Dump your fluorescents and incandescents for this amazing new LED bulb | Hacker News Reader

Dump your fluorescents and incandescents for this amazing new LED bulb

slate.com

96 pointsrobg105 comments

This kind of casual evaluation by a non-expert observer is incredibly meaningless. LEDs have a radically different spectrum from "regular white light". This shows up in weird looking colors when this light is used in real settings; the fact that the light "looks white" through the lampshade doesn't mean that it won't turn your off-white walls an ugly color, and your sofa, and the skin of your family members.

Saying that you "can't tell the difference" when casually looking at the two lamps, during the day in all likelihood, is something that's been done before with a million CFL and LED lamps. It usually doesn't actually hold up. In this case, there doesn't seem to be anything special about the actual LEDs - they just put many of them into a screw-in enclosure, for which they need liquid cooling to remove the heat.

If this was about actually saving electricity, and not about "environment theater" and a windfall for a few companies, they'd pass a steep progressive tax on residential electricity use, and let people figure out whether they want to have more efficient bulbs or to turn down their air conditioner, or perhaps just not build a McMansion which needs 1000 lamps to light it in the first place.

To be fair to the environmentalists who back this legislation: they almost certainly want to pass higher progressive taxes on electricity use. As the current budget debate is showing, it is all but impossible to raise taxes in this political environment. But also, a generic tax on electricity would probably never result in more efficient lightbulbs as they just don't take that up much electricity on a per house basis. People don't change their behavior over $5-20 a month in electricity bills. It's only at a larger scale that you notice the waste.

If this were purely a windfall for cfl manufacturers, the law would mandate cfl's. It mandates a minimum efficiency and lets the markets figure it out. If someone builds a better bulb, it will win out. If the casual observation of this article proves correct, this particular bulb will win out. If not, it is only a matter of time before someone builds a bulb that will. The mandate all but guarantees it.

First, to settle the question about this particular lamp: I found some specs on their website, and they claim a CRI (color rendition index) of 65 for their "neutral" version and 85 for the "warm" version. This is poorer than most modern fluorescents (which people complain about), and the better CRI for the warm version is at the expense of making the light yellow, which many people also don't like. So we're not looking at the at the coming of the savior of efficient light bulbs.

For the question of the law, I agree that the savings in electricity are barely measurable, which is why I think it's theater. The huge growth in residential electricity use has been driven by increase in average home size, and by the explosion in the use of air conditioning. Also, it's not even clear that bulbs like this save energy over their lifetime. Sure, one of these theoretically replaces 20 incandescents and uses less electricity, but an incandescent light bulb is just a tiny tungsten wire + steel leads + glass + aluminium base + a drop of epoxy, all made in one factory with extraordinary efficiency that has been polished through 100 years of manufacturing. This thing has 20 LEDs + mounts + cooling tubes + cooling fins on the outside + "non-toxic" liquid + a bunch of electronics for control and dimming + enclosures and wiring for all of the above, made out of plastic and ceramic and glass. All manufactured in probably 20 factories and transported to be assembled in yet another place. If there is a win here for the environment, it's super-tiny.

If there is a win here for the environment, it's super-tiny.

Presumably the higher environmental costs of the LED bulbs are reflected in their higher retail prices, which are still offset by the savings in electricity (the price of which should eventually come to reflect the environmental cost of generating the electricity).

LEDs have a radically different spectrum from "regular white light"

So do incandescents. The good thing about LEDs is that they're easy to tune; they're becoming very popular in the film world because you can make small fixtures that run a long time and that allow a cinematographer to dial in the desired color (via a multicolored array) instead of faffing around with sheets of colored gel, and they are capable of much better spectral power distribution than CFLs. It's not just the tinting and tunability of the LEDs themselves, but also the fact that they put out a steady light instead of CFLs' high-frequency flicker. Even when the color is slightly off from the ideal, the steady light source is much, much easier on the eyes.

I'm not ready to endorse these products yet, because their spec sheet just gives a single figure for color temperature. But I'm going to pick one up soon both because it looks cool and because the CFLs I have at home produce really ugly light, so it can hardly be worse. The first thing I'll be doing after ooh-ing and aah-ing over it's retro-modernist design is lining it up against other bulbs and shooting it against calibrated gray cards; I have the gear and skills to measure these things properly. If they're spectacularly good (or bad) I'll write it up.

EDIT: BTW LEDs can flicker too. It's probably more accurate to say fluorescent lights have a low frequency flicker since it's often slow enough to be easily seen by eye, especially if you use them with a dimmer switch - in which case the flicker is slow enough to interfere with film shutter speeds, CRTs and so on, giving rise to an unpleasant strobing look. I didn't mean to say that LEDs are completely immune from this, but I have found them vastly easier to work with than fluorescents in a wide variety of situations.

>> LEDs have a radically different spectrum from "regular white light"

> So do incandescents.

What do you mean? Incandescents have essentially the perfect black-body spectrum (the effects of the spectrum of tungsten are insignificant). This is what's called a CRI of 100. If you want a bulb with essentially a daylight look, you can get a halogen incandescent, and as a bonus, it's 30% more efficient. LEDs and CFLs don't come near this.

> The good thing about LEDs is that they're easy to tune; they're becoming very popular in the film world

The problem is that people can't tune the color filtering in each of their rooms with the expertise of a Hollywood cinematographer.

If you can test this product, great, but unfortunately, HN doesn't lend itself to this, since the thread will be long dead by the time you have results. I wouldn't expect anything amazing, anyway, since the CRI is not fantastic and the LEDs appear to be standard jobs.

'Regular white light' I take to mean a daylight spectrum, not the the warm orange of incandescents, sunset-like and pleasing though it may be. Halogen lights take time to warm up and their color spectrum shifts along the way. I like them, but they don't deliver a real daylight look. Any time I've ever shot at night using a HMI people's skin looks purplish unless you add in some incandescent light to warm it up. I am not saying that LEDs magically solve this problem, but that their limitations as stated above are really not that bad.

The problem is that people can't tune the color filtering in each of their rooms with the expertise of a Hollywood cinematographer.

It's an awful lot easier to do now. For old-style cinematography you were basically dealing with bulbs of a fixed color temperature, gels, and a dimmer switch. Building the lighting setup so that it looks good on camera was a very complex task - in fact that's one reason that the title 'cinematographer' got upgraded to DP/Director of Photography. (There's a joke: why don't more DPs smoke? Because it takes them 3 hours to light anything.)

Now with an LED array, you can basically just twiddle a few knobs and get the color you like. If you're shooting video it's especially easy, because you can just hold up a color card and match it against some other scene or even have a computer do the matching for you. But even with film, you can still shoot a polaroid and get the color matching done a lot faster on a small-to-medium set with LEDs. And a lot more cheaply, too, because everyone knows that LEDs are commodity components. If someone wants to do this at home with mood lighting (say, recessed strips around the ceiling with 3 tracks of red, green and blue LEDs connected to a dimmer - something that can be done for a few hundred bucks)...well, it ain't that hard.

As for testing, I can simply start another thread about it if the results are sufficiently interesting.

Hold on, HMIs are not incandescents, they are an arc-discharge lamp, and they have all the spectrum problems of that technology. They are probably some of the worst offenders in terms of peaky spectrum, so it's not surprising you have problems.

Anyway, I am not talking about cinematography; incandescents, whether regular or halogen, have an even, non-peaky spectrum, and don't freakishly exaggerate or mute different colors. This is what people want for their homes. The idea that we can regain this by installing banks of LEDs of different spectrum and fiddling with controls doesn't seem realistic. And psychologically, I don't want a huge continuous space of choices for my lighting, since it's impossible to settle on just one thing (there were studies about this). Again, I am talking about home lighting here, not professional videography.

Not to mention that by the time we finish installing the controls and the wiring and the LEDs and the electronics in every room, the environment will be at a huge disadvantage.

Hold on, HMIs are not incandescents

You are quite right, my mistake. I was, quite incorrectly, thinking of the halide mixed with the mercury vapor in a HMI, as opposed to the halogen around a tungsten filament in a worklight or so. The chemistry and mode of operation is so different as to be meaningless in this context, and I apologize for derailing the conversation.

It's not that I expect people to install banks of multicolored lights and tweak them endlessly, but that I disagree with you about just how peaky the LEDs are compared to CFLs. Cinematographers are hyper-picky about light and color (or should be) because the realities of both film and digital compression make it expensive and difficult to overcome a poor lighting choice after shooting has ended. LEDs have delivered considerably more convenience and predictability for small-scale lighting situations than anyone had dared hope for a few years earlier, and even when they require correction the output is consistent enough to keep the problem manageable. In short, they solve existing problems to a much greater extent than they create new ones. They are not so peaky in practice as to interfere significantly with skin tones, clothes, makeup or hair color in realistic-looking contexts (as opposed to sci-fi or 'stagey' lighting setups). That bodes well for their deployment in more ordinary settings such as businesses and homes. It seems to me that you're attaching too much important to the raw CRI score, which is after all an abstraction of a rather nonlinear curve (http://www.ledsmagazine.com/features/2/5/4/VLambda).

Whether by mixing or doping, ISTM that the ability to get a desirable result under demanding conditions with relative ease means that providing a broadly acceptable standardized product for the general consumer is a lot more feasible. Once again I am not claiming it's perfect. There are still problems, especially for film where you can't tweak the color balance during production, just as you can't tweak the colors of paint, wallpaper or furniture in the home. There's a comprehensive set of Macbeth charts at the AMPAS lighting project pages: http://www.oscars.org/science-technology/council/projects/ss... and http://www.oscars.org/science-technology/council/projects/ss... Despite the peakiness and pink tinge of the the multi-emitter 2 sample, and the slight darkening of the blended phosphor, you can see that they are quite close to the incandescent alternative on a range of colors. This is what I'm talking about: if you're concerned with perfect color fidelity - as you should be if you are, say, shooting a commercial that uses a well-known logo, or trying to provide accurate representations of nature, textiles, or artworks - then there are still shortcomings. If you do not need forensic-level quality, and most of the time you don't, then LEDs are an increasingly viable substitute. The biggest problem for the Academy is one of scale; what works in a medium close-up in an average size room is one thing, but when you get onto a soundstage (which you would use 5kw or 10kw incandescent lights to illuminate), then you have a whole different set of problems, and the deficiencies of LEDs are significantly magnified. That's not an issue for consumers, though, since few of them have floodlights installed :)

I'm not invested in any LED lighting companies, I'm just telling you that I've had very good experiences with the technology so far. Given the fact of the regulatory changes, which I am neutral about, LEDs seem much more likely to deliver the desired results than CFLs.

I am not invested into any lighting companies either. Maybe I should be, I am sure they stand to make a fortune on mandated fancy technologies. I am not sure why you think I advocate CFLs. I advocate letting people keep using incandescent bulbs, since they are the best residential lighting technology available, and will be for a long time. They are also very environmentally friendly, once you account for their use of only the simplest materials and manufacturing techniques. And you don't need to put up new factories.

LEDs are a very expensive technology. People in this thread are getting sticker shock from the $20 LED light bulb. I am shocked that they claim to sell it for this little. IMO they will quickly go bust. $20 is roughly the price of all those LEDs in quantity, unless I am guessing wrong about what they use, and they also need a powerful and compact control system to step down from 120 volts and control the LEDs, plus the liquid cooling tubes, plus all the manufacturing to get the LEDs and electronics in that package. Put this kind of technology in a flashlight, and it's going to be a $400 flashlight at retail very easily. And you don't get Moore's law to make it all cheap in a couple of years; these are power electronics, you can't just shrink them to 22nm.

All this, and the consumer state-of-the-art in white LEDs is a fairly nasty blue LED with a yellow phosphor to add some longer wavelengths and make it look kind of white. This is what Switch uses, and it's not a good light for homes. Even fancier and better stuff is on the way, but it's ridiculous to force people to go that kind of expense to get something which is still not as good as what they have. And don't get me started on the failure modes of LEDs - they don't burn out like light bulbs, but they grow dimmer and shift their color quite noticeably, and the electronics can start to buzz or flicker or, better yet, burn.

I understand wherre you are coming from, and why you support sticking with incandescent bulbs. I just mentioned that I wasn't affiliated with any lighting companies so you would know i wasn't pushing an agenda of some sort. I agree that the ecological/net energy benefits of alternative lighting are questionable; on a business scale there's a stronger argument, although businesses have been using cheap ugly lighting for years to keep their bills down. But since the incandescent phase-out is happening anyway and opposing it is likely a waste of time, LEDs look like the best of the available alternatives.

Incandescents don't have a "perfect" black-body spectrum, by far. Neither does the sun; both of which have 100 CRI. The CRI metric was also devised around the incandescent lamp to compare them against fluorescent lamps - thus the fact that an incandescent bulb has a 100 CRI has very little meaning in a broad context.

If you want to compare a light source to the black body spectrum you have to look at their locations on a color chart like the CIE charts - and compare their distance to the black body line. This is called Duv. With binned LEDs we can get a light source closer to black body color than fluorescents, and almost as close as incandescents.

If we are going to go into detail, then let's mention that the fact that you've matched the black-body line in the CIE color space does not mean that you've matched the black-body spectrum. Which means that you can have a light that looks "white" to an observer through a lampshade, but which distorts the colors of the surroundings (since nobody matches those to where they need to be in CIE 1931). This is exactly the problem with casual observation - "light looks the same" is not enough.

Anyway, all I meant to say is that the smooth spectrum of incandescents is what people like to light their homes. It allows natural perception of colors of the surroundings. The high-tech approaches to getting there are a) not as good, b) expensive, c) not so good for the environment, once you add it all up. Now, for specialized applications, I am sure that LEDs are wonderful compared to complex filters on incandescents, but that's not the product we are looking at.

One of the main questions is how wide/broad spectrum a light source is. CFL's, I believe, have strong & narrow emission bands[1]. LED's tend to be a bit fatter (also shown in link[1]).

I adore NIST's "Spectrally tunable lighting facility"[2]: a room with a huge variety of different narrow-ish-band LED's. You talked about a tuneable RGB array for selecting color, this is the same idea but comprised of 22 different colors, giving way better spectral control. Humans might not know the difference when they're looking at grey cards, but when it comes to seeing how different spectrums of light interact with real objects, I really like this idea of a tunable light source.

[1] http://web.ncf.ca/jim/misc/cfl/ [2] http://www.nist.gov/pml/div685/grp05/vision_lighting.cfm

The picture at link [1] is correct, but it makes the LED look way, way better than it is. It does show some light smeared through the spectrum, unlike with the CFLs, but it's not actually even in intensity, and has some sharp peaks (which you can't see in a test like this). Many people find LED light annoying, even more so than CFLs, when it's the primary light source. Of course, that's totally subjective, and some people even like the effect.

I've always wanted to build something like your NIST reference. I'd like to have a lighting system in my home that allows me to select color temperature in addition to intensity (as well as some other less-obvious parameters I can think of), without sacrificing CRI and subjective light quality.

But incandescents flicker too, even if they don't dim all the way. Google says that modern CFLs should all have electronic ballasts that first rectify the current and then run at over 10kHz. Maybe you've only experienced cheap ones with shoddy ballast?

Touring concert lighting designer here (when I'm not coding for money, anyway).

The above is absolutely right; LEDs produce a fairly narrow range of colors. If you look at a spectrograph of an incandescent light bulb (including halogen, which is a bit less yellow), you'll see a fairly flat curve that's a bit higher at the red-yellow (and infrared) end and a bit lower at the blue-violet end. If you look at a spectrograph of an LED, you'll see several narrow spikes. If you illuminate objects with LEDs, you'll find that certain colors appear almost cartoonishly vivid, while others that might look similar under incandescent or sunlight appear very dull.

An increasing number of small-medium concert venues are switching to lighting that's primarily LED-based, which I find very frustrating since there's not much I can do about the problems it creates. If all the LEDs had the same spectrographs, I could probably convince the performers to wear things that look good under LED light. They don't, of course.

Guys, this article is not about the existing LED light bulbs on the market today. It's a fluff piece about a very promising new bulb technology. The new bulb uses an unusual LED immersed in a "non-toxic" liquid. In theory it should have vastly better color temperature / light quality than existing LED bulbs.

I'd be happy to buy these things at $20 -- have my cake (carbon footprint) and eat it (light quality, no mercury release on disposal) -- but they don't exist yet. All the LED bulbs I can buy today are basically purple, like the early CFLs. Maybe it's fine for lighting a closet, but they're not adequate to night-time illumination of my home.

Call me when Switch bulbs are actually on the shelf!

They aren't usually purple, they are usually a sickly greenish.

The liquid cooling technology is what enables so many LEDs to be inside a small enclosure. I don't see how that improves the spectrum. They company doesn't even seem to make any claims along those lines (measurable claims that is - they do say "everything will look great"), they just say they meet certain color consistency standards for LEDs. As do the lamps you bought, probably.

I have a couple of LED bulbs and they are not purple, they are a very "harsh" (I actually prefer it) blueish white much like the neutral white CFL's I have purchased and use.

I hope so, because compact fluorescents, at least the cheap Chinese ones we get in the US, are awful.

I have no idea why I don't hear more about this, but I can smell a horrible chlorine-plastic smell from them after they've been running for a couple of hours. It is an acrid smell that makes my nose feel strange. I've tried many different brands - Sylvania, GE, Feit electric, Great Value - which are all rebranded Chinese manufacturer products, of course. I used to be able to find some bulbs that did not do this, but not any longer.

I'm pretty sure these fumes are toxic from how they make me feel. If you've never noticed it, go buy 4 Sylvania CFLs and run them in a closed room overnight. The room will smell like chlorine in the morning.

Does anyone know how they hold up in uneven power situations? As far as I can tell the electronic ballast is the number one reason CFLs fail so frequently, and I've seen plain old incandescents outlast CFLs in many shop conditions.

If LED bulbs actually last the 30 years, they're worth the extra in convenience.

The power supply is definitely the critical limiting factor in LED light bulbs. The thing is that to make it rugged for uneven power situations - as you say - you have to use large electrolytic capacitors, which then have a reduced lifetime of their own (and sometimes contain Polychlorinated biphenyls). These limit the lifespan to ~25000 or so hours and burn out while the LED array is still in good shape.

Compared to CFL's, the circuitry is less extreme - LEDs are low voltage devices unlike mercury arcs - and so the driver can definitely last much longer and be more reliable than a comparable CFL ballast.

But then you have a bunch of LEDs to play with in your hardware hacking! Clear win over CFL and incandescent bulbs.

LED & CFL are very much opposite situations for powering: LED uses very low voltage DC at high currents, and CFL uses high frequency high voltage low current. LED's have to be current limited.

I recently picked up a 2V 100A power supply off ebay for some LED projects. :)

I'm seeing more LED's in automotive use like tail lights, so that's a good sign. Seems like they've nailed down a stable driver that can handle voltage fluctuations, like on starting.

AFAIK, it only takes a zener diode and a resistor to drive LEDs in an automotive application (or any automotive electronic device that runs on < 12V DC). The power is already rectified by the time it leaves the alternator. Worst case is that they're a bit dimmer than normal on startup.

Many car LED tail lights are actually refreshed at (I'm guessing) 30Hz or so to keep from overloading the wire - They have a distinct look to them from the low refresh rate.

Those low-hz LED tail lights drive me nuts. The strobing effects are incredibly distracting and headache-inducing.

While it's certainly possible to do so, I doubt any engineer worth his salt would actually use a zener and resistor as a stable power supply for automotive lighting. Voltage in cars varies from pretty low (sub-8 volts or so) when the starter is cranking, to around 14-16 volts when alternator is running against a fully charged battery, to the occasional high voltage peaks of 40V+ when, e.g., the air conditioner clutch disengages.

That's a pretty large range to run a current-driven device (the LED) from a fixed resistor. Most useful applications these days will use constant-current supplies that feed the LED a fixed amount of current as the input voltage changes all over the place.

Does anyone else find many of these LED tail light designs too bright? I find myself squinting or outright having to look away from them, particularly when immediately behind the vehicle at an intersection.

I've seen a few designs that do not overwhelm in luminosity, but I find most of them to be an outright hazard by partially blinding those in following vehicles.

Voltage fluctuations don't matter since they're driven by a switching constant current supply. The voltage drop across one high brightness LED is only a few volts.

another comparison to CFLs. the mercury arc in a CFL is struck between two electrodes, and every time the light is switched on the electrodes degrade. So if you are energy conscious and only have the light on a few hours at a time - you will burn through the CFL much faster. This isn't the case with LEDs - so if you have a light that you use 1 hour a day, you could definitely stretch it out to 30 years.

CFL induction lamps might eventually resolve this, but at least for now they're priced out of the general market

Induction lamps are getting big - but most people want to get rid of the mercury vapor arc entirely and go for solid state, that's a big enough advantage that energy efficiency almost doesn't matter.

Unfortunately, Switch didn't figure out that the light is no longer "solid state" when they added liquid to it - and that a non-toxic liquid is probably worse in many situations than a toxic vapor.

I live in a motorhome and use LED lighting all over the house. This is, perhaps, the least even power environment I can imagine. They're running off of batteries, which ebb and rise throughout the day based on whether the sun is shining (I have solar panels), I'm plugged into shore power (which causes the battery charger to cycle off and on through the day and night), or I'm driving (the alternator is charging the batteries).

I don't know that they're going to last forever, but I do know my remaining incandescents die about every 6-12 months, and none of the LEDs have burned out. Because LEDs are so expensive ($20 per bulb vs. $1 for the old style) I only use them for the lights I leave on for some length of time, so I still have enough incandescents to make a comparison. So, the LEDs are working harder and not burning out as fast as the incandescents.

I recently bought an LED that has a much more pleasant color than some of the older ones. I don't know exactly what the difference is...they all have yellow tinted covers, but the new one is a more neutral light. Still blueish, but not as harsh. So, I guess they're all getting better at a pretty rapid clip.

I live on a boat and have virtually the same electric/bulb set up. My solar panels also feed a battery bank, and the light bulbs that get used a lot have been switched to LEDs, with the remaining bulbs still incandescent.

A couple of my "soft white" (aka yellowish) LEDs seem to be pretty close in color to incandecents, but a couple have much more of a bluish tint to them. I got an excellent price on the bulbs, so I figure this variance is part of the bargain.

Anyway, high five for living more off the grid!

The low voltage, direct DC environment that you have is far less harsh on electronics than built in AC-DC transformers with much higher voltages, variances and surges. Batteries usually don't peak, their voltage just drops off, which most electronics can deal with just fine. You also have a battery between your electrical system and your solar charger, which smooths things out considerably.

While it's true that batteries act as a nice big filter, the fact is that automotive DC voltage ranges between 9 and 18 volts, depending on what's providing the power, and extremely noisy. I don't know exactly how it all works, but I know when the generator kicks in the lights go from "ok" to "POW! Let's get some lights on in this mofo!"

I'm pretty sure it's a hostile environment...which I'm guessing is why the incandescents die at such an alarming pace. I've had some last as little as a month or two.

They're good. I've used them on film sets and with dimmers, and they're way easier to work with than fluorescent lights. Also, you don't have that annoying hum from the ballast all the time. If these new bulbs make the grade for domestic lighting then I'll probably put them all over the house.

Film and video have the benefit of being primarily sensitive to specific wavelengths (likely red/green/blue). It seems the problem with LEDs is that certain surfaces are not very reflective in those wavelengths. I have a somewhat similar problem with the CFLs in my kitchen: they put out too much orange light, making things like raw meat, cheese, and orange juice look decidedly odd. I wouldn't be surprised if they look identical to equally-luminous incandescents on camera.

They've got three main shortcomings

Four. They can cause fires. Consumer Reports this month had a recall of two brands that would catch fire. I have had several (not of the bad brands) do weird things when they burn out. In each case, they "smoked". It was unsettling, because I didn't know if the type of event could have led to a fire. Now I know.

Ive led bulbs fail almost immediately, by erratically going from too-bright to barely-on, finally settling in barely-on. No idea what the deal was.

Philips went a much simple way. They just painted the bulb enclosure yellow, and it works really well. The color gets very close to that of incandescent. The only issue is its price. It is very hard to bring oneself to pay $49 for a light bulb. Economics or not.

http://www.homedepot.ca/wcsstore/HomeDepotCanada/images/cata...

it's not exactly a yellow paint, it's a phosphor doping they embedded in the plastic and it's the reason the bulb costs so much. Theirs is very high tech but I don't think it can compete in the long run due to complexity and the costs associated with it.

"... it's a phosphor doping they embedded in the plastic and it's the reason the bulb costs so much."

It costs so much because if something cannot compete on cost, it has to compete on premium features at a higher price point.

For more on the physics behind "white" LED's, there's a pretty good summary at:

http://en.wikipedia.org/wiki/Light-emitting_diode#White_ligh...

The fundamental challenge is that the wavelength of light emitted from the LED is constrained by the band gap structure of the diode. To get white-looking light, you need more than just one narrow frequency band.

For instance, until some new materials were developed in the 1990s, this is why there were no blue LEDs, only red, green, and amber.

Bob Johnstone's We Were Burning has a fascinating look at the development of the first blue laser diodes.

http://www.amazon.com/We-Were-Burning-Entrepreneurs-Electron...

Why don't they just paint it then, or put in some simpler kind of yellow plastic?

The idea is that the yellow phosphor converts a short wavelength blue light from the LED into a longer wavelength yellowish light, called down-converting.

This is a special physical property of the phosphor material - not simply a function of its color.

Normally this yellow phosphor material is applied directly to the LED die - but they figured they would put it on a plastic shell outside the LEDs. It doesn't actually make any kind of performance improvement except that the light is more evenly distributed.

If you tried to paint the shell with the phosphor - you would most likely have blue light shining through it - that's why they need so much technology to accurately dope the plastic phosphor, so that the light output is an even and consistent color.

The LEDs are typically blue with a phosphor that is energized by the LED light. It's the phosphor layer that starts to glow and produce the white light.

If you "paint" a surface a particular color to alter the light transmitted through it, you're effectively creating a filter. If you want something to appear more yellow, you need to increase the amount of yellow wavelengths transmitted through the filter by reducing the other wavelengths. But if you have a single color LED, such as blue, there are no other wavelengths present (at least not in any significant amount). So if you put a yellow filter over a blue LED, you see nothing...

I have this bulb. The phosphor has one killer feature: it gives a much broader emission pattern than any LED+reflectors bulb on the market. Every other LED bulb I've tried has been too directional to use for general lighting.

This is exactly what I have wanted for years. I'm willing to bet $100 on a pair of them, just to see how it works out.

I bought one of these as an experiment. I've been trying out LED bulbs for a while, and this is the first one that's worked well enough to be a viable CFL replacement. Unlike others I've tried, this bulb has great color temperature, brightness, and non-directionality. It's also small enough to fit in almost any lamp or fixture. The only downside is price ($40 at Home Depot) and the lack of a 100W equivalent option.

I've been using a LED bulb for 2 years or so now (it's on 24/7).

It's currently half the brightness of when it was when new.

I am not impressed, unless they have fixed that.

Also, if LED is so efficient, why do they produce so much heat that they need massive heatsinks or advanced cooling?

While they are more efficient, all the waste heat is produced in the die. Incandescents on the other hand radiate infrared light.

The stuff about mercury in compact-fluorescent lightbulbs is basically nonsense. A typical CFL might contain 5mg of mercury. You ought to open up the windows for a few hours if you break one, but otherwise don't worry about it.

I do think that people ought to have the right to buy plain incandescent lightbulbs if they want, though. Energy-efficiency is very important and will remain important for a decade or more, but the contribution from involuntary speculative energy futures investment in the form of compact-fluorescent purchases is not significant; I think it's more important for people to be able to build E-Z-Bake ovens, make heat-sensitive invisible ink visible, have ready-made electric igniters, and be able to do their homework when they only have 50 cents to spend on replacing the lightbulb before payday.

I've bought several LED bulbs to try them out. The number one problem is they are about half as bright as necessary, so you need twice as many bulbs. But if you've got existing fixtures, there aren't twice as many sockets for the bulbs.

The second problem is you cannot install them in enclosed fixtures - because they'll overheat and fail. They can only be installed in open air fixtures.

These two issues pretty severely limit where they can be used.

Since this _is_ hacker news, can someone in the know explain the considerations that have to be heeded when developing LED technology for lighting environments? I'm much more interested in hearing about the science behind these bulbs and how it compares to fluorescents and incandescents than hearing 50 different anecdotes about peoples' LED experiences.

So the basic operating principle behind lighting LEDs is the correlation between light output and temperature. As the LED heats up, its light output decreases. If you want a single LED to produce more light, you have to pump more current through it - meaning more heat and a resulting higher temperature, shorter lifespan and reduced efficiency. Thus, you have a tradeoff between more LEDs and lower current (higher cost and more efficient) and less LEDs and higher current (lower cost but less efficient).

Switch tries to give you the cake and feed it to you - they give you lots of LEDs, running at high current, and they manage the heat by immersing the LEDs in a liquid [liquid water has a conductivity and capacity for heat ~5x greater than aluminum even with a lower density]. But this has its own problems that you can probably imagine - not least of which is that the liquid probably absorbs more light than it gives back by removing heat.

Based on my "in the know" status, I can tell that Switch will be losing big bucks trying to sell those lights at $20 but they are working hard to cement themselves in the market early. I'm very sceptical and look more towards smaller innovators like LEDNovation and LED Integration Technology.

I agree that this is more interesting. Some basic background is at

http://en.wikipedia.org/wiki/Light-emitting_diode#White_ligh...

I brought an LED bulb since I found the idea interesting.

Holy fuck that bulb sucked though. Not only was the color so bad it made my hand look like a body in a morgue but it wasn't powerful enough to light by study table unless the sun was up.

I don't think I will try again. A standard bulb works just fine...

Some of the LED bulb makers have taken deceptive packaging to an extreme. At Home Depot the LED chandelier bulb I was looking at said "Compare to 20 watt incandescent bulb!". So I did. It puts out 1/4 the light. I'm glad they tipped me off that it was useless except as a marker bulb, but "Comparable to a dim incandescent nightlight bulb!" would have been more honest.

I'm not convinced. Saving a few dollars a year on electricity is not worth any compromise in the quality of lighting for me.

If we are concerned about electricity, maybe we should generate more. It's not like we don't know how.

The concern isn't with a limited supply of electricity but with the environmental costs of how it's generated...

You know, I think I'd rather have an extra $19.50 right now than a light bulb that maybe pays off 4 or 5 years from now.

I didn't look at the guys spread sheet, but I wonder if he discounted the future savings to their present value and current interest rates.

But if you ran a stadium, or office building, or government installation, you would definitely be looking at spreadsheets and total-lifecycle-cost. It really only makes sense to do so when you have hundreds/thousands of bulbs to maintain.

I just replaced most of our bulbs with CFLs a couple years ago. I don't think I'll be jumping to Switch lights (or any other LEDs) for a while. There doesn't seem to be that much of lifetime savings between CFL and LED, but maybe when the CFLs need replacing, LEDs will be more cost competitive.

I've just replaced 3 outdoor 50w halogens with 3w "warm white" bulbs and 6 indoor similar with 4.5w bulbs. Total cost was about $220. Definitely less bright than the halogens, but these are areas that we leave on all night outside or for about 6 hours inside. They're expensive, but I calculate that the outside ones will pay themselves off in about 6 months and the inside bulbs 20 months. This excludes replacing the halogens, which happened relatively often. We still have many other down-lighters indoors, but they're not on very often and need to be dimmed, so I'll wait until the cost drops.

Well its practically an advertisement for Switch Lightning. (I sense PRs at work). There are LED bulbs in the market already and they are not new, but the legislation limiting incandescent lightning is new.

Yeah, they're still pricy. So far the only one I've replaced was one of the bulbs in the kitchen fixture. It doesn't really produce great quality light (although I guess some of them are pretty good). What it has done is save me having to get out the step stool and move the table to change a bulb on a couple of dark, before-coffee mornings. In that respect it's been worth every penny.

If I wind up in a house where I'm sure I'll be living the rest of my life, the economic calculus will change a bit.

If the LED driver handles dimming through PWM, no thanks. The flicker is very noticeable. Also, the 20k hour figure for LED's doesn't factor in "lumen maintenance": LED's can dim with use, especially high output use or possibly poor ventilation. Lumens after 10k hours may not be the same as new. Last, it's hard to find a good warm (2700k?) LED that is remotely efficient, though the Cree XP-G's seem to be getting there.

You are right about the PWM dimming - but it's a better alternative to the lack of dimming of CFL bulbs.

The 20k hour figure likely does figure in "lumen maintenance" - although Switch maybe stating a "consumer lifetime", the mean lifetime of the product, they are required to provide a lumen maintenance figure called L70 - the time it takes for a 30% reduction in light output. Chances are these light bulbs just broke before reaching that target, and the consumer lifetime is more useful to the market at large.

And, yes, the warm color with high CRI is the hardest LED to make - that's why they're waiting until next year to try and offer a warm white 100W equivalent bulb.

"If the LED driver handles dimming through PWM, no thanks. The flicker is very noticeable."

The flicker is unnoticeable if the chop rate is a few kilohertz or higher.

If you notice flicker, the PWM frequency is too low. I recently built a PWM driver for high power RGB LEDs and there's no flicker, even at very dim brightness. I don't know what's on the market, but probably I should get into it ;)

The dimming is generally done on the AC side before it even reaches the bulb. So it would be dimmed by either resistive (unlikely) or by cutting the AC waveform by timing the zero crossing point.

For more information, and a look at the waveform that is produced see this awesome page: http://www.epanorama.net/documents/lights/lightdimmer.html

You wouldn't even notice any flickering at a PWM frequency of 120 Hz. Linear (resistive) dimming is just not done anymore. Just about all lighting fixtures these days dim by either PWM or chopping the AC signal.

I'm a little confused. Even Ikea sells rather good (albeit slightly expensive) LED lighting, and I'm using some in my house right now. Heck, they're the primary lighting in our kitchen.

Maybe the Bulb Form Factor is still a work in progress, but in terms of actually lighting places up, I suspect a lot of people are already using LEDs. So why is this a revolution?

You might not notice because you're used to them, but bring in a lamp with an incandescent bulb and put it beside your IKEA lamps, and notice the vastly different color temperature. Which you prefer is largely a matter of preference, but most people grew up with incandescent bulbs, so prefer that color temperature.

I have noticed the color temperature. As a photographer, I am somewhat sensitive to the issue.

I'm saying, people are buying and using LED lighting all over the country and presumably liking it. Ikea used to have problems stocking the item in demand and only in the last 2 years have they really managed to consistently meet demand. I wonder if maybe the idea that most people are holding off on a bright LED light with a ruddy orange glow is really supported by data or if it's just assumed.

Heh. I saw switch at LightFair - and I asked the guy whether the envelope was made of glass, and what the liquid was made of. Both questions he skirted around.

Me personally, I'd rather have broken glass and a microscopic amount of mercury vapor, than broken glass and a huge "liquid" stain. Switch isn't over the hump yet.

My concern is whether the bulbs flicker in the the same way that LED backlit displays do when they are set at less than full brightness. I have found this effect to be migraine inducing when the frequency is too low. Does anyone know if this type of circuitry is employed in these bulbs?

You are talking about the PWM frequency. If you are having issues with that make sure to look for a manufacturer that has theirs at 30 Hz or higher, when I worked with LED's we tested that 50 Hz was enough to make sure that humans were unable to see that the light was not constant on.

I've built LED displays and the lowest frequency at which I wouldn't notice flickering was 500 Hz. This was at 1/8 duty cycle though, so you might get away with a lower frequency at a higher duty cycle.

I find your claim that 50Hz was enough rather strange. Even incandescent light bulbs have noticeable flicker at below 50 Hz and those have considerable warm up / cool down times, as opposed to the LEDs.

Hmmm, it is possible I am doing the math wrong based upon the timer/interrupt on the micro controller and the rows we had...

LED light bulbs have gotten bad reviews in the past, which is why they aren't some magic alternative to CFLs. The author kind of glosses over that part.

For instance: http://a.wholelottanothing.org/led-bulbs/

Your link specifically calls out LED bulbs as being too dim, and not listing their incandescent equivalent wattage (ie, a 10 watt LED == a 60 watt incandescent).

The article specifically lists a 60 watt replacement, and a 100 watt one soon.

My point was more that the author seems to have just discovered that LED bulbs existed, and that it is a new alternative to CFLs.

As long as it's a bulb that will put out enough light to light a room I'm all for it.

Talk to me when it actually exists.

I get so sick of seeing these revolutionary new products that never make it to market, have a much higher pricetag, or don't actually do what they claim.

But they do exist. You can already buy them. They are not cheap but they are getting cheaper.

LEDs are nothing new. And they have have replaced other types of indicator lights decades ago (the lights on older computers had incandescent lamps, and if you have seen those in practical use - you are getting old). Trafic lights are already LED here in Bulgaria and I'm sure in other countries too.

not really. this company's product is still in the pipeline: "Switch Lightbulbs will be available fall 2011".

there are other LED lightbulbs available, but they are more expensive. and the ones that I got all failed in less than a year.

Well they do exist. I was actually surprised to see this as 'news'. I got one at home for almost 2 years. Works like it was advertised. It's from Pharox http://www.pharox-led.com

These exist - but the $20 price tag is only possible using subsidization from investment funding. There are at least $20 worth of LEDs in the Switch bulbs - meaning they will be making negative dollars on each sale at $20. But that's how they make the stats look so appealing.

That, and they're made of glass filled with an unknown liquid, is the reason I'm not very convinced by Switch.

I would like it if "news" like this had a less biased title (or subtitle in this case.)

Anyone know how well LED bulbs handle cold (below water freezing) temperatures?

The LEDs themselves work excellent at low temperature - but you may have problems with your electronic driver's properties changing at lower temperatures.

I will do that when the prices come down from the sky. This guy can rationalize it all he wants at $20/bulb, but very few people will make the change. Where it does make sense is places where the lights are on all the time and places where the lightbulbs are hard to change (It's a huge pain to replace a bulb in a hard to reach place so if you have one that lasts forever it is extremely time-saving)

I think once you get below $5 you'll have takers, but we're probably a ways off from that. I do like the fact that they are dimmable. I haven't found any CFL's that work well with that aspect, and the ones that do are very expensive.

If it looks good, $20/bulb is cheap. A dimmable CFL costs about $13 and it makes an irritating noise. Regular CFLs are still around $6-7 and nobody likes the way they look. The electricity saving of LED/CFL isn't as great as over incandescent, so that won't resonate with homeowners straight away, but for businesses it's a no-brainer and that will create enough demand to bring the price down.

I have never paid more than $1 for a standard CFL in the past 3 years... I've gone to my local hardware store several times and replaced all of my bulbs, but none of them cost more than $1. Here's a 2-pack at Lowes for $0.98: http://www.lowes.com/pd_20845-371-60038_0__?productId=123352...

Dimmable ones have cost me $5.

You are right. I mixed up my memory of the prices from the last time I bought specialist bulbs, because I was putting them in regular lamps.

After stopping by a store yesterday and remembering I meant to get some lightbulbs, it turns out I was right after all. Incandescent bulbs are <$1 each, but CFLs start around $5. (Walgreens) Also, your link to Loews shows two CFL bulbs for $4.98, not $0.98. If that was a troll, then well-played; otherwise, it's time to visit the optometrist.

I think that's false economy. How much do you spend on nice furniture and interior decoration? Take 10% and spend it on good lighting instead, and your quality of life will improve.

Even if you're a starving student in a dorm with nothing but a secondhand ikea couch and faded Bieber poster, good broad-spectrum lighting will improve your aesthetic experience.

Reply on news.ycombinator.com