt, you need to go to the store, they will exchange it with one that is as bad as the original and it will burn again.
Why would that be? The warranty is against defects in the product(design or manufacturing) that prevents you from using it for the expected lifetime. Limiting it to way shorter period than the lifetime doesn't feel right.
I didn't investigate too much when I was hit by this, but apparently it's because these kinds of goods suffer from wear. Now obviously, it doesn't make much sense to allow a manufacturer to advertise 10k hours for a lightbulb while only offering 6 months of warranty, but here we are.
For a laptop battery I might have put in more effort to dispute with the manufacturer, but for two ~5€ lightbulbs I just couldn't be arsed.
You can have plenty of LEDs and a reasonable switching PSU in an E27 lamp, it's just harder to cool them with the typical design. Most LED lamps use ridiculously few chips, you could fit multiple times more even on the one-PCB designs, not to mention "corn cob" arrangements or those with LED filaments. That's not to say E27 is an ideal form factor for LEDs, I'm hoping we'll see more and more homes with LED-optimised lighting designs.
You can't have active pf correction for real in there (so pf would be 0.5). Pretty much all of them have bridge rectifier + capacitor. Most of them lack input filtering (common mode choke). The real issue is that even with the best light/power ratio, 2/3rd would be heat and that heat has to be dissipated. Electronics hate heat, even with full aluminium body, the light bulb requires plastic casing (as there is no earth available), so heat dissipation is heavily limited.
Personally I have mostly near ceiling lights in the house. The non-dimmable/remote control ones would be fully repairable. For ones with the remote controls I can't replace the microcontroller as I could not copy the firmware but all the power driver would be fixable (the controller can be replaced with esp-32 and blue tooth dimming, though). Of course, for most people that would be an extra hassle and fully replaceable lights are likely better. However, custom designs allow for the most optimal heat dissipation.
When it's only a 1 watt bulb it really doesn't matter. It will be massively eclipsed by your 3000 watt dryer or your 30000 watt car charger. And those will have a good power factor.
It is about how current draw is in or out of phase with voltage. Purely resistive loads won't affect the phase of the current. However if you have reactive loads like capacitors or inductors your current draw will either have its peak before the voltage has its peak, or after. This leaves you with less power.
In Germany going lower than cosphi 0.9 capacitive or cosphi 0.9 inductive requires special permission from the operator of the electrical grid.
If your installation affects the electrical grid in negative ways they can make you pay when something breaks or switch of your power to protect the grid.
On top of cosphi you also have ripple (basically "overtones" created by non-linear loads like unfiltered switching PSUs). If your THC (total harmonic current) exceeds 5% any power transformers in your net will have significantly reduced lifetime. But German grid operators allow a maximum THC of 15%.
So basically there are many ways how inductive, capacitive or switching loads can "deform" and shift the waveforms of the currents and voltages you are drawing, some of this can have negative effects on your own devices, some of it can have negative effects on the grid. Ideally you have a cosphi of 1 (so neither inductive nor capacitive) and a THC of 0% (no harmonics of the fundamental frequency present in the sine of the current drawn), but in reality you have to stay within what your grid operator allows.
Lower power factor means more voltage drop over distance as well as more power consumed per unit of work. For a AC 120v 20A single-phase circuit that is 100 feet long with a power factor of 1 (electric heat is pf 1, purely resistive) you can use #12 wire. For the same circuit with a very inefficient inductive load (a crappy motor) that has a power factor of .5, you need to use #8 wire, which is 4x larger (cross section area) than #12.
It also requires more power to extract the same amount of work as a more efficient power factor motor. Low power factor equipment wastes electricity and requires more copper.
In practice, motors in the US are generally .75 pf (fractional horsepower) or better, .90 pf by 20 HP, and .95 pf as HP approaches 500. These efficiency standards combined with variable-frequency drives means that new electric motors are pretty efficient at using power to do work.
It's the phase difference between the voltage waveform and the current waveform.
Power transfer is most efficient when the peak current is happening at the peak voltage. As it moves further away, the loss at generation point increases.
Put simply, lower power factors are inefficient. They draw more current, which means the components are either stressed or need to be bigger to compensate.
(This is all a big simplification): DC power is pretty easy to get your head around. In a resistive circuit you have a nice easy P=IV. But for AC circuits that include capacitance or inductance you sometimes get the current out of phase with the voltage. So you end up with "real power", "apparent power", and "reactive power". For some circuits you can balance out these inductive and capacitive loads. In the UK in the past factories that had a lot of motors (inductive loads) would also have big capacitor banks to help balance these out.
Simply put when there is lower power factor, there are higher losses into cables. And It is not always negative power that causes this (as many say). For a bridge rectifier the power is always positive.
I wonder if the power factor really matters for something as low power as an LED bulb. The grid has to be able to handle the many-kW loads from high-power appliances (ovens, stoves, kettles, car chargers, etc.) anyway. Doubling the current draw of a hand full of 1W-bulbs seems rather insignificant in comparison.
> It's not hard to make LEDs that actually last and have high lumen/watt ratio with off the shelf components.
Are you sure? It seems to me that most LEDs dim gradually over time. What do you mean by "actually last?" 10 years? 50,000 hours of operation? Less than 90% reduction in lumens on average over the advertised lifespan?
>It seems to me that most LEDs dim gradually over time.
Of course, esp. the warm white ones (they got more phosphor). For 50k hours, the LEDs are likely to drop to 70ish% of their original brightness, provided they stay below 50C[0] - lumen maintenance. That would depend on the spec. of the LEDs but it should be a ballpark number.
This nice video re-ignited my desire to find “the right” bulb manufacturer again (in the US market). I tend to go with Philips because I had pretty decent experience with longevity and customer service. But I wonder where do we stand now?
Are there any clear winners (considering the light quality, bulb longevity and of course the price)
A coworker and I both replaced almost every incandescent we had with cheap Walmart LEDs [1] about three years ago. That was 26 bulbs for me, and I believe quite a bit more for him.
All are still working fine for both of us.
I went around after installing them and measured their light output. A couple years later I went around and measured again, and there was little or no detectable reduction.
[1] they were a particularly good deal because the electric company here and Walmart had made some sort of deal that marked the 60 W equivalent and 40 W equivalent bulbs down to $0.17 each.
They are a good deal and I had replaced at least six. My assumption was it's just a manufacturing defect since it's low cost and mass produced. And I'll buy the same ones to replace them.
Also note, these are spotlight LEDs that in the past were halogen, and not your standard sized lightbulbs, which are probably a few generations ahead in terms of manufacturing.
The small spotlight bulbs (GU10 fitting) are the most difficult to make reliable, since there's very little space for air circulation to cool the lamp.
As this video says, Philips don't even make these bulbs with GU10 fittings, probably because they can't achieve the necessary reliability.
Got one too. Still alive and kicking after almost 5 years, but I feel like it's starting to show some slight 100Hz flicker, which it definitely didn't do when it was new. This is something that really annoys me so it's the first thing I check after buying one.
Still, everything noname I bought around the same time is either already dead, or just doesn't have as pleasing light quality.
Oh I love Big Clive, I haven't watched him in so long.
I wonder if we could somehow get the same under-powering effect using special bulb socket splitters that reduced the power. So you could have two bulbs in your lamp running at half the power each.
For a city that continuously runs AC units at every corner, even outside, this is a weird priority. Though I guess it's a good example of how progress can be limited by mutual interests of competitors.
The wording for the cookie banner is absolutely hilarious. "By using our website and services, you expressly agree to the placement of our performance, functionality and advertising cookies." The "expressly" there is completely nonsensical, as there's nothing explicit at all about it. It's like somebody read that the GDPR requires "explicit consent", and thought that they could just add "this implicit consent counts as explicit" to their banner.
https://news.ycombinator.com/item?id=25808959
https://news.ycombinator.com/item?id=25758652
Clive has also made a few more videos since then, showing how reducing the power at which LEDs are run at can dramatically increase their efficiency - they put out less light, but at more lumens/watt.
I guess the author isn't familiar with what caused dishwashers to become more efficient[0]. In general I don't see how anything except government regulation can force these improvements whenever consumers find themselves in an "any color as long as it's black" kind of position regarding market choices, and this looks like it is one of those situations.
EDIT: Also, before anyone thinks I'm blindly pro-government regulation, Philips got rich by exploiting the fact that the Netherlands did not enforce patent laws for a large part of the 19th century. This let them combine innovations that other competitors didn't dare to use due to fear of being sued by Edison, resulting in them having the best light bulbs available at the time (the fact that many competitors that Edison sued out of business happily joined Philips probably helped too). So it's all context-dependent.
[0] https://www.reviewed.com/dishwashers/features/why-obamas-dis...
The fact that you feel the need to add this edit makes me think about how the conservative framing of almost every issue is assumed to be true by both sides. Nobody would make an edit like this about being blindly anti-government regulation, because that's the default position. Government regulation is not bad by default and your overall point is excellent.
This is lobby, and it's bad for economy and the environment.
Do you have any concrete arguments for why these two cases of successful government intervention do not hold up? Do you have any argument to suggest that the situations they describe are exceptional instead of common and all over the marketplace?
That "might" is doing a lot of work. We're stuck with many mediocre options in many markets because they're locally maximal outcomes but not maximal outcomes for society, which is exactly why we need a strong government.
Most of the constant current driven LED would have similar characteristics when they are not overdriven. 25KHz+ PMW driven LED would have similar features as well. It's not hard to make LEDs that actually last and have high lumen/watt ratio with off the shelf components.
Overall the main culprit is the LED installation mechanism, e.g. E27 - there is just not enough room to make efficient light out of that. Near ceiling lights do look better (at least to me) and they can have both not overdriven LEDs (say 90mA per 5730), high (over 87%) power factor, decent passive elements... and they are repairable with some electronics background.
Out of curiosity, I teared down a few failed ones and they were all gone because one of the LEDs on the LED array blew up. Here is a picture of a failed one: https://imgur.com/aOg7D7S
I'm very doubtful about the longevity claims.
https://en.wikipedia.org/wiki/Phoebus_cartel
When you visit an eyeglass store in the US, it's likely all Luxottica frames, especially the chains they own:
https://www.luxottica.com/en/retail-brands
One of the mfg bulbs go bad. I think I'm at about a 50% failure rate with them. Something like 10 out of 20 bulbs. None of the other mfg bulbs have gone bad. Basically I bought bulbs that were on sale. The "bad batch", IIRC is when average 60W led bulbs were in the USD$2.5 range and I got them for about $1.0, so about 6 years ago. I bought some other bulbs at a similar price point/date, and they are still all fine.
So, it could just be you got unlucky with a crappy mfg. Try some different bulbs.
Also only the PCB is aluminium, not extra heat sinking. Is this GU10 socket?
This is at least better than the CFL situation. At least in my house they had an absolutely horrific failure rate. The whole shift to CFLs had to have been an environmental disaster.
I still buy LEDs because I like the light from them, but man, I'm basically paying 10x the price (I like the 100-watt equivalent bulbs) for the same quality bulb. It's depressing.
That would be the case only if the driver is exceptionally bad, heat kills them pretty much (which would be continuous use). Any usable driver should limit the inrush current.
@mrtksn, pretty much all LEDs you buy in the shop are heavily overdriven to get 1) high power, 2) shorten their life span, so you have to replace them, i.e. the planned obsolesce.
It's exceptionally rare to see e27/e14 designs that would be half-decent. The "dubai ones" are an example of not-overdriven leds with okayish electronics but they still are subpar.
--- (Edit) A simple solution for DYI, are LED strips - they are not super efficient as they have current limiting resistors (i.e. heat) but they can be dimmed easily (PWM) to last for long long time. Also they can be used in a lot of creative ways.
For a laptop battery I might have put in more effort to dispute with the manufacturer, but for two ~5€ lightbulbs I just couldn't be arsed.
50% failure rate in 2 years at domestic use is a very poor performance IMHO.
Electroboom recently reviewed / explained common failure cases: https://www.youtube.com/watch?v=lIK1jnYr0yw
You can't have active pf correction for real in there (so pf would be 0.5). Pretty much all of them have bridge rectifier + capacitor. Most of them lack input filtering (common mode choke). The real issue is that even with the best light/power ratio, 2/3rd would be heat and that heat has to be dissipated. Electronics hate heat, even with full aluminium body, the light bulb requires plastic casing (as there is no earth available), so heat dissipation is heavily limited.
Personally I have mostly near ceiling lights in the house. The non-dimmable/remote control ones would be fully repairable. For ones with the remote controls I can't replace the microcontroller as I could not copy the firmware but all the power driver would be fixable (the controller can be replaced with esp-32 and blue tooth dimming, though). Of course, for most people that would be an extra hassle and fully replaceable lights are likely better. However, custom designs allow for the most optimal heat dissipation.
When it's only a 1 watt bulb it really doesn't matter. It will be massively eclipsed by your 3000 watt dryer or your 30000 watt car charger. And those will have a good power factor.
In Germany going lower than cosphi 0.9 capacitive or cosphi 0.9 inductive requires special permission from the operator of the electrical grid.
If your installation affects the electrical grid in negative ways they can make you pay when something breaks or switch of your power to protect the grid.
On top of cosphi you also have ripple (basically "overtones" created by non-linear loads like unfiltered switching PSUs). If your THC (total harmonic current) exceeds 5% any power transformers in your net will have significantly reduced lifetime. But German grid operators allow a maximum THC of 15%.
So basically there are many ways how inductive, capacitive or switching loads can "deform" and shift the waveforms of the currents and voltages you are drawing, some of this can have negative effects on your own devices, some of it can have negative effects on the grid. Ideally you have a cosphi of 1 (so neither inductive nor capacitive) and a THC of 0% (no harmonics of the fundamental frequency present in the sine of the current drawn), but in reality you have to stay within what your grid operator allows.
It also requires more power to extract the same amount of work as a more efficient power factor motor. Low power factor equipment wastes electricity and requires more copper.
In practice, motors in the US are generally .75 pf (fractional horsepower) or better, .90 pf by 20 HP, and .95 pf as HP approaches 500. These efficiency standards combined with variable-frequency drives means that new electric motors are pretty efficient at using power to do work.
https://www.energy.gov/sites/prod/files/2014/04/f15/amo_moto...
Power transfer is most efficient when the peak current is happening at the peak voltage. As it moves further away, the loss at generation point increases.
(This is all a big simplification): DC power is pretty easy to get your head around. In a resistive circuit you have a nice easy P=IV. But for AC circuits that include capacitance or inductance you sometimes get the current out of phase with the voltage. So you end up with "real power", "apparent power", and "reactive power". For some circuits you can balance out these inductive and capacitive loads. In the UK in the past factories that had a lot of motors (inductive loads) would also have big capacitor banks to help balance these out.
All of them are pure resistivity load - resistive load has power factor of 1. Ok, stoves might be the inductive kind but those run on 3-phase.
Are you sure? It seems to me that most LEDs dim gradually over time. What do you mean by "actually last?" 10 years? 50,000 hours of operation? Less than 90% reduction in lumens on average over the advertised lifespan?
Of course, esp. the warm white ones (they got more phosphor). For 50k hours, the LEDs are likely to drop to 70ish% of their original brightness, provided they stay below 50C[0] - lumen maintenance. That would depend on the spec. of the LEDs but it should be a ballpark number.
[0]: https://datasheetspdf.com/pdf-file/1404455/BIVAR/SM5730UWDW0...
All are still working fine for both of us.
I went around after installing them and measured their light output. A couple years later I went around and measured again, and there was little or no detectable reduction.
[1] they were a particularly good deal because the electric company here and Walmart had made some sort of deal that marked the 60 W equivalent and 40 W equivalent bulbs down to $0.17 each.
Also note, these are spotlight LEDs that in the past were halogen, and not your standard sized lightbulbs, which are probably a few generations ahead in terms of manufacturing.
As this video says, Philips don't even make these bulbs with GU10 fittings, probably because they can't achieve the necessary reliability.
Still, everything noname I bought around the same time is either already dead, or just doesn't have as pleasing light quality.
I wonder if we could somehow get the same under-powering effect using special bulb socket splitters that reduced the power. So you could have two bulbs in your lamp running at half the power each.
I mean this is basically free PR
https://en.wikipedia.org/wiki/Phoebus_cartel
https://www.mea.lighting.philips.com/consumer/dubai-lamp
https://produkte.migros.ch/osram-led-special-t26-25-2700k-e1...
https://produkte.migros.ch/osram-led-special-t26-40-2700k-e1...
https://produkte.migros.ch/osram-led-retrofit-clas-a-60-e27
The Dubai one consumes 1 watt and produces 200 lumen, that's more than 2x difference in efficiency.