2. The LEDs tested are not the "warm white" (1500-2500K) commonly used as replacements for incandescent bulbs in the US. They're the "blue white" (>4000K) ones, or pure blue or pure green at specific wavelengths.
With those constraints, it's an interesting result. It does not mean you need to run home and rip all the lights out, though.
Yeah. The hypothesis seems to be that the blue end of the spectrum is the danger here, LEDs being relatively bluer than incandescents or common fluorescent phosphors.
But even so, the high energy spectral content of indoor LED lighting is much, much less than what you see from direct sunlight. If LEDs are really a vision risk to the modern office worker, they're significantly less so than the same risk of agricultural or other outdoor workers.
But, just to play devil's advocate, it seems plausible that sunlight might activate some kind of defense mechanism in the eye that protects it from the effects of the higher energy part of the spectrum.
If nothing else, the overall brightness levels would cause your pupil to shrink more in sunlight than under LED lighting.
It's definitely a good point. Dark sunglasses that don't block UV light can actually increase damage to your eyes, because your pupil dilates in response to visible light only.
However, the ganglion cells that influence dilation and melatonin happen to be mostly sensitive to blue light (~470 nm), so in theory this would only be a problem for LEDs largely concentrated at even shorter wavelength.
I have worn polycarbonate lenses my whole life, well since infancy. Polycarbonate lenses block UV light. If I go outside without my glasses or sunglasses my eyes hurt, even in winter. I don't know if my eyes are super sensitive or they just aren't used to the UV light and the micro-damage it causes.
That's not really a defense per se, just a warning that something is wrong. I had pain in my left eye when I burned my retina with a laser, too, but that didn't reduce the damage.
If you don't mind me asking, what was the damage, and what wavelength/power was it? Laser eye incidents fascinate me since there's not too much information about them.
[edit] My email's in my profile if you'd prefer not to say publicly, although since a few others upvoted my comment it looks like it's a popular question :)
I don't mind at all! Happy to oblige. It was a 405nm (violet) 5mW diode laser, housed in a pointer of a sort that's sold cheaply on eBay [1]. It was focused through a spherical lens taken from an old scanner and another from a broken magnifying glass. With such a short focal length, it should've been very hard to injure myself, but I had my head quite close to the bench and was also messing with a couple of mirrors. Given such a golden opportunity, Murphy rarely fails to oblige, and in this case he obliged me right in the eye.
(The idea there, if you're curious, was to build a light collector with a relatively wide incidence angle that'd focus the beam onto a mostly decapsulated violet LED reverse-biased to act as a photodiode and tied to the base of an NPN transistor, so as to switch a GPIO on my entertainment center Raspberry Pi and use the laser pointer as a kind of remote control. The circuit worked quite well when illuminated, but a ~1mm target is hard to hit from ten feet away, hence the messing about with focusing optics.)
I don't have a detailed characterization of the injury because I'm rather overdue for an optometrist's appointment, but later experiments with the same configuration (and goggles!) showed that, so focused, the beam would deform and start to burn black ABS plastic within a few seconds, and took no perceptible time to burn a hole in a sheet of black construction paper, in both cases producing damage in a radius well under a millimeter - I found that, unless placed precisely in the focal plane, the beam was too widely spread even to raise smoke from the paper. The effect on my retina would therefore have had to be the same sort of closely localized heating, a hypothesis borne out by the effects I observed to result.
Beyond the pain already described (which lasted a good couple of hours), those effects included the click sound I mentioned, which I believe to have been produced by the sudden vaporization of a small volume of aqueous humor adjacent to the retina, and a quite noticeable but thankfully non-foveal bright green floater which did not cease to be perceptible until about a month had gone by. Whether this latter is the result of the injury healing, as I gather such minor retinal injuries often do, or rather of my visual cortex having grown a new filter, I won't know until my next slit-lamp exam.
Warning signals do reduce damage though: if you hadn't gotten the warning then maybe you would have continued burning your eye. I certainly would have continued looking at the UV light if I hadn't felt any pain.
It was a visible-wavelength laser, so I would've known there was a problem as soon as it hit my retina, even without the audible click and the sudden and disquietingly misplaced feeling of having been kicked pretty hard in the nuts. But I take your meaning.
(And I own a pair of wavelength-matched laser goggles now, as should you if you're going to do anything at all with a laser and any kind of focusing optics. Carelessly enough handled, even a 5mW pointer beam can pose a real risk. Don't make the same mistake I did!)
Oddly enough, yes! Not in the same area, obviously. But if you recall the last time you caught a blow to the business, then imagine how that'd feel in your head instead of your belly, that'll give you a decent idea of what it was like. I don't know what else would, short of actually replicating the injury; I've had enough shiners to know what that's like, and while it's not at all pleasant, I'd probably choose it over another laser burn if I weren't given the option of declining both.
Fascinating though it was, I have to say I found the experience with very little to recommend it, not least because it disquiets to realize that the click you've just heard, conducted through the bone and tissue of your head, was produced by a small volume of fluid inside your eyeball suddenly boiling.
Probably "overwhelming" pain. Where you almost complete loss of rationality, a torrent of sensation that blocks out all other thought and sensation - vision narrows to a point, sound becomes choppy and discordant, limbs feel far away or even absent (no, I don't mean the one that's potentially chopped off or something ;) , the other ones)
Oh, it wasn't anywhere near that threshold! (Although it was well over that for swearing the air blue.) See my sibling comment to yours; I just meant that the nature of the sensation was almost identical to that produced by a shot to the area, just in my head instead of my belly - and, yes, that really does feel as weird as it sounds like it would.
It feels like the kind of pain when you injure your testicles, but you feel it in your eyes. It is similar to the kind of pain you feel when you press your thumb and index finger between the bones in your forearm (under your wrist).
Our eyes are extremely advanced. When I stand near a laser cutter, of course wearing protection glasses, I can feel it in my eyes whenever it is turned on.
2) Spectrum: LEDs can have very uneven spectrums with spikes at particular frequencies (particularly in applications like street lamps that don't use phosphors). This is something an eye in nature would never experience.
This is why I didn't jump on the spiral CFL bandwagon and why I won't jump on the LED bandwagon for some time to come. Good old-fashioned incandescents may be less efficient to run in summer, but they are heaven to my eyes compared with the crap being sold today. I'll gladly pay for lighting that doesn't make me nauseous or make everything look horrible.
Decent LEDs have very similar spectra and colour reproduction to incandenscent. And everyone on this forum stares at LED or CFL light whenever they look at a computer screen.
Which bothers me less, as it's much darker, contains less blue light with flux or night shift, and doesn't flicker as many bulbs from the hardware store do.
There's still the flicker and glare, and an unnatural "look". LEDs are not as bad as CFL's, I'll grant you that, but still not as good as an incandescent for the living room or bedroom.
If the spectrum is good, and there's no flicker, I think this could make sense in the kitchen or hallways. I haven't been that pleased with the LEDs I've seen but they are still better than CFLs and show more promise.
In the living room or especially the bedroom and bathroom, I prefer dimmer lighting with less blue spectrum. My 29W halogen incandescent on the night table doesn't exactly burn through a lot of electricity and has the most pleasing light in my eyes. Until there's a LED that can perfectly match that, I'll stick with that until these bulbs get banned, too.
The LEDs should be in 2 or more groups that each pulse at AC speed but offset by a fraction of the wavelength, that way there is little to no flicker. Or convert to DC and power it that way.
That's WAY too complicated. Just run the power through a high-frequency tank oscillator to convert from 60Hz to a much higher frequency with just a coil and capacitor.
Or, do what I did. Just make the rectifier directly out of LEDs itself - https://www.youtube.com/watch?v=jDnCHyF7o5U And from there you can power your devices with the leftover DC power.
LEDs are seriously far more robust than they were a decade ago.
Yes, there is "leftover DC power" as a diode only allows power in one direction and whatever is left after the LED voltage drop is rectified DC. If I put 12V of LED on a 120V leg, I'll have 108VDC rectified left over after the LEDs take their voltage drop. If no other device uses that power, it just gets dumped out to ground, but it's still unused DC power.
That's not PWM in the video. That's raw mains frequency.
What's the frequency of the 12V AC supply in that video? Is it the same as 110V AC going into the transformer? What is causing the lights to pulse every few seconds?
Yup, same 60hz power mains, it's a simple 10:1 step-down transformer with a resistor and filter cap wall-wart. My camera is recording at 29.97 FPS so you get to watch essentially which sets of diodes on the rectifier are operating on the waveform peak/trough (or would actually see which ones had I had this on an o-scope.) The weird star-trek transporter effect is just the intensity of the lights screwing with the camera + the progressive-scan nature of the camera sensor.
Here's one directly on the mains power. NO CONTROL CIRCUITRY.
You don't want to power LEDs with DC if you want the highest light output. This is because the junction heats up and lowers the efficiency of light output. When LEDs are pulsed at a high enough rate, the junction temperature can stay lower at higher currents. This leads to higher light output.
Its funny, I replaced 2 spot lights in my kitchen with leds and when I use one certain kitchen device (high speed rotation thing) the leds start flickering visibly.
One or the other doesn't have proper grounding or the cable for power to your appliances is run too close to your cable for lighting (which should be on separate circuits, mind you!)
Different rules here. There is proper grounding, but both the device and leds do not use grounding.
Grounding is just for the metal cases, the lamp holder for example.
How do you ground a light bulb, and a motor inside a plastic case?
Ground is for safety in AC, when there is a short the ground should take the current away resulting in power loss through the earth that in turn makes the breaker flip. Without grounding the wire that is exposed and touching something will (depending on material etc.) not transfer any current because of too high resistance, then as you touch it it will shock you and then the breaker will go off.
(A loss power breaker, not a draw power breaker)
"How do you ground a light bulb, and a motor inside a plastic case?"
Same way you complete any other ground - since ground and neutral are literally on the same leg (go look at any installation. You've got TWO wires coming in at the mains, ungrounded.)
So, tell me, how do you ground those ungrounded 240/277V wires??
You don't. Both of them ARE the natural ground dependent upon whether you're at the peak or trough of the waveform. That's kinda how AC works.
Where I am from we have installations differently.
phase and neutral come in the box, they go through a loss breaker(RCD) and then split up in to multiple breakers.
Then those wires go to the socket.
Then near the box there is a metal rod in the ground with a wire going to the box, this wire connects to the ground wires coming from sockets.
The ground wire, is not connected to anything but things that are not supposed to have any power flowing through.
Yes Neutral is grounded (outside my house, on the main)
But when we say we ground things we mean the third connection.
Sorry, didn't know it was so different over there.
edit: Saw the correct name for the "loss breaker" in other post, it is RCD :)
edit: Question, what happens if you hold the neutral to the ground wire there?
Yup, it's different over here. RCD (we call if GFCI here in the USA) tends to be built directly into the outlet, main breakers installed at the box. You guys generally tend to run 240/277V series-ring wire runs, though, yes? We run parallel circuits here.
240V, What we usually do here is run wires to a light point and fan out from there (the light fixture has a small box above it), small rooms can be grouped together by connecting the two boxes. Heavy appliance sockets get separate circuits, and kitchens usually have most circuits.
So parallel mostly.
ground and neutral are literally on the same leg (go look at any installation. You've got TWO wires coming in at the mains, ungrounded.)
Forgive my ignorance, I'm no electrical expert, but did you just say that the ground wire is connected to the neutral somewhere in the circuit? Or did I misunderstand you?
That's exactly right. Ground is always tied to neutral in a single-phase 120V installation. Open your breaker box and remove the front panel. You'll see it. http://i.imgur.com/hCY1jV0.png
EDIT: That "always" is assuming the house is modern wiring and not older-style two-prong wiring for the wall outlets. Those systems were just main and neutral at the box and controlled by a fusible link.
Intriguing. I'm from the UK and I believe we do it differently here because I have an RCD on my earth bar in the consumer unit. The RCD detects current on the earth wire and cuts power to the circuit instantly. An earth fault doesn't trigger the circuit breaker, it triggers the RCD. A fault between live and neutral would trigger the circuit breaker.
If I'm understanding you correctly, in the US the ground wire is functionally identical to the neutral due to the bonding. Therefore (this is my own deduction and I'd appreciate being corrected if I'm wrong) you could, in theory, invert the ground and neutral wires in the plug of an electrical appliance and there would be no change in the behaviour or safety of the appliance?
"you could, in theory, invert the ground and neutral wires in the plug of an electrical appliance and there would be no change in the behaviour or safety of the appliance?"
Generally, no. I've seen backwards wiring jobs cause 48-80VAC to run through the casing of microwaves (they usually use the casing as a floating ground.) Any wiring inversion will usually cause some issue somewhere.
This is fascinating, I'm reading lots of resources attempting to understand how this works. Thanks for your responses, I appreciate you taking the time to do that.
That device must be causing the powerline voltage to dip harder than normal (or otherwise screw with the voltage waveform), more than the capacitors in the AC/DC converter for the LED lights are designed to handle. And thus uneven current = flickering light.
> If LEDs are really a vision risk to the modern office worker, they're significantly less so than the same risk of agricultural or other outdoor workers.
Those outdoor workers probably ought to be wearing UV-blocking sunglasses regularly to avoid increasing their chance of cataracts:
My optometrist was quite emphatic about this because she said the damage is thought to be cumulative, so if you're going to have a lifelong outdoor career or avocation, the difference could be significant.
Oh, agreed. I'm just saying that the "LED risk" being discussed is at worst of an order already seen by large existing workforces. So we office workers are at worst at a risk comparable to our farmer relatives. We should investigate and take what precautions are reasonable, but broadly: we can deal.
I do think the theory elsewhere in this thread about the different spectral distribution could be significant. This is why apparently non-UV blocking sunglasses can make UV damage worse: the pupils open wider when visible light is blocked and so let even more UV in. If you just skew the spectrum of some light source toward the UV without increasing the overall luminous intensity, presumably it could cause some health risks in a similar way.
This isn't correct. Most white LEDs use a blue LED which is shone on some phosphors which then emit a broader spectrum of yellow light.
The ratio of yellow to blue light determines the colour temperature perceived. Some of that blue light will always get through, so white LEDs (even "warm" ones) will always have a blue light component to them.
One group of people who could potentially be affected: people who use anti-SAD lighting (especially people building souped-up DIY rigs: https://meaningness.com/metablog/sad-light-led-lux). Those use LEDs and are specifically meant to be very bright and very blue.
Hopefully this turns out not to be a big deal. Having to choose between depression and retina damage would sure suck :/
Most anti-SAD lighting (your average full spectrum bulb not weird things like blue light beds) is not designed for high lumens like this but for broad-spectrum (avoiding spectral 'spikes'). Which should probably be beneficial on this issue
Also, those lights are suggested for skin exposure, not for staring directly into or for illuminating your space... still though, proximity would encourage more harmful interactions.
They are not intended strictly for skin exposure. They are marketed and used as a way to expose your eye (brain) to the high light levels that (may be) needed to regulate diurnal activity.
The Phillips light specifically has warnings about not looking directly at it, but rather having it in your peripheral vision. It calls out that eye damage can happen if staring directly at the light for prolonged periods.
The lights that I've looked at (look up liberty vt20 for example) have all suggested that eye contact with the light, peripheral or otherwise, is essential for results.
They got damage at 6000 lux with pretty much all the light sources. The interesting result is that they got significantly more damage with the white LED source than with other sources at 500 lux.
Someone else was nice enough to post a link to the paper:
Important to note that this was done with dilated pupils, so the difference is seen in a condition not normally encountered.
Overall, I think the focus should be on light spectrum, not LEDs per se. Warmer LEDs has far less direct emission, with most of the light being re-emitted from the phosphor.
For comparison, typical artificial light levels in an office are more like 300 lux. IES recommended illuminance for a professional indoor basketball court is 1000 lux.
Daylight ranges up to 120,000 lux, so 6000 is several times the lighting in most spaces, but still nowhere near bright sunlight.
That said, from the abstract (I don't have journal access), it sounds like they found a result in both 24H high intensity, and in on/off "domestic levels".
Is the 6000 lux you mentioned their value for high intensity, or is that what they considered as domestic levels?
multiple experiments were made, one at 6000 as the GP mentions, but also one at 500 lux. There's more information here but in French[1][2]:
On the other hand, by exposing the same animals to a luminous intensity similar to that usually used in dwellings (500 lux) for 24 hours, only the LEDs appeared harmful
500 lux is also much greater than what reaches your eyes in a typical household lighting scenario. If you stare in to a 100W incandescent light bulb from one meter, you're getting about 100 lux to your eyes.
We don't usually stare at light sources; we use them to illuminate objects. A 100W incandescent from 1 meter on to reading material or other work is considered fairly brightly lit, and that's not getting the full 100 lux back to your eyes - only what the surface reflects.
Daylight is far brighter, but most of us have been warned from early childhood about the danger of staring at the sun. 100,000 lux may be present, but we're not directing it in to our eyes.
Not to take away from your fundamental point, but to add information to anyone shopping for lighting:
4000K isn't "blue white" or even what's usually called "cool white". It's warmer (that is, more weighted toward the yellow-red end of the spectrum) than sunlight at midday (5000K-ish) and much warmer than LED and fluorescent sources that are noticeably blue (6000K+).
Beyond that, sources with a flatter or fuller spectrum (most commonly measured as color rendering index, or CRI) will have less of a blue peak even at cooler color temperatures.
And yes, 6000 lux is a lot of light. Staring at a 100W incandescent light bulb from 1 meter away will put about 100 lux on your eyes, by way of comparison.
> And yes, 6000 lux is a lot of light. Staring at a 100W incandescent light bulb from 1 meter away will put about 100 lux on your eyes, by way of comparison.
That puts 6000 lux firmly in the "my eyes hurt, this cannot be good" category.
Take a high-powered flashlight - the kind the police use. Hold it at arm's length. Point it straight in to your eyes. That's about 6000 lux.
They had also dilated the rats' pupils with atropine, making the acute exposure test equivalent to, or perhaps worse than staring at the midday sun for 24 hours straight.
Even the lower-level chronic exposure test used 500 lux with light from all directions so there's no way to look away. We don't use light that way in reality, and while I'm sure the intent was to accelerate the test, I can't help but suspect it's the equivalent of testing whether bathing in warm water might be harmful by boiling rats alive.
> The data suggest that the blue component
of the white-LED may cause retinal toxicity at occupational
domestic illuminance and not only in extreme
experimental conditions, as previously reported.
The domestic illuminance they used was only 500 lux, which they describe as the "domestic classic light intensity".
Can't see the full paper, just the abstract, but another thing I wondered if they took into account is the effect of strobing. They didn't include a control that doesn't strobe (such as halogen), since CFLs strobe in the kHz range. LEDs may or may not strobe depending on how they are driven and that isn't stated in the abstract.
It does however, seem much more likely that this is to do with extra energy in the UV end of the spectrum, which IIRC WW LEDs cut a lot more of (as well as the additional, visible blue light that makes CW seem so much brighter).
I can tell when LEDs are strobed at the (double) mains frequency; e.g. Christmas LED lights that likely blink at 120 Hz, since they don't have any electronics beside the LEDs. These do give me a slight nausea.
Compared to these, I can conclude that my home LED lights likely strobe in kHz range or above, or maybe not at all.
Cheap Christmas LEDs with no electronics look so awful because they strobe at 60Hz and are completely off for a full half-cycle (while the LEDs are reverse-biased).
I can't speak for you but I find the full-wave rectified ones not to be a problem. Of course the difference isn't just the doubled frequency, it's also a much higher duty cycle.
Oh wow, thanks for pointing that out. I thought I was having a migraine aura when I noticed vaguely flickering christmas lights out of the corner of my eye.
If the issue is confined to the uv, it's a dead-simple to fix. A uv-absorbant filter (ie like on all eyewear since the 80s) should eliminate the problem. Costs would be next to nothing. Conversely, anyone worried could don any eyeware, even swimming goggles have such coatings.
Yeah, I looked and this and thought "well of course they can't handle as much light as humans, they're supposed to spend most of their time inside walls et al".
Not terribly scientific, but pretty relevant when transferring this knowledge to humans.
6000 lux of BLUE light is a _ton_ of photons. The light meter they used applies the human sensitivity curves to the photo frequencies. The sensitivity of the human retina to blue light is very very poor. Most of the time you see graphs of sensitivity vs wavelength the curves have been normalized. Here is one graph where the blue channel has only been inflated by 10x: http://www-i6.informatik.rwth-aachen.de/web/Misc/Coding/365/...
So 6000 lux of blue is approximately 30x as many green photons. For blue LED plus yellow phosphor that 30x "yellow" will be the signal that you will use to decide "that's enough light to perform work."
The rat's eyes were definitely cooked by all of the blue. Imagine if they had used near UV or near infrared LEDs and the same meter to calibrate the intensity. Cooked rats.
Although if you want to measure an effect in a reasonable time frame, you'd up the concentration. You won't measure much damage from a "reasonable" amount of lead in a short time frame, either, but over longer time scales you'll probably conclude that you want to rescale reasonable downwards.
I don't have access behind the paywall, but I do run engineering for a luminaire manufacturer. Do you have access to the full details of the illuminance source and output?
> We used two types of lighting devices. For exposure to white LED, commercial cold white LED panel generating 2300 lumens during 24 h was used. The LED panel was placed above 8 transparent cages, placed on white surfaces, leaving enough space for air circulation and constant temperature maintenance at 21 °C. The illuminance measured at the rats’ eyes position was 6000 lux (Photometre DT-8809A, CEM, China).
> For long-term exposure, specific devices were built and characterized by Statice, France (Fig. 1A). Metallic boxes contained rows of LED with a diffuser in order to improve the directional uniformity of the radiation and avoid punctate sources. Alternatively, CCFL or CFL were uniformly distributed around the metal cages. Each cage was placed in a metallic device that was then placed in a ventilated cupboard allowing for a constant 21 °C temperature control (Fig. 1A). The light intensity was controllable and the distribution of light in the cage was homogenous whatever the rat position. Different types of LEDs were used: cold-white LED (pure white 6300 K), blue LED (royal blue 455–465 nm), and green LED (520–35 nm) (Z-power LED, Seoul Semiconductor, Korea). Exposure intensity was spectrophotometrically measured by Statice.
> Exposure protocols
> Acute exposure: LE and W rats were maintained in a cyclic light/dark (250 lux, 12 h/12 h) environment for 21 days. The day before light exposure, rats were dark-adapted for 16 h. The next day, pupils were dilated with 1% atropine (Alcon, Norvartis, Rueil Malmaison, France) under dim light, and rats were isolated in separate cages containing enough food for one day. After 24 h of exposure, rats were placed again in a cyclic light/dark (250 lux, 12 h/12 h) environment for 7 days and sacrificed for histology and immunofluorescence analysis. Control rats were submitted to the same pre conditioning protocol but not exposed to light. Different types of light sources and light intensities were used as detailed in Fig. 1B. For cold-white LED, different light intensities were tested from 6000 lux, to 1500, 1000 and 500 lux. Blue and green LEDs were used at 500 lux which is the domestic classic light intensity. CFL was used at 6000 lux and 500 lux, CCFL at 6000 lux. Illuminance was measured at the level of the rat eye.
> Long-term exposures: Rats (LE and W) were maintained in a cyclic light/dark (250 lux, 12 h/12 h) environment for 21 days, then placed in specific cages for chronic cyclic exposure to different types of light at 500 lux: CFL, white, green and blue LEDs. Animals were sacrificed right after 8 or 28 days of exposure. For the long-term protocol and in order to be as close to domestic light as possible, rat pupils were not dilated.
It sounds like their control rats were kept literally in the dark... so unless I'm reading it wrong, this is an interesting result, with a terribly bad control that renders (IMNSHO) their conclusions invalid based on poor controlling of the variables.
This sort of thing should have had 2 control groups. One kept in the dark, the other exposed to the light protocol they used for artificial lighting, using natural sunlight as the light source.
Its not clear from the wording above exactly how the control group was exposed in pre conditioning, and what the nature of their post exposure protocol was.
500 lux in cyclic test is not outside the realm of reason for representative real world exposure.
Personally, I'd like to see a control under natural sunlight, and one under a man-made filtered blackbody radiator.
Wow, guess you can get LED poisoning :-). The question that I've heard being bandied about is that people use brighter (higher luminous flux) LED lights and displays for better visibility and the question was whether or not the pupil dilation reflex was appropriately triggered (sort of like going blind by looking directly at a total eclipse of the Sun). If a cause could be identified and reproduced it seems the "fix" would be simply adjusting the spectrum of the output to clue your eye into the fact that it was really bright.
Anyway, I wonder if these guys got a follow up grant, does anyone know?
I'm curious why LEDs used in light bulbs could cause eye damage, but not for backlights in computer monitors or other common uses? I kindof get the impression that the blue-tinted led's are being considered 'evil' and studies are 'showing' that they are 'harmful'. Certainly these blue led's have far far far less of any wavelength at all of output than say the sun at noon...
I wouldn't be surprised to see that this study is never reproduced.
Just a guess how this could happen with LED bulbs but not the sun: at noon, there is certainly more short-wavelength power in sunlight but if you're outside without eye protection your pupils are going to shrink down to limit retinal exposure. Out of the sun, your pupils can be open wider but the short wavelength light is still there with these LED bulbs. This is a pretty new development: humans have of course had artificial light sources since before recorded history, but flames (and much later, hot wires) had much less blue output than these specific LEDs they're testing.
Another guess: they might actually find similar retinal problems from sunlight exposure in people who spend most of their working life outdoors, but since most people in developed nations don't work in agriculture anymore their baseline population doesn't reflect a lot of sunlight exposure.
Turn the lights off in a room, and see how much of it your monitors can illuminate. Our eye's logarithmic response to light can end up confusing us about the mile-wide canyon between a glow and a light.
I think, although I'm not entirely sure, that LED backlights work in one of two ways. Either there's a white LED that is behind red, greed, and blue filters, or there are separate red, green, and blue LEDs. In the former case, the filters are likely to reduce short-wavelength light and, in the latter, case, the red and green LEDs would be inherently safe and the blue LED would most likely be selected to be longer-wavelength for efficiency reasons.
(IIRC blue and violet LEDs are close to 100% efficient, but our eyes are much more sensitive as the wavelength gets farther from violet. This means that the "efficacy" of less violet LEDs is much higher. The backlight power will be set to achieve some desired apparent brightness, so a longer-wavelength blue primary would save power.)
Many studies have already produced similarly disturbing findings, although I don't know if this specific study has been replicated. See my other comment.
Having now skimmed through the paper, there's one very reassuring thing that no-one's pointed out yet:
Long-term exposure to LED at 500 lux, in cyclic
(light/dark) conditions induced retinal damage only in albino rats but not in pigmented rats
So, at household levels WITHOUT artificially dilated eyes, only the albino rats (who one would assume are more sensitive to light) suffered any damage.
Given that, as another commenter mentions, rats are likely to be more sensitive to light than humans anyway, this looks like it's a reassuring rather than concerning result. Just don't artificially dilate your eyes then stare at a super-bright LED bank for 24 hours!
One area where this can matter more than standard lighting is light therapy devices. I have a circadian sleep disorder and for those light therapy is one of the few medically recommended treatments (it is also recommended in the winter for seasonal affective disorder). There are two basic types of light therapy devices, a bright light that you set on the table nearby and a less bright light in a visor that you put near your eyes. I think LEDs are universal in the visor style ones and quite common in the table ones. The blue spectrum is more effective than other wavelengths so they are frequently focused on the blue spectrum although some of the table ones are white light. They are only supposed to be used for a half hour or hour at most.
I just tried for a couple of weeks a visor style light that is just over 500 lux at 500nm dominant wavelength (blue-green; it looks green and is a longer wavelength than most such devices). While it didn't seem super bright, it did seem fairly bright and actually caused me mild pain, similar to what bright light occasionally causes. The lower 315 lux level did the same at first. I started with that a few days and it seemed like my eyes adapted quite a bit so I moved to the higher level. I got spooked after noticing the pain later in the day one day after using the visor on high in the morning, so I haven't tried the lower level again to see if that still causes pain. The manufacturer said that in studies about 3% of people report that problem.
Anyway, it seems to me that my use of the visor might be similar to the lighting in this study. I haven't noticed this issue with other artificial lighting, although I haven't been around LEDs that much that I know of, mostly compact flourescents. The closest is a led light alarm clock that I have stared at at close range for a while and is listed at 300 lux, although I'm not sure if that is what I am getting even a few inches away. The visor light does have a 50-166 hz pulse so that could be related; I'm not sure about the light alarm, but I never noticed that flicker the way the visor does.
Anway, 500 lux at eye level at least can be quite a bit different than normal lighting.
The visor did actually seem to help more than any other bright light in the morning method I've tried (including getting natural light in the morning). So I wish it didn't hurt my eyes :(.
I don't like the light given off by LEDs. It looks unreal/surreal to me, and makes me uneasy if it's the only light source. Especially the blueish ones. Before installing them I was very enthusiastic about all the benefits of low power lighting, but now I want to switch back to halogen bulbs, or at least very warmly colored LEDs.
I'm oversensitive though. E.g. slightly flickering energy saving lamps make me dizzy, exacerbated by ACs making my eyes dry (hello, Ikea).
I hope those nano-coated regular light bulbs will become a thing.
The light from good quality LEDs should be indistinguishable from halogens, and better than any CFL. I've replaced my entire household with Philips 2700k LEDs of various types and they are excellent.
Poor quality LEDs are awful - the 50/60Hz flicker/strobe effect that many produce is very annoying and fatiguing and it doesn't surprise me that they're considered a health hazard.
2700K CCT means that the light from the diode is being mostly absorbed and re-emitted by the yellow phosphor. The lamps people have trouble with are those with a CCT of 4000+K, which emit several times more short-wavelength light for the same perceived brightness.
Try quality LEDs suggests that there are reputable manufacturers of LED products. Last time I tried Philips Greenpower, they said "40W!" yet the Kill-a-watt showed ~30w and it was nowhere near as bright as an 18w unit some cheapo Chinese maker had. General Electric markets a 7w UV LED - there's one 1W UV LED, and four 1.5W white LEDs, behind a fake woods glass. Yet neither the bulb nor packaging makes note of the white LEDs. It instead says directly on the bulb "395nm UV Light" as if that were all it emitted.
Quality certainly doesn't mean Brand-name in this day and age of unscrupulous law-buying companies.
watt is a poor measurement for measuering the quantity of visible light.
Well I mean just knowing the lumina of a bulb doesn't help since it's a subjective measurement and can change depending on the kelvin of the light. For myself I usualy look for 600-800 lm and around 2600K well for the bath I go lower and only take a look at 400lm.
As said measuring the light is extremly subjective for different kinds of people
"watt is a poor measurement for measuering the quantity of visible light"
Actually, it's a great measurement if you only work with the monochromatic ones. Knowing the LEDs efficiency at bin testing conditions will tell you what to expect - E.G. high-bin 3w 460n LED with 50% efficiency at 3w drive and typical operating temperatures (~70C junction temperature) means you can expect 1.5W of light. We toss in Avogadro's constant and a couple other formulas, do some multiplication, and suddenly we know roughly how many photons are being put out.
This gets harder with white LEDs as one must figure out the relative percentage of each wavelength comprising the light, basically forcing you to do the math a few hundred times over, but it's still rather reliable. I just do rough calculations in my head for White LEDs.
Not only that but about 5 years ago we replaced most of the house with Philips LED lights, most of them failed/are failing by now (they start by getting a darker patch in the bottom of one of the tubes and then it slowly spreads out and the light shuts off completely).
5 years for a product that is advertised as lasting for decades as this usage rate.
"they start by getting a darker patch in the bottom of one of the tubes"
Tubes? Presumably you're talking about CFLs, not LEDs? 5 years ago, household LED lamps were still in their infancy and were really expensive.
My experience with CFLs was pretty similar to yours. They didn't last, and dimmable ones especially would often fail within months. And the quality of light from fluorescents was never all that good anyway.
But out of around 25 Philips LEDs I've installed, most of them dimmables, not a single one has failed or had any problems so far. The oldest ones were installed around a year ago now.
The Philips A19-style bulbs kinda look like futuristic tube-ish light bulbs. It's just a circular ring of LEDs inside. One section goes out, then the next, in a cascading failure.
And yet, here I hold some cheap (literally $0.99 each) 9w Chinese LED bulbs, several of which have been operating 24/7 for a couple of years, and still working just fine.
I'd recommend you look for high-CRI LED bulbs – I've had similar reactions to yours but I'm satisfied with some of the better/higher CRI bulbs(color rendering index https://en.wikipedia.org/wiki/Color_rendering_index). They tend to be harder to find, but they're out there. GE Reveal bulbs are higher CRI, for example and I've found other brands through some specialty LED dealers that almost match halogen bulbs.
Parent is correct, high CRI[1] is what you want to look for. I'm embarrassed to reveal how much time and money I've spent down this rabbit hole and ultimately went back to halogen.
As you research, you'll discovered that there is in fact more to the single CRI value. It is actually an average measurement of a range of different colors.[2] R9 in particular is the one that is hardest for LEDs to reproduce. Some manufactures goose their CRI value by getting high scores for the other values even though their R9 or R13 (skin tone) is poor.
Be careful, though. Soraa, for example, makes fancy LEDs with CRI 95, and they use violet LEDs behind the phosphor. The problem is that the phosphor lets a decent fraction of the source light through. Soraa even has a nice graph on their spec sheet, and it has a violet peak.
Also, a neat fact: all CEC qualified LEDs are high CRI.
(P.S. I bought a GE Reveal once and it had the worst color fidelity I've ever seen. Some normally bluish surfaces were distinctly green. I don't understand how it has a high CRI.)
GE has been busted by myself falsely advertising their UV LED bulbs. In fact, I'm currently looking for more people that bought these A19-style UV Black Light bulbs from GE so we can start a class-action suit for false advertising and unfair competition.
In short, I wouldn't trust GE for LEDs. In fact, I'm so disillusioned with almost every company out there (Cree, Philips, GE, Sylvania) that I've simply gone back to building my own lights for my specific purposes. I can't trust any of these companies to be accurate or truthful in their marketing.
Maybe you could point me in the right direction. I'm looking to construct a panel that can make it look like day, but has infinite dimming.
Making daylight is easy enough. The Nichia 219B LED in 5000K R9050 (CRI 90+, R9 50+) is beautiful. 12-18 of them will comfortably make over 5000 lumens without having to be driven especially hard. It's not terribly hard to come up with options to wire them up and heatsink them properly either.
What I haven't been able to find is a suitable power supply and driver. I want to plug it in to 100-240V and have a dimmer knob giving me 5-5000+ lumens via current regulation, not PWM, though very fast (> 10kHz) PWM might be acceptable. Yes, I'm aware the tint will shift a bit through the brightness range without PWM. I'm OK with that.
"I'm looking to construct a panel that can make it look like day, but has infinite dimming."
Infinite? Not happening, but if you want to get CLOSE, your best bet is in a simple benchtop voltage and current regulator. I use four Nichia 219B LEDs to light my 55-gallon aquarium, are you SURE you really need 5,000 lumens? You could probably get away with just four of those, a fixed 12V power supply with adjustable current, and be done with it, they're ungodly bright and a quad of those can match up with the Cree MK-R.
It need not be truly infinite or stepless as long as it has hundreds of steps. I want to dial in the ideal level, rather than pick from say... 10 options.
I'm sure I want over 5000 lumens. When I want my room well-lit to promote a sense of alertness and compensate for shorter days in winter, my best current option is a flashlight that produces about 4200 lumens bounced off the ceiling. No, that's not one of those BS marketing numbers: the light is a Jetbeam T6, which in stock form uses four Cree XP-Ls powered from four 18650 Li-ion batteries. Independent tests, including my own have found its claimed performance to be accurate. I swapped the XP-Ls for Nichia 219Cs (5000K, CRI 80+, R9 unspecified) and measured about a 2% loss of output, but increased intensity. This works as a room light, but requires an external cooling fan to run on high for longer than about 10 minutes (it has a thermal sensor and stepdown), and the batteries can only keep up with that output for a bit over an hour. 219Cs are more efficient than 219Bs and can be driven much harder, but aren't as pretty. Even the recently-released R9050 (CRI: 90+, R9: 50+) version just doesn't look as nice to me.
Four 219Bs definitely can't produce 5000 lumens. The 90+ CRI versions peak at about 600[0], but I don't want to run them at peak output; the harder they're driven, the stronger the spectral peak of the underlying blue LED. I want to run them at about 300lm/ea. 12 of them will get me that, but I'd actually rather have more than 5000 lumens.
Thanks for responding. I was hoping for an off the shelf driver, but it looks like I'm going to be improvising a bit more.
I thought about getting one of those bigger arrays, but I'm concerned that the light they produce may not be as pretty as the 219B even if they have the same color temperature and CRI on paper. The 219C, for example is not, though it's still very nice and capable of quite a bit more output. Output per emitter and overall efficiency aren't major concerns for me: we're talking about around 90W for 6300lm driving 18 219Bs at 1500mA each.
It's also difficult to find the 048Z available in single quantities. I didn't find the R9050 version on Octopart, Other versions cost $50-60, while 219Bs in several tints are readily available from suppliers of DIY flashlight parts for $3/pc. Getting 219Bs and triple or quad copper MCPCBs is easy, and they can be configured for any multiple of about 3 volts, while the 048Z requires a 50 volt power supply.
I was expecting to spend over $100 when all is said and done.
"It's also difficult to find the 048Z available in single quantities"
Just ask Nichia for an engineering sample. :) Boom, free sample if they've got one for you to utilize. I don't know how many Cree, KingBright, Nichia, Osram, and Epistar LEDs I've got just because I ask to test them.
As for the 50V power supply, it'll run on a 48V 1.25A driver. You'll be fine and those are about $20 each.
I have hundreds of dollars worth of flashlights I got by offering to write reviews, but asking for an engineering sample never crossed my mind. I may have to try that.
If you point out to them reviews you've written (some of them mentioning their LEDs favorably) you'll get the engineering sample no problem, I imagine. They like people that promote their products and bend over backwards to get them tiny things that cost them far less than what they're going to get in potential advertising.
LEDBenchmark [1] does a good job with their own spectrum and flicker graphs. So far, the LED bulbs I have gotten based on their results have been good.
They might be unreliable, I've had some bad luck with Feit bulbs myself in the past. But if the price is right it might be worth the gamble, it was in my case. My bigger fear is that they're stretching the truth with their CRI number, I don't think there's any independent verification of their claims. There might also be some degradation over time as the phosphors wear out.
> using different LEDs (Cold-white, blue and green)
I wonder what the results would look like for warm LEDs ~ 2700K. I used to be a Cold-white aficionado but moved to warms for everything. If I experiment by suddenly switching between my cold and warm lights its pretty jarring (warm seems more relaxing).
Lab rats are brown rats, which aren't nocturnal, but active day and night. I'm not sure that's important for a test of damage to the eye, but there you go.
Sounds like you have some background on this. Would you elaborate? In particular, how the differences make the research unlikely to be meaningful with respect to humans?
I have no background in eyes but I have on rats. I can't affirm that the differences would make that study invalid.
I can however tell you a few facts about rat eyes.
- Their eyes are sensitive to ultraviolet.
- They have a dichromatic vision, two sets of cones instead of three. (some short wavelengths of ultraviolet/blue and a lot of greens)
- They cannot reshape their lenses.
- They have a lot more photo-receptors on their retina than us.
- They are something between farsighted and nearsighted.
- Poor binocular vision but big field of view.
Albino rats have more differences but I doubt they would have used any of those in a study on eyes.
Rats do get the same kind of eyes disease humans do (cataracts, etc.) however their sensitivity to light is really different. I don't think that it is ideal to use such different eyes to test LEDs sensitivity.
I think it's important to recognize that this is far from the only study pointing to this, and that similar problems are associated with CFL's.
For example, here is a 2014 study, "White Light–Emitting Diodes (LEDs) at Domestic Lighting Levels and Retinal Injury in a Rat Model" http://ehp.niehs.nih.gov/1307294/
Links to eye toxicity, macular degeneration, and blindness are not the only problems associated with new lighting either, which tends to be very heavy in the blue components. Lots of research links usage of blue-heavy white lighting to damaging circadian rhythms, which can lead to significant health effects. https://www.sciencedaily.com/releases/2011/09/110912092554.h...
Does anyone else use something like flux or redshift permanently? These programs change the color temperature of your monitor based on the time of day, but I use them all day.
Otherwise my eyes start watering after awhile and get itchy and irritated. I think I must be sensitive to blue light. I used to set it around 4500 color temperature but now I lowered it to 4000.
I use f.lux on a rather high setting 24/7 at work, most of the day at home (where I have better lighting), and at night on my phone. Using it all day at work with the 20/20/20* rule has greatly reduced my eye stress.
* I made a simple website using the Notifications API which reminds you every 20 minutes to take an eye break.
I use it for the majority of the day (especially now, in the winter, where the day is just a few hours long).
> I think I must be sensitive to blue light.
My eyes hurt after a few minutes of looking at my girlfriend's iPad/computer screen. She's the opposite and thinks my screen is constantly stained yellow.
If blue light exposure causes Macular Degeneration I wouldn't be suprised if environmental and genetic effects would cause some people to be affected by it and others not.
This is the case for noise related hearing loss. Everybody loses hearing acuity a bit as they get older, but exposure to noise accelerates this process. Exposure to ambient sounds in the high 60s (decibels) is harmful to a small segment of the population. That's not very loud, but fortunately most people are not hurt at that level. Occupational exposure levels are set considerably above that point.
Just a word of caution not to draw precipitated conclusions since this is a study in preclinical phase tested on rats. Of course it means something and requires further study but just wanted to emphasise the fact that it wasn't applied on humans.
Two more situations where this spectrum and intensity is in use: on stages (music,TV) the old PAR lights are nowadays nearly fully replaced by strong led spots shining directly on the actors. Still you don't spend 22h on stage of course.
The other situation would be artificial gardening system where a heavy bended light spectrum is used to grow lettuce etc.
Oh great, no doubt this will picked up as a sensationalist news item that'll inevitable lead to a futile debate with my elderly inlaws about how dangerous LED lights are, because this is proof, no matter how much my inlaws are NOT nocturnal albino lab rats.
How does this relate to staring at OLED at LED screens? Are they potentially dangerous as well? Could you prevent harm by running a tool like F.lux and lower the brightness?
Everything is potentially dangerous. The article reported an experiment done on albino rats; that's not much indication of a serious risk to people; it only means that someone should look into it further.
Color temp = 5000K. Output is 13000 lumens. The sales page assumes 10 ft x 15 ft illuminated area (install height not specified but probably 8 ft) = 250 ft^2 or ~23 m^2.
Doing the math yields 565 lumens / m^2. Should be a little more at standing (eyes ~5ft off the floor) height.
Is it possible that this is just a problem with LEDs that flicker[1] because they don't rectify the AC? The blurb didn't mention this problem, but presumably that's not what's going on here.
Normally, I would not have thought much about this kind of an article. But, I happen to be wandering around CES today and there was a booth somewhere that was promoting some glasses to protect your eyes from LED lights.
If I happen across that booth again, I'll stop and look.
Car headlights seem to have gotten much brighter in recent years. I know this is about domestic lighting, but... any concerns about car headlights and streetlights? Those tend to be much much brighter than indoor LEDs.
Looking through the literature it looks like it might be Dry AMD (macular degeneration) which is bad because it's currently untreatable. Wet AMD has some treatments available.
When new phones or displays come out, the manufacturer will invariably brag about the increased color gamut. Before I get blinded with science, it's time to adjust my f.lux and night shift settings.
Name me a thing, a biscuit, a light bulb, a gadget, a piece of metal, where there isn't some sort of agency that claimed that it increases the risk of cancer by 0.3% based on a study based on a sample of 8 patients and an opinion from the author's grandma!
1. 6000 lux is a _lot_ of light.
2. The LEDs tested are not the "warm white" (1500-2500K) commonly used as replacements for incandescent bulbs in the US. They're the "blue white" (>4000K) ones, or pure blue or pure green at specific wavelengths.
With those constraints, it's an interesting result. It does not mean you need to run home and rip all the lights out, though.