One of my favourite Big Clive moments was when he turned up on Barry Lewis's channel (a thoroughly nice and decent British amateur cook and kitchen gadget tester) when Barry had plugged a 120v rated grilled cheese toaster into a UK 240v household supply, near kitchen conflagration hilarity ensues:
"The best bit about this video is seeing you plug a 120V appliance into 240V without realising it, and then watching the horror unfold as the grossly overloaded appliance makes loud stressed noises and emits flames. The appliance doesn't just run at twice it's normal power, it's closer to four times the power. Good job. Very entertaining."
I've followed Big Clive for about 5 years now and he's a great presenter. Especially his "what cheap shite I bought from my local Poundland on the Isle of Man" videos...which actually kinda reassure you that Poundland's "cheapo" electrics aren't that terrible or are going to burn your house down.
Offtopic but serious question: how do you downvote someone on HN? I thought they flat-out turned that off as a design decision, like you see on some subreddits.
At least 500 karma points are needed, but even with that, nobody can downvote a reply to their own comment, which is what your parent commenter was referring to (albeit rudely).
You need a certain amount of karma before it is enabled. I can downvote comments - but not posts. I've got just over 500 karma. I think 500 was the tipping point.
The guy I replied to spent 3 years thinking downvotes didn’t exist on this site simply because he wasn’t enough of a windbag to earn 500 karma. That’s not fun, it’s a tragedy. What would be fun is a discussion that leads to them changing this lame rule. Quiet people shouldn’t be punished for being quiet.
It put a big smile on my face that big clive is the first comment here :-) His videos are hypnotically good! He's tied with aVe IMO as the best teardown channels on youtube.
I admit he got me good with his "Scottish Army MRE" video though in my defense, who expects an April Fools video in January? I really enjoy his teardowns and the accompanying circuit diagrams - amazing stuff.
He has a lot of other stuff too, but I agree, he has probably made the largest and most in-depth collection of LED lighting videos on YouTube. Unlike a lot of other review videos, he doesn't just take the lamps out of the box, plug them in, turn them on and make some comments; he takes them apart and scrutinises them in a lot of detail.
That whole stuffing 240v into a BabyBel cheese and Co-op mincemeat pies thing was quite amusing, I guess the festive season can be "slow news" days even for video bloggers. That said, back in the early 90's when I was a field engineer, a mate and myself used apply 240v to foodstuffs in the company workshop "just to see what happened" when things were a bit slow. The resulting smells were somewhat odd.
Have you ever seen "PhotonicInduction"? He did a lot of very dangerous and borderline insane stuff with electricity like running up a 20KW bulb in his house: https://www.youtube.com/watch?v=LT5_-A0m8_U (or, more famously, chucking a cinder block into a washing machine)
He married someone from India and the UK Home Office wouldn't grant his partner a visa initially. I think it's all sorted now and she's now living with him in the UK:
The cheaper the design the higher the failure rate. The heat seems to kill these things over time. For example I just ordered another warranty replacement from one brand called Hyperikon that uses similar types of designs for floods in my Kitchen. These have a much higher failure rate. Others seem to last forever. Between using electrolytic Capacitors and other tings that are prone to heat and small enclosed areas it is no wonder that the newer, cheaper bulbs fail quicker.
Since one big reason for going with LED is to be more energy efficient and greener, it would be interesting to compare the carbon footprint of these cheap LED bulbs that fail quickly with an incandescent and CFL over the entire lifespan. If the bulb lasts longer the numbers are pretty easy, but given short lifespan of some and their use or rare earth elements it would be interesting to see the full analysis of the environmental impact of each one.
It's ironic that LEDs themselves last for 30,000 hours but bulbs created with LEDs can wear out much quicker than incandescents because of crappy power supply design. I can't help but think manufacturers must have initially worried that LEDs would kill their continuing revenue streams. When they finally figured out they could design the power supplies to fail quickly, they must have been dancing in the streets.
I would consider market forces a more reasonable explanation: most people are going to pick the cheapest option. The success of cheap, flickering bulbs attests to that, I think.
It doesn't help that the old Edison screw is terrible for cooling.
There is no sane option but to pick the cheapest option. You go to the shop here and see 3 or 4 brands, and looking at the details they all seem identical. You choose your voltage, colour warmth, and if you want to pay a premium for the only brand name you recognize.
You can pick the cheapest within a range, but not the cheapest overall.
E.g. I get relatively good bulbs by requiring CRI > 90. Or just requiring no flicker, as a feature. When the producer has more than one parameter to optimize, they seem to come up with more well-rounded designs. To say nothing of the useful features I mentioned.
Yes, I likely pay 2-4 times more for these bulbs. But they perform, and they stay for years of intense use. The cheapest overall bulbs are designed for other purposes.
Good idea. I have never noticed flicker on any of my lightbulbs. I wonder if some of us are unable to see (most) flicker, even though we may otherwise have great uncorrected vision?
But little do you know, the brand was bought out by a different company so they could sell cheap junk using the reputation of the brand you chose and the review is now out of date.
Unless there's some regulatorially-mandated set of minimum standards, given the thin information channel of price ("what will I have to pay out of pocket right now for Product X, with no other indicator of quality, longevity, or potential risks"), the market will pursue low costs with an iron determination, over all other factors.
Consider that a corrolary of Gresham's Law, which in its essence says that given a nominal currency value, currency of lower determinable quality (lower specie metal content) will tend to drive higher-quality currency from the market.
"But wait, Dred," you say, "You just said that there was no other indicator of value. Why is low specie currency driving out high specie?"
Because that's not the relevant value, and it actually is apparent (or at least sufficiently so to a certain number of traders in coin). It's the nominal value that's the determinant for circulation. In the case of coin, it's face value which determines market exchange. Put another way: in a world of copper-painted zinc lozenges circulating as pennies, a diamond-encrusted gold penny will still only net you $0.01 in exchange value as currency. You're better off holding on to it or trading it for its specie / gem value.
With bulbs or electronics, the logic shifts around a bit, but the result is similar. Price becomes exchange value, product quality factors (colour registration, colour range, flicker rate, dimmer compatibility and performance, heat dissipation, any electronics buzz or hum, diffusion pattern, and most of all, that one factor you cannot determine in 3-5 minutes at a store: product life (or more properly: product-quality-life, the period of time it functions adequately before driving you bananas, even if it doesn't fully fail).
You're standing in a store aisle, or staring at an utterly uninformative webpage of quite probably counterfeit and fake-reviewed Amazon LED bulbs as well as whatever LED-bulb-adjacent search results Amazon's Alexa Search Engine has decided it needs to distract you with today, and you see two things:
- LED bulb.
- Price.
And "you" is the typical shopper. Which is to say, one who's not read this particular screed.
What are you going to do?
You're going to pick the lowest-cost bulb you see and offer a prayer to His Noodliness.
This is, incidentally, a chief reason why pure market solutions tend toward abysmal product offerings in the absence of some external regulatory, ratings, or similar systems.
There's an alternative, and some retailers follow this. Marks & Spencer in the UK long made a tradition of carrying only one exemplar of a brand, generally represented as "the best there is". That is, M&S undertake the research and quality determination costs for its customers. Your research problem is reduced to "shop M&S".
In the US there are retailers who follow a similar strategy, though emphasizing the price-quality relationship. Costco is among these, where typically only a small range of products within a category is offered (Trader Joe's operates similarly, though only in food). Friends purchased a convection toaster-oven from Costco recently -- not the cheapest model out there, but one which performs quite admirably, will handle a frozen pizza and most meals for 2-3 quite well, and (whilst slightly tricked out on controls) is fairly simple and straightforward to operate. On balance, "shop Costco" has been justified in this case.
The pure-pricing model typically works best (if at all) for nondifferentiable commodities of minimum complexity. Foodstuffs, bulk construction materials such as sand and gravel, coal, oil, etc. And yet even within each of these in developed economies you'll find grading and quality control, some imposed by government mandates, some self-adopted by industry.
And outside of bulk commodities -- in services, in wages, in rents, in interest and risk markets, in public goods -- pure market-based pricing absent various corrections, breaks down entirely or leads inevitably to pathological conditions. See "Iron Law of Wages" and "Law of Rent" -- the twin prods of numerous revolutions.
The problem with your "sane option" is that at mass market scale, say, the 100 million or so households in the US, or the 300 millions or so in China, it simply does not scale. Information and expertise implies some level of clustering and trust relations. Commodity pricing doesn't provide for this.
"This is, incidentally, a chief reason why pure market solutions"
Yes, but what is a "pure market"? I like to believe that my local home improvement or grocery store has some kind of concept of their own brand and puts some effort into getting decent products on the shelf, whether store brand or not. So, I hope that their markets are not as "pure" as online sites where anyone can sell.
You might read "pure market" as "self-interest seeking", as opposed to some entity acting as a regulator in the public interest.
In the case of a brick-and-mortar store, part of what the retailer is paying is for a physical retail location, which you might think of as access to a specific cohort of local traffic (foot, vehicle), and customers, in the sense of patronage of a retail establishment by custom (https://www.etymonline.com/word/customer).
Since that is a finite set, and there are often (though not always) competing retailers, practices which alienate customers tend to be counterproductive.
I was actually thinking of the local B&M hardware store where the staff know their products, and customers, and I've been making a series of (to date, satisfactory) LED bulb purchases whilst writing my earlier comment. What is ultimately carried is a function of the store's specific traffic, the parent corporation, and other arrangements (there are periodic government and utility incentives for low-energy bulbs), but the local staff and management have some agency.
Existential Comics' "Freedom Monster" thought experiment is worth consideration:
"yet even within each of these in developed economies you'll find grading and quality control"
I think it's an oversight to mention bulk commodities as an exception to the need for quality information. Things are bulk commodities because there are standards and regulations that enable them to be traded easily.
"Bulk commodity" in trade means that a good is comprised of units (discrete or continuous) of equally-interchangeable quality, as opposed to a good which is comprised entirely of unique entities.
If I take a brick, or a shovelful of sand, or a barrel of rainwater, a ton of iron, a bushel of wheat, a dollar or yen or pound from you, and return another to you later, it doesn't matter if you get back the same unit or a different one, so long as the quality is equivalent.
If I take a Monet or Picasso or first-edition Gutenberg Bible or the Hope Diamond, you're going to want that specific item back, not a replica, no matter how exact or precise.
If I order a package of washing-up liquid from Amazon and that goes walkabout, Amazon can simply ship another unit of soap and I'll be happy.
If I order a pair of prescription eyeglasses and those go walkabout, Amazon cannot simply redirect another customer's order to me -- the specific nature of the glasses -- prescription, frame choice, fit, lens coatings -- matter.
Bulk commodities are highly undifferentiated. Within a commodity class, they're infinitely substitutable. Once a price is determined, what you get for that price really doesn't much matter. In other words, it's the realm of market-based exchange most amenable to price-as-the-only-signal you can find.
And yet, it still doesn't rely on price-as-the-only-signal.
Even amongst commodities, there are quality measures. You want your sand to be sand, not dirt or plastic. For specific applications, river rather than beach sand. Grain is controlled for moisture content, spoilage, and insect fragments. There are quality standards imposed by mechanisms other than price.
Your (apparent) objection to my comment was actually the point I was trying to make.
I think it's interesting to consider how consumer electronics (that is, brand name, widely distributed products) share the qualities of a bushel of wheat or barrel of oil. They aren't simple or interchangeable in that there are a vast number of unique types, but they are easily traded online because the important characteristics are all available and can be relied on. They kind of seem like they share salient characteristics with commodities. So are they commodities?
Except when you go to buy the bulb that is more expensive, hoping it works better, it just ends up dying FASTER. There seems to be no correlation between price and quality anymore.
The problem is that the heat from the power supply needs some place to go. So installing the bulb upside down traps heat, and putting them in an enclosure also traps heat. Probably 90% of installations is one of these two.
I've put in some "filament" style LEDs, these look like they will last longer as there doesn't appear to be a power supply taking up half the bulb (although there are electronics down in the screw part).
At a given efficiency, LED heat and light output are directly related. Human vision is not linear at normal brightness levels. So, less bright bulbs last longer, use less energy, while putting significant amounts of light.
Also, I see many people getting extremely bright bulbs and putting a large fully opaque light shade over them. That’s rather counter productive.
I found a supplier for access lights of commercial/high-end buildings [1] that mentioned 200,000+ hour L70 ratings.
There apparently are 200,000 hour L70 fixtures [2], but I haven't found any 250K, 300K, 400K, or 500K hour units.
If we ignored form for pure functionality to get the full-rated life of the bulbs, I wonder what the design/specs would look like. The longest DC power supply I could quickly find a mention of is an Antec line, so I wonder what a really long-lived (50-100 year) DC power supply design would look like.
Ironically, two of the most recent failures I've had (and I've only had a half dozen over the last few years) were heavy Philips bulbs that felt like they were solid metal.
The only LED failure I've had so far was a first gen Hue. Even then it's "just" a noisy power supply. All of my LEDs are either Hue (1st gen color changing or newer white-only ones), Ikea (dumb and smart), or some T8 drop in replacements that Home Depot had. The Ikea ones are, IMO, the best value out there. They turn on quickly, are cheap, and seem to be fairly long lived.
Just a heads up, the makers of Hue lights are now Signify. They license the name Philips from the health company Philips. Given my recent customer experience, I now believe Signify operates very differently from Philips Ligthing.
Philips is selling its small domestic appliances business and hoping to strike a similar deal regarding the licensed use of the name Philips. They are centering solely around the Health business now.
Yes, though it has absolutely no link with Philips apart from the name licensing, contrary to before the renaming. This isn't crystal clear to consumers, I think. And it only submerges when there are talks of selling the rest of the consumer business. Or when contacting the Signify support, which is insanely worse than usual Philips support. (And that is to say nothing of the online buying / shipping experience which is awful.)
I am reading this post on my laptop, in bed. In my grand parent's house.
There's a bedside lamp next to the bed and the bulb is at least 30 years old. Two generation of children have used that bedside lamp to read stuff at night or walk to the bathroom.
Meanwhile my LED - whatever the brand - are failing between 6 months and two years.
Somwething seems wrong if your non-cheap LEDs are dying that quickly. I have an entire house of LEDs, many going on 5-6 years old, and I only ever recall replacing two (and they were a pair, used outdoors -- only one failed, but I replaced both so they would match).
If you have 5-6 year old LED bulbs, then they represent the state of the market 5-6 years ago, and aren't necessarily representative of current quality.
Perfect! A great picture of the lightbulb. Its easily identifiable not :)
Nonetheless in the last 20 years all my normal light bulbs had to be replaced on a regular schedule which was often enough that i got slightly annnoyed by it.
An under-fed incandescent could last centuries. However it would waste 95% of its energy as heat instead of producing light. That's actually not much of an issue if you either live in a cold climate and the heat is useful to you, or if the light is only lit on occasion (attic, cellar, closet ...).
My wife and I were talking about what lamps we had that were LED vs fluorescent vs incandescent; one of the bulbs that we have that is of the latter is in our closet, and has been there since before we bought the house in 2002 (the house was built in 1973).
18 years now, maybe longer, but it still turns on and has yet to burn out. Of course, we don't use it much, and it's only something like 30 watts (clear bulb).
it's very rare for me to have an LED bulb that last less than 2 years... but I have 4x 18-watts recessed led fixtures in the kitchen and one burnt out after about 2 years and they are no longer manufactured so I have to find one that look-a-like....
I have some spots in my house where it's a major hassle to change the bulbs, so the long lifetime of LEDs was an important selling point there (in fact, much more important than the energy savings). Thankfully I've had good luck with those (all Cree bulbs), but I'd be pretty annoyed if I bought LEDs for those fixtures and ended up having to replace them after a year or two.
If you're in Canada, I've had good luck with Luminus Elite (different from the Luminus non-elite available at non-Costco locations) PAR20 and GU10 bulbs from Costco. Or I use Philips warmglow from Home Depot or comparable Ikea bulbs (which have brighter GU10 than the Luminus Elite line) in locations where I want non-ugly dimming.
Yes, the "lifetime" of the bulb tends to be quoted in terms of the expected life of the specific light emitter, i.e., the LED itself. In reality, something else almost always fails first, and much much faster.
The LEDs stated lifetime is based on an expected gradual reduction in brightness, with an industry agreed cut-off at, IIRC 90% of original light output. If you have LED lamps that don't fail (typically due to heat) they will just keep getting gradually dimmer forever.
Prior technologies have expected failure built in. An incandescent bulb's filament will fail eventually (making the filament more robust reduces light output so now your bulb lasts longer but is less useful or costs a lot more to run) and either type of fluorescent very slowly loses the gas that makes it work.
I have about 20 Hyperikon dimmable BR30s that are 4-5 years old and have been excellent. But I bought some cheaper ones for the kitchen more recently and they have had a bad failure rate. The older Hyperikons were excellent, the newer ones, not so much.
I had pretty much this exact thought recently after a couple of LED bulbs failed in fairly close succession in my house, despite a purported 20/50K hour lifespan (I can't recall which exactly, but neither reached).
They're own-brand ones, sold by a major supermarket chain here in the UK (Sainsburys), so I'd expect them to be 'ok' at least. Certainly not 'cheap'.
Compared to ye olde incandescents, the complexity and material make up is significantly less recyclable and looks to my eye to require a lot more invested energy to manufacture.
-
I've wondered this about so-called 'smart meters' too - how much energy does it take to make one, and how long to 'pay off' that investment?
A crude way to estimate embodied energy is just from the cost. It had to take less than $1.25 worth of energy to produce and distribute. That might have been something cheap and polluting like heat from coal, but it's still not much. You could look up the cost of that in India or China or wherever it's cheapest to get the worst case amount of energy or CO2 emissions.
Alternatively, plastic takes about 100 kJ/g to produce. So if the bulb has 1 g of plastic, that's about 0.03 kWh of energy. Nothing compared to the savings in electricity using it.
Where this gets interesting is when governments decide that if LED energy efficiency is good, more LED energy efficiency is better. There's a push for minimum energy efficiency standards to be really close to the cross-over point where consumers just barely save money compared to less efficient bulbs if they meet the nominal lifetime specs, on the pretense that it's helping out consumers who are too stupid to realise that the higher upfront cost saves them money - at least in the EU and the US pre-Trump. Given the combination of more complex designs and more incentive for cost-cutting, I can't see those bulbs meeting the lifetime specs, and I do wonder if it'd actually save energy.
Were smart meters about saving energy? I sorta assumed they were just there to make things easier on the power company. Although I suppose if it means that a person doesn't have to drive a car to every meter every month, that's probably a win.
I think they can help with saving energy indirectly: by letting power companies give customers more detailed information about their usage. But I don't know how often people take advantage of this.
I've shaved off around .15 kwh of energy that was used 24/7 ($25/month in California) based upon usage information from our energy company.
My 'smart meter' is Zigbee enabled, and I own an adapter that spits out data from the meter in XML format. I can see power consumption with ~400 Watt resolution every 10 seconds without messing around in my breaker box attaching things.
Downside is that we're a decade into net-connected thermostat and HVAC systems and none of them support any of this.
Smart meters in themselves don't do anything to save energy. However, they make your energy consumption level more visible, which in turn might encourage you to use less electricity.
Not to mention that the "wasted" energy of incandescent light bulbs, might actually not be, at least not entirely - since they pretty much work as electric heaters...
I'd imagine on the whole it's wasted. Half the year they aren't just generating wasted heat, they're increasing cooling costs. The half of year that the heat is wanted, it's usually above you, and I imagine that heat is being generated less efficiently than a purpose-built device.
> that heat is being generated less efficiently than a purpose-built device.
“efficiency” is meaningless when heat is desired. if you pump 100W into a device, and it “wastes” all that power, you have successfully generated 100W of heat. only heat pumps can be more efficient.
Exactly. And if you have a whole house/apartment electric heating system it is very likely that it is in fact heat pump based. So you are better off using that for heat than the lights.
If your whole house/apartment heating system is gas or oil based, that will probably be even cheaper than the heat pump electric version.
>if you have a whole house/apartment electric heating system it is very likely that it is in fact heat pump based
Hmm, where? I have never knowingly been in a house with a heat pump in the UK, and lots of cheaper flats have electric radiators and immersion heaters.
Not quite – some energy was lost converting from its source form to electricity. If you heat your house by directly burning natural gas (the most common way in cool countries like the UK) then it can still have an efficiency advantage over converting electricity to heat.
This is true but - especially if your house is not well insulated - getting heat in the right place matters too - most people don't need to heat their roofs or ceiling spaces. There's also a more minor point on time of day you want heat or not.
The only thing that can be more efficient is one of the most common forms of heating devices.
During the summer, it's also extra inefficient to pump 100W into a device producing heat and then pump even more power somewhere else to cool down the room.
I had a friend with a condo in Chicago with a 400-500 dollar a/c bill plus another 200-300 year round bill because the builders hated the earth and universally used halogen bulbs. But on the plus side the heating bill was probably 50 bucks for the year.
Heat pumps have higher than 100% efficiency because they extract heat out of air / water. 100% is a very low efficiency for electric heating. Anything lower than 400% is considered below average.
Resistive heaters are very low efficiency when compared to heat pumps. How much more efficient depends on the climate you are in, of course. So, in net, they require more cooling in the summer, and the heat they provide in the winter is more expensive than the heat pump that you are likely to be using (in a home anyway).
During cold winter months, when one is otherwise attempting to add heat to the indoors, the 'waste' heat from incandescent bulbs is not really 'wasted', since it contributes slightly to maintaining the interior room temperature.
If they are resessed down lights you mostly end up heating roof or ceiling space. And in summer not only do you consumer power for heat you don't need, you end up consuming power to remove said heat with AC. Also in winter there are much more efficient ways to heat than electricity, such such as heatpumps.
For incandescent, the carbon footprint all depends on the source of your electricity, because the difference in efficiency is just enormous.
For CFL, it's a bit more complicated, but keep in mind that they do have quite a bit of glass (takes energy to make and form), need some mercury, and at least as much electronics (if not more) as LEDs. Furthermore CFLs don't necessary fail completely, but degrade over time. Thus they're a bit less efficient than LEDs to begin with, but a lot less after a few years.
So overall, unless the LED is so shitty as to consistently fail within weeks, there's no doubt their total energy bill is competitive, and it's even more clear if you limit the comparison to products of similar quality.
I think it's more about how well balanced all of the individual LED's in the bulb are. Unbalanced circuits have more current flowing through some paths than others and leads to premature failure of individual elements and it's usually rapidly downhill from there for the rest of the bulb.
LED bulbs that aren't current controlled are rare now. You're right - just sticking a bunch of raw led's in parallel over a "12v" source is a recipe for disaster.
Yep, heat definitely apepars to be the biggest factor in these bulbs dying. I have Hyperikon BR30 bulbs in various locations. About half of the bulbs installed in insulation-contact recessed cans have died, but those in exposed fixtures are fine.
Philips bulbs seem to do much better in the ceiling. If only they'd make 3000k bulbs...
I’ve had great luck with Hyperikon bulbs. I consider them a premium brand, due to simultaneously delivering high CRI and lumens per watt. Zero failures out of at least ten (over 2-4 years, bought in three batches).
Hopefully you bought a problematic model, and they’re not uniformly going downhill.
Exactly this - I ended up going with a brand in the UK called LAP for the downlighters in the bathrooms in my house. Not a single one has failed since 2016 when I installed them.
They weren't cheap, but they weren't mega expensive either.
I have had a few (3 or 4) LED bulbs of a reputable brand fail. To their credit, after one email to support a new package of bulbs has arrived at no charge. So no milking their revenue stream in that case.
Even open to the air might not be enough. I just had a couple of self described "revolutionary thermal design" [1] bulbs fail after less than a year in a one end open glass fixture.
I’ve installed Phillips LEDs in various enclosures and had no issues in 3 years. Frankly, the hum issues that I have encountered with dimmer switches and LEDs (even when not dimmed, and even when using supposedly dimmable LEDs) make them more suitable for use in an enclosure. That way you can’t hear them buzz, like in an open chandelier.
Some of these cheap LED bulbs have horrible flicker. For a fun time, record a slow motion video of a cheap bulb on your phone and play it back. Like half the damn video is black (pretty sure it is because they literally chop out half the AC sine wave with a diode). LED Christmas lights are also really bad.
Then compare it to something like a Phillips hue bulb. The difference between the two is pretty big.
On a side note I sometimes wonder if other animals have a quicker “refresh rate” on their eyes and see nothing but very visible flicker on these bad LED bulbs. I also wonder if some animals could see the raster scan in action on old CRTs...
It’s not like these bulbs are flickering in a way you can perceive just by slowing it down by 2/3/4x. They’re flickering at 10000+ Hz. The reason cameras perceive the flicker is that they take discrete samples, with an (electronic or optical) shutter in between. This creates a beat frequency with things that also flicker, in effect “tuning into” a HF flicker and “lowering it” into the visible range, the same way a superheterodyne radio tunes into a HF carrier wave and brings it down into the audible range.
The 10khz+ flicker only comes about if the bulbs have a proper switching power supply (and if that switching power supply itself runs at 10khz+).
But for the 'real cheap' bulbs, they likely (due to being "real cheap") have either a half wave or full wave rectifier (i.e., no switching PSU) which results in the LED's having a flicker at power line frequency (either 50hz, 60hz, 100hz, or 120hz depending upon which combination of line frequency and full/half wave rectifier is present).
Now that incandescent bulbs are almost gone, I wonder if it will become common to have wires carrying DC in the ceiling instead of every bulb having to implement the AC to DC conversion as cheaply as possible.
With renewable energy and local batteries, it would make sense to have DC wiring. It used to only be that off-grid systems had batteries (usually lead-acid), and they always just use an inverter. But now with battery packs such as the Tesla Powerwall, even grid-connected houses have DC storage.
Unfortunately, Powerwalls and other similar products are made with built-in inverters and connect only to AC, there is no DC tap. And there aren't any standards around DC wiring and small appliances, so it isn't likely to get traction.
I'm mentioned before (https://news.ycombinator.com/item?id=21109247) that I've seen a DC installation at a friend's house, he used 12V cable lights and DC bulbs, so it seems to work.
But where is the DC coming from? If it is a cheap power source (a rectifier plugged or wired into AC), the DC into the lights could be intermittent, and so cause flicker. For true continuous DC, you need a good rectifier or get the current directly from batteries.
Then we can also use this for our video camera’s and wireless access points. Maybe we can add network to those cables. Let’s call it “power over ethernet” ;)
I have 960 fps video of some LED bulbs in my house. It's very apparent that some bulbs do not flicker (perceptibly), some flicker at 120Hz, and some flicker at 60Hz. At night I can sweep my eyes across the bulbs and see the dashed lines resulting from them turning off and on.
Yep ... it's the digital (meaning on-off) version of the slow moving line you see if you point your camera at a TV (or computer) screen - the speed that it happens at is the sum and the difference of the two frequencies (and perhaps at the sum/differences of their harmonics).
In the CATV industry, we used to send 110 analog channels across roughly 800MHz of spectrum (54MHz to 860MHz). Other than background and ingress noise, a major source of noise (effectively) was distortion - non-linearities in the amplifiers that kept the signal at a reasonable level. Composite Second Order (CSO) and Composite Triple Beat (CTB) distortions were two values that equipment transporting these signals would generally call out - they are analogous to the summing and differencing of two signals and three signals (respectively) as described above for the lightbulb, but imagine doing that with 110 different frequencies simultaneously.
Filament LED bulbs usually flicker at 120Hz (with high duty cycle) and Christmas LEDs at 60Hz (with a short duty cycle). Other LEDs are, indeed, high frequency like you describe and are perceptually effectively constant.
I guess you are right about the beat frequency bit, but I still assert that the cheap LED's are cycling on and off 60 times a second (or 30.... math is too hard early in the morning). That is a bit low for my taste.
No, as reported by other posters, most bulbs that flicker do it at 120 Hz or 60 Hz. This is easily confirmed by eyesight alone because 60 or 120 Hz can be perceived when an object illuminated by the bulb (eg. your hand) moves quickly in front of a dark background.
Yes I find these cheap LED bulbs nearly intolerable-the flicker is especially noticeable because they transition so quickly from fully illuminated to fully dark. I’ve always perceived a slight flicker from fluorescent lights but because the phosphors smooth it into more of a fade in and out it’s much less distracting. I definitely think people perceive these things differently...I recently stayed in a hotel room lit entirely by these cheap LED bulbs and it was the first thing I noticed from walking in. It really bothered me that every light fixture strobed as I moved my eyes, but my girlfriend barely noticed it and only after I pointed it out.
The OLED displays in current smartphones also exhibit a visible strobe effect at lower brightness settings. I figured I’d get used to it but ended up returning my iPhone 11 pro after a week because it was driving me nuts. I’ve met a few people who perceive it as well but most seemed to have no idea what I was talking about.
The filament bulbs are almost all like this. Not too bad if they're using an H-bridge (120Hz with high-ish duty cycle), but Christmas LEDs I think might use no separate diodes at all, just relying on their own diode-ness. That means they're 60Hz with a less-than-50% duty cycle.
Filament LED bulbs (the ones that look almost exactly like clear incandescent bulbs with a visible filament when turned on) do this because they have to fit their electronics in the tiny metal base. Room enough for some diodes (and maybe a capacitor), but usually not a full set of high frequency switching electronics like in most other LED bulbs (although I've seen some filament bulb tear downs that might show more advanced electronics). But again, doesn't bother me due to their high duty cycle and relatively okay refresh rate.
(At least, this is all my impression. Please correct me.)
Dogs apparently have a visual sample rate of around 75Hz, meaning they wouldn't see much in an old analog TV, but newer HDTVs might look realistic to them.
I’m very suspicious of this 75Hz number. We’ve also heard that 60Hz is the limit of human perception but I think we all know now that we actually perceive higher than that (seeing flicker at 120Hz, blind A/B testing of 240Hz monitors, even perceptual flickering for some people in 1000Hz PWM LEDs).
Even if that 75Hz number were derived from deep understanding and measurement of the entire visual pathway, there are variations in perception for people, there most likely are for dogs as well.
60Hz is the 'flicker fusion rate', meaning if you were changing the frequency of a flashing, stationary LED, ~60Hz would be the frequency where perception transitions from visibly flickering to apparently continuous. It's a lower threshold for refresh rate in the human eye. When you have complex refreshing images (a 2d computer screen rather than a point-like LED, diverse motion, depth, etc), you are likely to notice flickering (or tearing, jittering, non-smooth motion) if the refresh rate is near this minimum.
People routinely watch movies which refresh at 24Hz and motion is apparently continuous.
Perceiving flicker has a higher threshold. I can perceive 60Hz flicker in my peripheral vision easily enough.
Other sources online (you can find lots of them but I didn't see an obvious authoritative article) suggest that 16Hz is the flicker fusion rate for motion in humans.
Flicker fusion for continuous brightness (CFF) is somewhere around 30Hz per https://www.nature.com/articles/s41598-017-15034-z and it rises with brightness - this paper is about testing how it changes in different circumstances.
This is often exacerbated horribly by the fact that the 50 or 60 FPS output doesn't divide well by the 24 FPS source material. It was smoother in the cinema, when it was projected at 24 FPS...
The limit depends on the size of the source, contrast, location on the eye of the source (rods are much ‘faster’ than cones, so we see flicker better out of the corners of our eyes), and probably a few other factors.
1000Hz I find hard to fathom, though. Would want to see what these signals look like.
Classical 50/60Hz lighting looks like a stroboscope to chickens. There is a whole field of research on the effect of light regimes, light color and flicker on poultry farming efficiency. I couldn’t find a scientific link, but see https://agrilight.nl/lichtadvies/pluimvee/?lang=en&-en
Had a dog that would press her nose right up to an ipad screen and move her face around to follow the squirrel, bird, or dog that interested her. She would get very excited about it. She also watched TV. At one point sitting on hind legs trying to elevate off the bed to be inline with the center of the image.
Only one of my dogs appears to recognize things in the TV. She'll bark at something on screen, and my other dog will run to a window or door trying to figure what he missed. Always wondered why...
This is actually a pretty big deal if you work with high speed or rotating machines. Using an high flicker LED bulb over your lathe (for instance) can cause it to appear static when it is in fact running due to visual aliasing. There are lines of flicker free LED bulbs for industrial spaces for this reason.
Funny you mention animals. I often wonder if home environments have become screaming dens of high-frequency noise for dogs and cats. I know that flourescents tend to osciallate around 40kHz, and I definitely have an LED bulb that whines for some reason. Not to mention cheap laptop power supplies that also whine over time.
Are all these cheap SMPS driving our pets insane? (or just annoying them? or neither?)
I wonder if I could rent some super-high-quality mic from a local sound reinforcement studio and look at a 20-60kHz FFT ... or do fancy mics not really go beyond human hearing?
High quality recording mics and even mics designed for acoustic measurements usually (and deliberately) only go up to 20-40 kHz though roll off more gently than human hearing. Usually there is a mechanical resonance that's tuned to get a flat frequency response up to a point rather than a wider but rolling off response. A few specialist mics get up to 100 kHz, or some sort of laser vibration measurement...
I don't know about the bulbs, but commercial LED drivers from companies like Meanwell typically let you select a range between 100Hz and 1000Hz for the PWM frequency.
I can tell when I am around these bulbs. It is the same with fluorescent lights for me. They make me (more) grumpy. The more expensive LED lights have full wave bridge rectifiers and capacitors with no flicker. I've considered just building my own for my next home.
Some LED lights/bulbs also incorporate phosphorescent materials and use UV LEDs to stimulate the phosphor which helps too. This is also one of the ways better colour reproduction can be achieved. This comes at a slight efficiency cost though, I think.
Most LED lights use the UV+phosphor technique as its cheaper than doing RGB (If the LED is yellowish when switched off, its using this technique).
The best flicker-free lights are ones that have low ripple constant current power supplies (somewhat expensive). LED lamps intended for use around rotating machinery used to all be this way until someone figured out that if the PWM frequency was wildly unstable it would prevent the problems with strobe lights around spinning things at a fraction of the cost.
Are there any statistics how many people perceive this flicker? I see them complain in online forums but I'm not sure if it's just a very loud small minority.
What does it take to make them not flicker? I.e. how much of a cost increase is required?
Based upon this article were the retail costs of a cheap flickering LED bulb is approaching the single dollar mark, the cost increase to use a proper AC/DC converter could be orders of magnitude more expensive (at a rough minimum).
Honestly it's starting to feel like a most cost effective long term solution is to build homes with a master AD/DC converter (or several depending on sq/ft) and simply wire 12v outlets and light fixtures. The energy savings of swapping all incandescents or compact florescence bulbs with cheap LEDs with cheaper power converters is starting to scare me knowing just how shoddy and fire prone some of these power converters can be.
I have been thinking about this for years! I think we need a new marketing term along with a independent organisation to push for a new LED Standards.
Let say it is called CLED,
And it requires the following features / spec before it could be called CLED. From PWM, Colour Correctness, Energy Efficiency, heat etc. The different between a low price light bulb and higher priced is relatively minimal , the difference in LEDs is massive. To be point they should not even be allowed into the market in the first place.
You don't need to buy Hue... Their 8W 800lm bulbs are €4 a piece in Europe, and a lot less in bulk.
So 4x more expensive at worst.
Philips' discount brand (Attralux) is even cheaper, I recently bought more of these 8W 800lm for less than €1 a piece. For now I cannot tell the difference.
True, however they are worth the price. I used various brands of LED bulbs for a long time (such as Cree), basically whatever Home Depot was selling. The failure rate was extremely high. In 2017 I switched to Philips LED bulbs, and just last night the first one failed. Overall they are worth the extra up front cost, they'll save you money in the long run.
I have three weighty Philips bulbs mentioned in the teardown as carriage lights on the exterior of my house. They've been on 24/7 for maybe seven years now.
That is just the stroboscopic effect. It’s an artifact of your measuring tool. I’m not saying that your lights aren’t flickering but your tool is introducing a lot to this system and so you can’t use it as evidence.
> an incandescent-bulb-lit house slowly pulses from light to dark
This sounds amazing to me, but also confusing. My understanding is that an incandescent bulb works entirely by heating up the filament. There should be no way that the filament is cooling down and heating up in anywhere close to the rate that it would require to get some kind of wagon-wheel effect, which is what it sounds like you're describing.
Ignoring 'complex power' concerns, the AC voltage driving a tungsten filament bulb is a sine wave at line frequency (60hz for the US). Because it is a sine wave, and because the tungsten filament is a resistor, the current through that resistor rises and falls in sync. with the AC sine wave. The current will be zero at the zero volt crossing point of the wave, and maximum at the peaks of the sinewave.
The varying current through the tungsten will result in a varying power consumption (varying at a 60hz rate). The varying power consumption will result in a small amount of varying temperature on the filament.
Now, the filament does not cool down instantly, so the result is its temperature will vary by a few degrees, but likely not enough to ever be perceptible to human eyes. But with a sensitive enough (and fast reacting enough) temperature probe, one could likely measure the temperature rise/fall of the filament that is synchronized with the line frequency.
Right, the wave filter effect caused by the thermal mass of the filament was my point.
I'm not sure you've shown that this would be visible to a housefly as a "slow pulse from light to dark," as the post I was replying to claimed. First of all, you said the power consumption will vary at a 60hz rate, but it seems to me it will vary at a 120hz rate, since the current will be at its max magnitude twice per cycle (as you say in the first paragraph). But more to the point, I don't know whether the temperature/lumens will decrease enough in 1/120th of a second to be visible to houseflies or anything else.
Trying to research the answer, I found this physics lab worksheet [1] from Pasco (the makers of sensors), for a high school lab measuring the output frequency of incandescent bulbs vs fluorescent bulbs, but without doing the lab I don't know the results.
The numbers here show a >25% difference for an example bulb.
Also, even ignoring how the cooling starts off at its fastest, with radiation scaled to the fourth power of temperature... incandescent lights switch off in about a tenth of a second. Even if that was completely linear, it would be very visible. Even a 1% flicker would be visible.
Huh, good find. I wish the poster had some sources, especially for that initial temperature graph, but there's no reason to doubt it. Thanks for the research.
Incandescent bulbs store thermal energy in the fillaments, lowpassing the signal in the same way as a capacitor would on an LED bulb. A 60Hz flicker would be perceptible and if incandescents had that problem nobody would use them.
Likely. As with dragonflies, there is a minimum speed below which flies don't perceive motion. In both cases, if you move very (very) slowly, you can approach almost to physical contact without provoking any reaction.
Anecdotally from my experience as a macro photographer, the threshold for flies is lower than for dragonflies; flies often startle in response to movements almost too tiny to perceive having made, while dragonflies are relatively easy, usually requiring only a few minutes to approach from a distance of a meter or so to the ~12cm minimum focus distance of my best macro lens. (It's hard to know; the passage of time isn't of much interest in the focused flow state that's required for this sort of activity.) It helps that disturbed dragonflies tend much more often to return to the same perch, but with some practice it's possible to make the entire approach without disturbing them at all. The only really tricky part about it is that, when you're moving slowly enough, they also tend to land on you, which can be somewhat distracting if you're ticklish.
In any case, the existence of a minimum rate of change for perception of motion suggests that flickering light below a certain frequency might well be perceived as strongly discontinuous.
(It also merits mention that my macro rig includes three very powerful flash heads mounted around the lens front element. I've never observed dragonflies, wasps, bees, flies, or spiders to react to these in any way, even when firing from a distance of six inches; the only reaction I've seen has been from fall webworm caterpillars, which displayed a communal defensive response, and that may have been as much due to the shadow I cast, or to the polistid wasps hunting nearby, as to my flashes firing. The wasps notably did not care at all about me or my flashes, especially after they also found the nest and busied themselves with its rapid depopulation.)
We used to catch them by the wing by slowly approaching. Interestingly, medium ones were easy, but the big ones (different species, much rarer) wouldn't let me approach at all.
Pigeons have a "refresh rate" of around 100 frames per second. It is believed that a bunch of smaller animals and even insects have similar or faster visual processing time, though I'm not aware of specific numbers.
I used to be able to detect the flicker by eye, or sometimes by waving my hand. But they pack so many into the demo display at the store that when I moved I had to use the slomo feature of my phone to pick the right LEDs.
It's pretty amazing how powerful small chip-on-board (COB) LED modules have become. I recently built a 6000 Lumen aquarium light with 7W - 850 Lumen COB modules. The active surface is just 5x5 mm, so the illuminance when held in front of your eye (1 cm² surface) is about 17.000.000 Lux, roughly 100 times brighter than direct sunlight on a summer's day at noon (so don't do it as it might be the last thing your see with that eye). And those modules are actually at the low end of the available power, you can get up to 35 W packaged into such a tiny COB, which is just crazy. Good thermal anchoring of the chips is therefore paramount, as they will quickly heat up to the point of being destroyed by electromigration if they're not properly cooled (and I imagine for the control electronics in the light bulb the high temperatures are also not beneficial). Anyway, it's still impressive how much light you can get out of such such tiny devices.
This has impacted the video production business in a positive way over the last couple years. Previously, we had to put on gloves, find a 30 amp fuse and use hot tungsten lights to put together a production. Now folks are using high-quality LED-based lighting, running off battery power, and significantly smaller than a generation ago.
850 lumens per 5x5mm square isn't overly spooky. It's only about 1/50 as bright as the sun itself, and comfortably under the level where you risk burning spots on your retina.
I'm not really sure what happens when you hold it right up to your eye. I'd be worried about the total amount of heat my retina can dissipate more than the lumens, but I have no idea what that number actually is.
It depends on the distance. Even at 1 m an 850 Lumen source with 120 degree radiation angle is painful to look at, and at 1 cm distance the illuminance is 10.000 times higher again. The peak illuminance of the sun at the surface of the earth is around 120.000 Lux (Lumen /m2), significantly smaller than such an LED when viewed from a short distance.
I just don't think illuminance is the most important number here. Luminance is more relevant to eye damage in the vast majority of cases, because that measures the spot intensity on your retina. After all, the sun's light would be completely harmless to the eye if spread over a five degree circle, despite having exactly the same illuminance. (Well, you would want a UV filter, but the brightness would be completely neutered.)
The sun gets up to 1.6 billion candela per meter squared. 850 lumens in a 120 degree cone, from a 5x5mm surface, is about 11 million candela per square meter if I did the math right. With a retinal danger level around 100 million, you'd need some very strong secondary effects to damage yourself with an 11 million source.
Secondary effects like boiling your entire eyeball, or overloading the cornea while the retina is safe. And I don't have the knowledge here to say if that happens, or if your body can easily sink a spread-out watt of light and nothing bad happens.
I don't think that's right. If I hold this LED right in front of my eye the retina will be exposed to the entire 850 Lumen. Sunlight with an illuminance of 120.000 Lux on the other hand yields around 12 Lumen on the retina, assuming a surface of 1 cm². It's the relative distance to the light source that makes this so dangerous.
Even if you only let a pinhole of sunlight in, enough to illuminate a surface to 120 instead of 120k, it will burn your retina.
Normal retinal spot burns are not about total light coming into your eye. They're about incoming light per solid angle. A light that is bright enough per area is dangerous at any distance, until it's so so far away it looks like a single point.
I'm sure that there's some level at which total light is damaging, even if there are no super bright spots. But that level is going to be significantly higher than 120k. It's maybe a range this LED can reach, maybe not.
If we can for a moment digress from light bulbs, I have had recently some experience with (el-cheapo) led panels, those that can be inserted in ceilings (false ceilings), they can be either round or square and are typically 12 W (around 17 cm diameter) and 18 W (around 22 cm in diameter), there are even larger ones (24 W/30 cm diameter).
They make a very good amount of light, like 1100 lumens for the 12 W and 1650 lumens for the 18 W, and have a separate "led driver", constant current, 280-300 mA, voltage 36-72 V for the 12/18 W driver and 54 - 96 V for the 18/24 W driver.
They can be found everywhere for anything between 4 and 10 Euro each.
It is now three years I installed some 60 of them (in a restaurant/hotel, think no less than 10 h/day on for 300 days/year) and 15-20% of them failed in the last few months.
In ALL of them the faulty part is the "driver" (which basically is a transformer and a rectifier plus a current limiting chip and a capacitor, which seemingly is the actual component that always fails[1]), spares (a new driver) can be luckily be found for 2 or 3 Euro each, the actual leds are perfect.
I cannot see why it is not in production a male/female E27 attachment containing a "driver" and a "driverless" bulb to match.
The amount of electronics thrown away would at least halve.
Does anyone know where to find actual good LED bulbs? All of the Feit bulbs from Costco I bought eventually died within two years. I tried buying Phillips (non hue) off Amazon, and they are also starting to die in about a year to a year and a half. They really try to sell LED bulbs as lasting "forever", but in reality they seem to die just as fast.
You might want to check your home for electrical issues (specifically spikes). I've yet to have a single LED light bulb die ever, some have already exceeded their supposed 10 year lifespan.
And that's not one brand. I have GE bulbs, Philips (Hue and not), and random brands from Amazon.
Were they selling cheap LED bulbs over 10 years ago? By my recollection they were all expensive back then. It seems plausible yours were expensive and higher quality than all the cheap ones everyone is buying nowadays.
Speaking for myself, I've bought a variety of brands over the years. Some brands seem to die after a year or so. Others are still going fine after 3-5 years.
Same here. I put 2700K philips all over my house 5 years ago and I haven't had any issues, but I have no idea of the quality of the power coming into my house from my provider.
(Anyone know how good Austin Energy's electricity on the west side is?)
Could also be fixture choice. My Feit can lights from Costco have had a couple failures so far, a few years after I bought them, but mostly when I have an LED bulb fail it's been in one of my enclosed fixtures. That's my choice, though, I take the risk. I buy decent quality bulbs, but I like them bright so I end up with 1600 lumen bulbs in an enclosed fixture, which is toasty. They only last a couple years before malfunctioning. Sometimes I can get more time out of them by moving them to table lamps.
I monitor the electric service at my house and it's very steady at 240V.
I replaced the mish mash of bulbs in my house with all GE "Reveal HD" LED bulbs of different temperatures. Nothing super fancy, just name brand hardware store bulbs.
In the variety of old bulbs, the Feit ones in particular were causing issues with dimmers and also getting slightly discolored and buzzy. Only one had outright died, but seemed that more were on their way.
All the new GE ones have been great so far (only a year), but I'll let you know in 10 more years.
Refreshing all the lighting in the house with modern, matching, actual-dimmable LED bulbs (along with nice dimmer switches) has been a really nice ambiance upgrade.
I've had good luck with Cree, Home Depot's brand. I don't think any of them have died since I converted about five years ago.
Lowe's is stocking GE these days, but I'm not familiar with their quality. The reason I'm not familiar is that they exclusively stocked Feit initially, and yes, those things are pure garbage. They had high failure rates and were not "instant on" compared to others. I stopped buying them and switched over to the Home Depot brand.
I've tried a few of the GE bulbs over the years, and the quality and CRI is always not that great (sometimes easy to tell by taking photos with a good camera and seeing how little color you can get out of certain channels). Cree bulbs have always been pretty good, and Phillips bulbs have been the best.
There are a few other brands you can get online that have better rendering (CRI), and are more accurate to the stated color temperature over time, but they cost sometimes double what even the Philips bulbs cost.
I also wanted to add a data point that I've switched every light in our house to LEDs over the past decade, and only had two bulbs (out of maybe 50+) fail in that time—both were in outdoor fixtures which range from 10-100% humidity, and -10°F to 110°F through the year (quite a torture test).
Just another speck of anecdata: I've had to replace almost all of my < 10 years old, <~10k hours of use Cree bulbs. The last 5 years of LEDs (feit, Cree, GE) haven't failed yet.
one other concern not mentioned by others, vibration. this can come from being too close to a door or being mounted on something with moving parts, ceiling fans and garage door openers.
Also tulip like lamp shades where the base is at the top can concentrate heat and wear the electronics down; canister mounts do the same if no venting.
> A fused neutral design would leave all of the bulb electronics hot. Strange.
Depends on what country you are living in.
It's not new for me to see a product passing safety certification in one country, failing in another, but still ending up in it because of messup by people running the OEM industry.
Second to it, in a normal country, for an electrical appliance, failing short, and triggering a short circuit protection is a relatively safe alternative to having a dedicated fail safe circuit, but in countries with building codes not saying anything about wiring safety, it often means a fire.
Honestly, I think it's better that we view polarization as a historical oddity. So many people don't understand the concept.
Even I have trouble understanding the real difference between ground and neutral, because neutral connects to ground in the breaker panel. I'm not even sure if neutral goes through the breaker?
Note, the below will be for the US -- rules will be different in other countries.
> Even I have trouble understanding the real difference between ground and neutral,
Neutral is the normal return path for current. Under normal operation power is delivered to a device via the "hot" conductor, and travels back from the device to the source over the neutral conductor.
The ground conductor is present for safety purposes. Under normal operation no current should flow over the ground conductor at all. The only time the ground should have a current flow is during a safety event (i.e. a short from hot to the case of a device with a metal case).
> I'm not even sure if neutral goes through the breaker?
It does not, the breaker is present in the "hot" conductor path, and tripping the breaker disconnects just the hot conductor from the power source.
1. It's to provide protection in case of neutral failure. If you have an appliance with the case grounded via neutral and the neutral wiring fails, you now have the case connected to live (through any appliance that is switched on). With separate ground, nothing happens as you are only allowed to connect ground to neutral at a point where failure of the upstream wiring is practically impossible (i.e., wires so thick that they practically won't break).
2. It's to provide a path for ground fault currents that does not pass through the RCD, if you have an RCD. While neutral doesn't go through the circuit breaker, it does go through the RCD, so as to measure the difference in current between live and neutral. If your appliance is grounded vial neutral, any ground fault (connection from live to the case) looks like regular load current to the RCD, if it is grounded via separate ground, the fault current bypasses the RCD and thus causes it to trigger, even for a small current that would not trigger over current protection.
Neutral exists separately from ground only because if there is some residual resistance in the line, there will be voltage on the neutral.
So you have a separate ground that normally sees no voltage at all.
You are correct, the neutral does not go through the breaker. It is usually wired inside the breaker panel though.
Some localities want the neutral and ground wired separately (to separate bus bars), with a single point connecting them, but most don't care and the neutral and ground are attached to the exact same place inside the breaker panel.
Electrically, the neutral and ground are at the same potential, assuming magic wires with zero resistance, but in the real world the wire does have resistance, especially if it was poorly installed, so they are not always at the same potential. There could also be capacitance in the line.
The neutral handles the expected load, while the ground is designed specifically to be an alternate current path from the case (assuming it is conductive) that has low resistance and will immediately flow enough current to trip the breaker in case the hot somehow comes into contact with the case.
As an amateur radio operator, I can hear the RFI generated from that cheap bulb just by looking at it. I upgraded my home from CFL to LED and found out the hard way that cheap IKEA bulbs are not conducive to a radio hobby. Philips Hue and Lifx have been generally much better.
I recently replaced a bunch of bulbs in my recording studio with LED and it was a huge mistake. They are basically RFI cannons. I removed them after realizing what was going on. I spent a couple of days hunting down a really nasty rogue noise and it ended up being a LIFX bulb I had forgotten about. Now I have to find a huge pile of black market incandescent bulbs...
Some designs eliminate the electrolytic capacitor. That's the first component to fail. If you want 10+ years of operation, it has to go. You can get rid of the electrolytic, but you have to add an inductor.[1]
The cheap bulb teardown linked to by indiantinker [1] shows that it contains a 100uf electrolytic cap for smoothing. A quick search shows that 6.3 V 100uf MLCC capacitors are readily available for ~20 cents. Maybe you can only afford that sort of thing starting in a "two dollar" bulb but it shouldn't require buying fancy $10 bulbs to ensure that the capacitors live as long as the LED.
Or are there other considerations that mean non-electrolytic capacitors are unsuitable?
This needs a 100V capacitor. 6.3V ceramic caps are easy to source, but not higher voltage devices. Multilayer ceramic capacitors for high voltages are still quite expensive.[1] Not available in surface mount, either.
Energy stored is voltage x capacitance, and capacitor size and cost scales according. You have to store enough energy to light the bulb for half a power line cycle in something.
If you used 3-phase power to light bulbs, no need for energy storage. 3-phase is always on; you just need a switching regulator to smooth out the ripple. Some large arena lights work that way. Not worth the trouble at home scale.
Well, for MLCC caps specifically, 6.3V is definitely not going to cut it (that's less than the voltage across 2 LEDs, and there will be more than that in series). You'll find that 100uf in 200V ceramic caps will be substantially more expensive ($10 - $100), plus you get fun effects like ceramic capacitors have less capacitance the higher the DC voltage across them.
Electrolytics are used when you have no choice, and if you are trying to store a large amount of charge in a small volume (especially on the cheap), you have precious little options.
Thanks! I was thinking of a single LED. Looking at other teardowns where the actual light emitting elements are shown, I see that bulbs contain multiple LEDs.
> A fused neutral design would leave all of the bulb electronics hot. Strange.
Makes me wonder if the resistor is really a fuse?
BTW, I recently bought a plug -> bulb adapter on Amazon. It did not preserve neutral at all, and instead had a higher plastic ridge making it impossible to touch the metal part of the bulb. (The part that's at risk of touch, so it's typically connected to neutral.)
I have often seen resistors used as fuses in cheap designs, but I have never seen a wirewound resistor used in this way. It is for current limiting to reduce peak load on the cheap components. Although it would probably be the next thing to blow if the IC failed short.
In my experience, the electronics _around_ the leds tend to fail rather fast.
Few years back - around 7 - we expanded our house slightly (A room, a small living room and a small bath) so I put led bulbs in there, plus I replaced some of the bulbs in the rest of the house.
As I bought several of these (5+) when I placed them I wrote on them with a permanent marker the date of installation. Bear in mind these were not extra cheap, more like mid priced ones.
6 months from installation I had to start replacing them, and at the 18 months mark I had replaced most of them.
Bear in mind this coincided with a big legislation in my country that made "normal" light bulbs illegal (Because ecology, global warming and shit).
I doubt they took into account how much trash these cheap bulbs generate.
Also, whatever money you save from less energy consumption is probably a lot less than the money that it costs replacing them this often.
Hijacking the comments. I have a project that would involve about 15+ smart Lightbulbs. Anyone would know any decent/cheaper alternative to the Phillips Hue bulbs?
z-wave "behind the switch" modules are super easy to install (you pull out the switch on the wall, and you put a small box in between the switch and the mains wiring), and you can control a whole circuit with one ~$40 switch. Depending on which one you get, they also support 3 and 4 way circuits (although with some caveats).
And it comes with the additional benefits that it will always work with the manual switch (turning it off or on), and it will work with any lightbulbs you may want to use in the future, but the downside is that you can't really get full RGB colors in the bulbs if that's something you are going for.
This is more or less my preferred solution. I use z-wave switches, not modules, though. I don't get RGB control, but I do get home automation that fails gracefully back to "works just like a light switch always did" if something goes wrong.
Can't speak about the quality of the "smart" parts in particular, but the Ikea LED bulbs have proven to be of generally high quality for a good price. I have a bunch of them in operation, and had some in permanent use (powered 24/7) for years now. Only a single bulb has failed on me after 4.5 years of continuous use, and even that bulb failed gracefully (it became darker and darker over time, while getting hotter and hotter, obviously converting more energy to heat than light).
Hence I would not hesitate to give their smart bulbs a try as well, assuming they use the same circuitry and LEDs for the "non-smarts" of the bulbs as for all the others.
One thing to consider with the Tradfri bulbs is that some people (myself included) have experienced them turning on randomly. Internet says it might have something to do with automatic software update, but that is hearsay.
Also, for a smart bulb I find their control requirements to be weird. Pair with this device first, but only up to X devices, then pair that with the hub. Everything else I have is Z-wave so maybe this is par for the course for smart bulbs and I'm just not used to it.
I think there's a way to pair them directly with the hub. I did it recently when I only bought a hub and outlet (no remote)[1], and it worked fine. One rumour I've seen a few times is that the Hub was an afterthought[2], which is why it is designed around pairing with the remote, but it's definitely a weird setup (they're ZigBee, so should work without the remote - they do when using the Hue hub).
If you need discrete bulbs (vs a strip) the Merkury brand sold at Walmart have been working great for me. $20 for 2 full color lightbulbs vs $50 for 1 Philips Hue.
They operate on 2.4ghz wifi and after inspecting the traffic do not appear to be funneling my data to China.
Do you have to control the bulbs individually?
Do you need to address (talk to) the bulbs individually?
Are the bulbs far apart or close one to another?
I'm glad I found this and of course I'm familiar with Big Clive (I watch his stuff regularly), EEVBlog and others mentioned here.I've had mixed success with low-cost LED lighting and often discuss that and related subjects on my own blog at https://tech.scargill.net as well as reading what others have to say. I get a lot of feedback on that subject.
I note comment below about Poundland. Related to that (I guess) is PoundStretcher and up to now, the highest failure rate I've had in LED mains lighting are the bargain LED lamps (Hitachi) from Poundstretcher (and I get samples or buy LED lights from China, Europe and various locations in the UK so I'm no stranger to the subject).
Not knowing much about less, are any of these cheap ones just as good as the premium ones? I'd also like a bulb that could be changed to an orangish hue at night to block out blue light. From what I read a lot of them emit little light when changed to orange.
They activate 1-4 LEDs in series depending on where in the half wave of the 120V AC we are. That's understandable, but I don't get how about 60V is okay for one LED?
One LED has a forward current of max. 3V, doesn't it? By partitioning the up to 120V into 4 phases, it does not cover a mere 4 LEDs in series.
Do they have to put 15 in series for the first section in the schematics and another 15 for the thign called "LED 2", etc.?
Are there LEDs with forward currents of 60V available?
Dave at EEVblog just toredown a BenQ freebie monitor nightlight the other day, maybe yesterday IIRC.
Also, Bigclive as mentioned. The hotdogger and USB charger with shoddy transformers passing 240V should make any sane person to realize it's cheaper to stick to name-brand USB chargers.
Disclaimer: Many moons ago, I was literally tearing-apart clocks and electronics when I was 5. Too bad I didn't have a Youtube channel. ;-) I think it's genetic as my father and grandfather both built a bunch of HeathKit projects.
I don't think this is the design for the next hundred years as the author claims. Instead, screw-in bulbs based on a form factor dictated by 19th century glass making techniques and incandescent lighting will probably die out.
They'll surely be available as a specialty item, but cramming a power supply, emitters, and heatsink into such a suboptimal package only makes sense for legacy compatibility.
Integrating a non-replaceable lamp is the trend. You buy the whole unit, it works for say, ten years and then you buy a different one when remodelling. This made no sense when a light bulb was just something that needs replacing every few months in normal use but LEDs mean you can engineer that out.
Replacing a "light bulb" will seem as weird as replacing the backlight inside your laptop. Obviously a technician can do it, might even make sense to pay for that rather than buy a replacement for the whole unit, but it's not something regular people do themselves.
New wiring standards are not necessary. The most popular successor is as tialaramex described, an integrated lamp hardwired into AC power.
For more sophisticated users, AC/DC power conversion, power regulation/dimming, and the LEDs themselves can all be separate components. This allows more customization, and the possibility of replacing individual components when something fails.
Personally, I've been using Viltrox brand photo/video lights for interior lighting lately. These combine two sets of LEDs in different color temperatures with dimming/regulation, and an external AC to DC power supply. These have much better color rendering than most consumer stuff, and I like having adjustable color temperature. The bigger ones have remote controls, so hanging one from the ceiling is possible while retaining access to all of its functions.
> A fuse should be put on the hot leg as close to the power entry as possible. The reasoning being that if the fuse pops it removes the hot voltage from as much of the assembly as possible.
This only works if the power system has a defined neutral and live wire. European Schuko plugs do not - as the plugs are reversible - so you need to break both wires to safely cut power to the device.
It would be better to run a dc circuit throughout a house just for lights and do away with the rectifier circuits inside these led globes and It could be free of transients.
I'm all for DC supplies in houses, but it doesn't quite solve the issue here.
Any DC house-wide bus would see voltage drop along the wires, so all connected devices would have to handle an acceptable voltage range.
Since LED's have a very steep IV curve, they really need either current control or a very big resistor to handle that voltage variation.
Very big resistor = too much heat.
So current control. And the only realistic way to achieve that is with a switched mode power supply.
There is one benefit left for DC: The switched mode power supply could have much smaller input and output capacitors, and operate at much higher frequencies, therefore being much smaller.
It solves the problem of your power disappearing every eight milliseconds.
No rectifier, no higher voltages, the ability to use capacitors that are a thousand times smaller.
And even if you did use a resistor, it wouldn't have to be too wasteful. If you use normal 14 gauge wiring, your resistance is probably half an ohm. At 48 volts, with a dozen 60-watt equivalent bulbs, your voltage drop is less than one volt.
Has anyone here done any rigorous testing wrt flicker in particular? Of the reasonably priced brands (which excludes Hue), what's the lowest-flickering warm white bulb?
No, because rectifiers are cheap commodity items, and you can wire all the LED lights in a house on a single 15A circuit and #14 wire (~1400 watts). You can also put ~140 32w 2’x4’ LED fixtures on a 277v 20A circuit using #12 wire. It’s cheaper and easier to convert at the fixture or lamp and just use standard wire for everything.
If there was anything to be gained from doing it that way, it would already be done in commercial build outs.
Low voltage requires low resistance to transmit reasonable current. The only way to get this is with substantially thicker wiring.
5V DC isn't a plausible voltage to send the distance from a fusebox to light installations. One could do 48V and step it down at the site, but there's not much in the way of gains to be had. AC/DC circuits are quite efficient as-is.
> PoE standard provides up to 15.4 W of DC power (minimum 44 V DC and 350 mA[2][3]) on each port.[4] Only 12.95 W is assured to be available at the powered device as some power dissipates in the cable.
84% efficiency sucks. It's a reasonable solution for some cases, because if you're running Ethernet to a camera already, hey, why not provide some power?
But for, let's say 300W of lighting that runs 12 hours a day, you're wasting about 15kWh of electricity per month. Residential wiring has internal losses under 1%.
PoE is popular because electricians are expensive. That said I think there is a lot of potential to use the newer higher power PoE standards or similar for smart houses. The newer standard go up to 100W.
One thing about efficiency is low voltage DC/DC converters are always fighting for efficiency. It's much less of a problem for higher voltage DC/DC's. This is due to fixed voltage drops and I squared R losses.
You'll probably like the product teardowns of youtube user AvE. He usually does tools, but also the random appliance or product from time to time. His teardown of the Juicero is a classic: https://www.youtube.com/watch?v=_Cp-BGQfpHQ
I'll take my "internet-connected abomination" over having to wire in a dimmer switch that might change the tone of the light as you dim, make buzzing sounds, cut out the light early, or shorten the lifespan of the bulb.
Hue has been the first dimming system that actually works well. I have a few colored bulbs, but rarely need them, the biggest benefit to me (aside from automation: dusk auto-on/off) is dimming different shades of white/yellow. Great system that has no real analogue competition.
