Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

This is one of those topics that I researched a lot when I bought a new home recently. While LED lights seem awesome at the first sight (low power consumption! No warmup time!) when you dig a little deeper you'll find that most LEDs are far from perfect:

1. Lots of LEDs flicker. Shitty ones with twice the frequency of the main current (so 100/120 Hz) While you might not notice this it causes eye strain and can lead to eye fatigue, headaches, migraines.

2. Most LEDs have a low CRI (Color Rendering Index, or called Ra which is one standard to measure it), this means that they do not emit light in certain frequencies causing objects to look dull/gray. Most LED lamps have a Ra value of 80, meaning that they emit on average in the 80% of a spectrum that covers most of the visible light.

If you look around you'll be able to find reasonably priced LEDs with 90+ Ra and no flicker, but its a pain in the ass to research it. Especially if you need something more special like a LED light strip for your kitchen.



> 1. Lots of LEDs flicker.

No, the diodes don't, the drivers do. You are speaking to a technical audience ;)

> means that they do not emit light in certain frequencies causing objects to look dull/gray. Most LED lamps have a Ra value of 80, meaning that they emit on average in the 80% ...

Hardly. It means that they reproduce a set of 8 or 16 fixed frequencies from the ral pallete. It doesn't cover anything inbetween so it's just a rough rule of thumb. It doesn't cover colors in the far red (or was that violette on the other end of the soecteum?) at all.

Overall, overexposure of some frequencies, arguably to appear brighter and fuller, has damaged musuem pieces--Radiation stress, really. All they can do now is individual tests with photometric instruments, because they can't trust CRI. Same for horticultural lighting, which hasn't really taken of anyway.

Edit: Some manufacturers publish reference curves in the datasheets for their emitters, however I don't know how accurate these are under varying conditions and across batches.


Having made a cheap DIY spectrometer[1], I found the spectrum of LED around my house quite good, not completely smooth as sunlight, but smooth enough (you can gain some efficiency for not being totally smooth by better matching eye receptors). It's a far cry from fluorescent spectra that have extremely sharp concentrated spectra (and I suspect the risk from narrow spectra of causing biological damage is greater, since it could in theory disrupt very efficiently some molecular energy transition that happens to be well matched with that frequency).

[1] See e.g. https://ocw.mit.edu/courses/electrical-engineering-and-compu...

https://lavinia.as.arizona.edu/~mhammer/outreach/cdSpectrome...

(diffraction spectrometer from used CD/DVDs, very easy to make, and it costs ~$0)


Could you explain what you mean by ‘disrupt molecular energy transition’? Would be interested to know.


I'm not an expert, so take it with a grain of salt.

Electrons in molecules have discrete energy states they can occupy, per quantum mechanics (the discrete set of wafunction solutions)[1]. Consequently they can only transition between states by absorbing or emitting photons of specific energy values (I believe it doesn't have to be exact due to scattering and various QED dynamics, but it has to be very precise). If your molecule is exposed to broadband light, a few photons will match those ionization energies, however the quantity should be small[2]. If your light source contains peaks though, and there happens to be important molecules whose energy transition matches your source (a big if), you're going to get a lot of ionization.

I should have mentioned a more established issue is fluorescent lamps emit more blue/violet and UV than LEDs (LEDs are usually packaged in plastic anyways, so even if the emitted there'd be no risk of uv exposure). In common molecules the more prevalent ionization energies are UV and beyond, that's why the risk is greater.[3]

[1] https://en.wikipedia.org/wiki/Energy_level

Electron states are central to molecular dynamics and chemistry.

[2] Infinitesimal ideally, just small in reality because of tolerances as mentioned

[3] https://en.wikipedia.org/wiki/Fluorescent_lamps_and_health#U...

https://www.scientificamerican.com/article/can-compact-fluor...


That's how photosynthesis works, too, only the electrons are free to move in a wider band [http://www.ucmp.berkeley.edu/glossary/gloss3/pigments.html]. But there's more to the whole process, other molecules and what not.

Plants are known to have been "burned" by LED, and not due to heat. Some turn pale if they don't get a broad spectrum, if I remember correctly.


Interesting. Thanks. I had no idea that compact fluorescents let out so much UV.


If you want to check for flicker, there is a great app called VISO Flicker Tester


Unfortunately the app seems to be broken on iPhone 8+. Fails with "App could not detect camera."


It doesn't detect the camera on my iPhone :(


> No, the diodes don't, the drivers do. You are speaking to a technical audience ;)

I've seen 'drivers' that are nothing more than a resistor in series with the LED. So the LED, acting as a half-wave rectifier, causes the flicker. To your point, in this case the LED should be considered part of the driver.


No, the LED itself literally flickers. Now, that may be because of a shitty driver design... but it’s not wrong.


Thanks for the corrections, yep I was a bit sloppy. But I hope the main point came trough - which is that you should choose carefully when buying LED bulbs, lots of crappy ones on the market.


> No, the diodes don't, the drivers do. You are speaking to a technical audience ;)

Pulse width modulation. ;)


> 1. Lots of LEDs flicker. Shitty ones with twice the frequency of the main current (so 100/120 Hz) While you might not notice this it causes eye strain and can lead to eye fatigue, headaches, migraines.

That's with a full rectifier. It's even cheaper to go half-rectifier and end up with 50/60Hz flicker. Being cheaper, this happens a lot.

I'm super sensitive to this flickering and will get nausea after sitting in some cheap LED lighting while other people hardly seem to notice.

I jokingly say that with cheap 50Hz LED lighting and 60fps screens, virtual reality now has a higher frame rate than real reality.


This use of half-rectified lighting is especially true of LED Christmas lights. I haven't yet dug in on how to change them to at least full-rectified, which look much better to my eyes.

Is it just a matter of adding a bridge rectifier? I guess I'd need to add some resistors in there also to dim the emitters since they'll be on twice as often (to prevent premature burn-out).

Has anybody looked into this? I'll admit to being a bit eccentric.


How does this effect compare to old fluorescent tubes? Where I live we still got a lot of those around (specially in government offices, which is as cheap as you could imagine, and then some), so LEDs seem like a win, even when horrendous.


Fluorescent tubes flicker at 100/120 Hz, but the flicker is less pronounced because the phosphor keeps giving off light for some time after the electrical current stops flowing.

White tubes are made of a mixture of phosphors, and each of them has a different time constant, meaning the flickering is noticeable as repeated changing of colors. Take a photograph (with a rolling shutter phone for example) of a fast moving object under fluorescent light and you'll see blue and yellow shadows from the changing light[1].

[1]: https://photo.stackexchange.com/questions/60701/why-do-i-get...


LEDs also use phosphors -- in that way an LED is like a miniature fluorescent tube. Just like fluorescents, LEDs (at least those used for illumination) have an undesirable emission spectrum that's converted to "white" with a phosphor coated enclosure. (LEDs generally have a very narrow spectrum, and the GaN system used for lighting can produce blue and shorter wavelengths, depending on doping.)


They are 100Hz, as I found out while trying to do an optics experiment, but they never bothered me directly. I think it has to do with LEDs going completely dark, where fluorescent just has slight fluctuations. Or maybe it is a 50Hz vs 100Hz flicker thing.


> If you look around you'll be able to find reasonably priced LEDs with 90+ Ra and no flicker, but its a pain in the ass to research it

For people in North America, I'd recommend https://www.hyperikon.com/ for LED products from screw-in bulbs to large area lights with high CRI and no flicker (note: a few products don't have high CRI, so do check the product page). They offer a good range of color temperatures too.

> Most LED lamps have a Ra value of 80, meaning that they emit on average in the 80% of a spectrum that covers most of the visible light.

That's not what Ra means. It's not a percentage - in fact, something with a perfectly flat distribution covering all the visible spectrum would not have especially high CRI, because that's significantly different from the spectrum of blackbody radiation, on which CRI is based. The actual algorithm for calculating CRI doesn't fit nicely in a comment, but it's in the Wikipedia page: https://en.wikipedia.org/wiki/Color_rendering_index

Edit: missed the negation - CRI is not high for a perfectly flat spectrum.


I just replaced all the lighting in my House a few months back with Hyperikon LED lighting, and really felt like it was an improvement over the previous CFL lighting.

This comment alleviated some of creeping anxiety from the above posts that I maybe hadn't done enough research (I hadn't, I just got lucky).


Note that CRI (Colour Rendering Index) is a bit bogus. It came about when florescent lights became a thing and was originally based on a sheet of printed colour swatches. RGB LED lighting promoters have pointed out that CRI doesn't really reflect the goodness of light sources other than florescent in practice.


I'd say CRI is a bit bogus in that a light source can have fairly high CRI and still look bad. For example, I've seen LEDs used in a flashlight with a CRI over 90 that had a significant green tint.

On the other hand, I have not seen a light source with low CRI that looked good illuminating people, natural materials, or anything else where subtle differences in color are relevant.


Couldn't we use something like KL divergence to measure the difference between a given bulb's spectral distribution and a perfect black body?


There are several other ways color rendering can be measured, but the ones used in the lighting industry tend to be based on comparing the rendering of specific samples much like CRI. One that's seeing a lot of use lately is IES TM-30, which is based on a much larger number of samples and includes a measurement of color gamut as well as fidelity.


I'm very happy with yuji [1] products. They have all kind of high CRI LED based products, including light strips.

[1]: https://www.yujiintl.com


I wonder if someone has curated a list of good LED bulbs with high Ra and no flicker. I have been reluctant to go 100% LEDs at home.


This is a fantastic site that tested the spectrum and flicker of LED bulbs. The author seems to have stopped, but I think it’s still useful.

http://ledbenchmark.com/


Lutron (the inventors of the dimmer) have a pretty comprehensive database. They test a lot of lights, and not all are listed. If you're buying their dimmers, call and ask if you have specific lighting pairings in mind.

http://www.lutron.com/en-US/Pages/LEDCompatibilityTool/Compa...



We went 100% LED back in March 2014 with Philips CorePro bulbs and haven't had any issues. Either they don't have these issues, even back in 2014 (they were certainly highly regarded at the time), or perhaps my wife and I just aren't sensitive to this flickering.

Edit: Incidentally every single bulb (15 of them) is still working too, we haven't had any issues with them failing.


I have no issues with Osram Lightify GU10 bulbs (EU version), on the other hand I have proprietary fittings with Cree LEDs and Osram drivers that do flicker. So even in a single brand there can be issues...


The Wirecutter reviews a few, but it's not a complete list: https://thewirecutter.com/reviews/best-led-lightbulb/ .


Regarding flicker: Use the slow-mo video function of your smartphone camera to check for it.


Some cameras have slow-mo modes which try to remove lighting flicker by having a special high sample rate light sensor whose purpose is picking up the flicker and then using software to compensate the rest of the image data for the flicker pattern, using as input the time the shutter opened and closed for each row of pixels of a video frame, and integrating the light sensor value over this time.


This; state-of-the-art mobile cams should be "beating" the flicker, even in slo-mo. I'd wager the latest iphones can, although I don't have one to test.

A lot of places use fluorescent lights which will look real weird if caught in the frame without some kind of flicker compensation just as one example so its pretty important.


Uncited rumour is iPhones for the last ~3 years have had special flashlight LED's (made by NXP?) which also embed an RGB colour sensor for exactly this reason. The LED's when reverse biased send out the results of the RGB sensor. It has been suggested the reason they didn't use an extra pin was to hide this anti-flicker feature from competitors who were sure to do a detailed teardown of the phone.

There are x-rays of the LED here: https://www.chipworks.com/about-chipworks/overview/blog/true...


Or just look at your finger while you wiggle it vigorously.


What would you see with, and what without flicker? I can wiggle my finger vigorously, but now I still don't know whether my LEDs flicker...


Maybe not, but if you do it hard enough while listening to some terrible music, you'll have replicated my 90s clubbing experience.


With a continuous (non-flickering) light you see your blurred finger. With a flickering light, you'll see your blurred finger with gaps in the blurring. This effect is especially pronounced with sharp-flickering light (like in LEDs) as opposed to gradually-flickering light (as in incandescent bulbs). CCFLs are somewhere in the middle.


You'll see extra discrete instances of your finger, as though lit by a strobe light.

This is how we used to be able to measure speed of spinning objects too. :)


I did this during my quest to find a flicker-free LED light bulb. I found that basically all of them flicker. Some appear to alternate between banks of LEDs to reduce the dynamic range(?) of the flickering, but they all flicker. Home Depot has a rack of LED bulbs that are always turned on, so that's a good place to test many models at once.

My first thought was: why don't they just filter the output? My guess is that perhaps LEDs are more efficient when being driven full ON or OFF rather than with a steady DC.


> My first thought was: why don't they just filter the output? My guess is that perhaps LEDs are more efficient when being driven full ON or OFF rather than with a steady DC.

It's actually the opposite, efficiency goes down when current goes up.


Mine are all Brightgreen [1] high-CRI warm white downlights and they’re pretty great. They’re pretty pricy compared to the cheap no-name brands though (going for the “designer”/premium market I suppose).

1. https://brightgreen.com


> Lots of LEDs flicker. Shitty ones with twice the frequency of the main current (so 100/120 Hz)

I'm wondering how annoying those might be for a dog.


Is there a reason a dog would be more sensitive to flicker than a human?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: