Hacker News Reader

Comment by atoav | Hacker News Reader

One thing: most LEDs suck at mimicking daylight. The cheap ones are bad at delivering the full light spectrum.

The higher end of LED is slowly getting there. The measurement one should be looking for there is CRI ("Color Rendering Index"). The sun has a CRI of 100. Any lightsource above 90 to 95 CRI is (to my experience) indistinguishable from daylight. The best stuff there currently is are (of course) the Skypanels by ARRI for laughable 6800 USD per panel (film equipment is expensive as usual).

Blasting a 2.5 kW HMI lamp ("Hydrargyrum medium-arc iodide lamp") trough a window from the outside is a good emulation of daylight. So good in fact that the poor souls shooting films inside will have their bodies in confusion as they exit the room and realize in horror that it is dark outside.

The electrical bill of anything emulating the sun is no joke tho.

View on HN

The trick my local bar uses is to put artificial lights behind a stained glass window. No matter what time you go in it looks like late afternoon, about an hour before sunset. Unless you have a watch or line of sight to the outer door when someone opens it, you can’t tell what time it is.

I’ve never seen behind the stained glass but am guessing they have warm light fluorescent bulbs (no heat) and something diffusing the light because it looks evenly distributed.

There are no windows on the outside of the building so it seems something you can set into a wall.

Our casino takes a different approach, block all the sunlight so that no matter the time of day, the artificial light is the same.

After a very long day playing poker, I walked into the Sky Ceiling area in the Venetian and had a really strong feeling that something was very wrong. (It was probably 2 AM, but the room is convincing enough to make it feel like late afternoon and the disconnect was strong enough that I had to leave the area.)

Very true, color rendering, or god forbid flicker, are the main culprits to me.

Over the last few years a ton of players have entered the film lighting space, with many offerings exceeding Arri's color rendering for far cheaper. Yuji LEDs or iFootage for example make some of the most accurate lights at 98+ cri.

The film industry will still go for Arris for other reasons, but at home we have lots of options nowadays.

Amazon listings will always claim ridiculous cri numbers, try to find a 3rd party measurement. https://indiecinemaacademy.com/complete-led-color-database-c....

https://youtu.be/qCyKdIgRVTk

>or god forbid flicker

That's just terrible drivers - likely an unisolated AC with a single coil. Flicker is trivial to defeat sacrificing the power factor (just add filtering caps). Overall there is nothing hard about using constant driven high-freq (over 30KHz, so no sound interference, either) LEDs.

Side note: in the days of home video calls, you can tell how many have terrible lights as the flicker is visible on the even camera.

The problem just is that it's really hard to find good LEDs. Maybe they're all being sold to industrial consumers, but it's been a pain to find good LEDs if you're looking for anything more "exotic". E.g. dimmable LEDs work and are available, but there is so much cheap trash out there while good brands are often sold out. And for many sockets, there's maybe one manufacturer offering it, if at all. I've had instances where only cheap vendors offered LEDs for a certain socket.

Hmm, It'd appear you wish to retrofit LEDs in existing fixtures. That's the rub - they are not designed to work well - there is so much space for a LED driver + LEDs + cooling + looking nice. Most LEDs are overdriven to show higher lumen numbers (and watts) on the tin, which means a short life. Dimmable LEDs with a standard triac based dimmer for incandescent bulbs is a bad idea overall.

The decent LED fixtures are designed to be a single unit, e.g. near ceiling light, dimmable w/ remote/blue tooth. Unfortunately, not possible to replace parts of them, and if they are poorly built - also overdriven. OTOH, they are trivial to repair with some electronics background - often times, the driver is replaceable entirely.

LEDs need to stay cool (below 60C) to have prolonged life - say 40K hours to stay at 80% brightness. That means large heatsinks, enough space to dissipate the heat.

I've already replaced dimmers to work with LEDs. But changing all fixtures would be way too expensive, it'd then be cheaper to just stick with Halogen. And I don't need any "smart" features, an LED that can be dimmed via an external dimmer would be enough. But they're harder to find than all these "smart" devices that might stop working after a few years and can send data to who-knows-who.

>I've already replaced dimmers to work with LEDs.

While not impossible - the LED driver is a switch mode power supply, most of them would be universal one taking 90V-260V with rather variable frequency. What dimmers do - they chop the sine wave, normally that would not affect too much the driver (aside made them hot or require more bulk capacitance, which they lack) - so they have to know that the sine wave was chopped and react by reducing the duty cycle, reducing the brightness of the LEDs. That complicates the driver, however if you don't keep the LEDs at full brightness and they are not hot on touch - they should last.

The other option would be communicating through the zero cross of the sine wave, but I doubt very much it's the case.

The external dimmer is not a smart feature but more like necessity, unfortunately.

Check these out:

[1] https://www.osram.com/ecat/Professional%20LED%20lamps%20with...

God, LEDs must be the worst scam foisted upon us by governments. Flicker, lies about the lifespan, and now you say I must replace the entire lamp when it fails rather than just the bulb.

I don't think there's any malice. They're far superior when it comes to energy consumption. And they generally work well. They're just part of a general trend that it's hard to find high-quality products anymore. Unless you have industry expertise, it's extremely hard to determine which appliances are usable. Traditional DYI stores sometimes provide that expertise, buying online it's nearly impossible to search for quality.

It is malice. The government wants you to consoom while simultaneously lowering your quality of life. "Traditional [DIY] stores" are something else the govt is desperate to get rid of.

LEDs are great when done well, corporations produce flimsy garbage. As if they have some interest in things breaking apart so you have to buy new ones.

> LEDs[' dark sides]

...Interference with medical equipment (this lab I knew found out that their EEG malfunctioned when the new lamps were on...)

Poorly built drivers, yet again - no EMI suppression. Virtually any switch mode power supply that has no suppression would have done that. The LED driver in virtually all cases is some SMPS.

I know one of the tricks to detect flickering: look at the emitter through a camera, e.g. that of a phone.

How would you instead check EMI - in general? (I am not a specialist in this area.)

>I know one of the tricks to detect flickering: look at the emitter through a camera, e.g. that of a phone.

This would be a really poor driver, effectively using the AC freq. and shutting off w/ each cycle/zero cross. Unfortunately, many are exactly like that. However - some cameras do have compensation for exactly this issue, and they may hide it.

>How would you instead check EMI - in general? (I am not a specialist in this area.)

Use an AM radio. (That would depend on the frequency and what not, but AM radios are a pretty neat trick to check)

"Take an electronics testing lab with you next time you go shopping to be sure you're not getting some cheap crap"

I miss the days of a restive wire shoved into the AC circuit.

I just want LEDs that don't noticeably flicker (i'm talking 3Hz here) after a year of operation.

My bet: terrible capacitors and the driver ran hot (which caused the laps to leak out). Bad driver.

Honestly I have no clue why LED light are so poorly built nowadays. There is absolutely no technical reason for it.

There are plenty of high-CRI LEDs available at reasonable prices. Bridgelux [0] and Yuji [1] are a couple of examples off the top of my head.

So I don't believe it's the LEDs that are the problem.

And yes, by modern standards they can suck some juice but even a large ARRI Skypanel only draws 1500W [2], which is similar to a small air conditioner.

[0]: https://www.bridgelux.com/

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

[2]: https://www.arri.com/en/lighting/led/skypanel/s360-c#F0.00S3...

And yes, by modern standards they can suck some juice but even a large ARRI Skypanel only draws 1500W [2], which is similar to a small air conditioner.

Why would you compare it to an air conditioner though? Those aren't equivalents. You need to compare it to an alternative light source. A bright LED bulb uses about 13W. You could have 10 of them and still be using less than 1/10 of the energy...

Typical LED lighting is approximately free, or close enough to zero that calculating a percentage change in cost isn't especially, ah, enlightening. The goal of the comparison is to attempt to contextualize the absolute rather than relative cost, relying on the monthly cost of air conditioning (approximately "a lot", or at least noticeable) being a more familiar value than the monthly cost of 100 light bulbs (which are individually so cheap as to vanish into noise).

The reason you'd compare it to a small air conditioner is because it's an amount of power that you can get from a branch circuit at home.

Normal indoor LED lighting is almost a rounding error on your electricity bill these days. Daylight simulation uses much more power.

But that probably emits at least 1000W in heat, so you'd then need that air conditioning, at least unless you only use it in winter.

It should be supplying closer to the full 1500W in heat, since any light it emits will just get converted to heat when it is absorbed.

But why would you compare when the goal is not to maximize your power usage before tripping a breaker. It is to get light.

Yes, the goal is to get light. And since these lights only use 1500W, you can plug them into an ordinary house electrical outlet in the US.

Yes, the goal is to get light.

Therefore a valid comparison is to measure it against other sources of light, and not air conditioners. But anyway, saying it's "only" 1500W is incredibly odd. If your goal is just to "get light" then using 115* more energy than a 13W LED bulb is completely stupid. I understand that you might want more light than one bulb gives you, but even scaling up to 10 bulbs you're still a huge margin away from a Skypanel.

If you're running an art gallery or a showroom where the quality, color, and coherence of the light makes a difference then it makes some sense. If you're at home and you just want 'some light' then it's utterly ridiculous to use that much energy.

> Therefore a valid comparison is to measure it against other sources of light, and not air conditioners.

I would definitely say that this is a false statement.

> If your goal is just to "get light" then using 115* more energy than a 13W LED bulb is completely stupid.

The goal is to get artificial sunlight. A 13W LED bulb doesn't do that. An air conditioner doesn't do that either.

However, it's nice to know how much power these lights use, so we compare them to other devices which use a similar amount of power. Air conditioners and space heaters use a similar amount of power.

> If you're running an art gallery or a showroom where the quality, color, and coherence of the light makes a difference then it makes some sense.

I don't think a SkyPanel makes sense in those applications. If you are just running an art gallery or showroom, get some ordinary high-quality lighting.

The skypanel is really really bright. The link says the 360 gets 80 lumens per watt, which is right around what LED bulbs get.

I added some Sunlike LED bulbs (the GE "sun filled" brand) to my office a few months ago. They aren't expensive, and they are nowhere near as bright as true daylight, but they do the trick.

Sunlike LED elements are available in different formats around the world:

http://www.seoulsemicon.com/en/technology/sunlike/casestudy/

If you're running 1500 watts in your office all day long you'll probably need a small air conditioner too!

not so small in the summer, too

A lot of people barely notice poor CRI lights. Heck people will mix bulbs of different color temps and not even bat an eye.

The issue is definitely the power usage. It’s not reasonable to dedicate that much power for the relatively minimal gain.

Also, CRI is a relatively poor measurement of color quality because it measures a very desaturated color palette. You can have two high CRI lights and they will still not match. There are newer color rendering indices like TLCI, SSI and TM-30 to account for this.

'A lot of people' here being 'Many American men, in engineering / programming who are often unaware'. A lot of the women I have worked with (in dev/engineers) notice and are annoyed by this kind of mis-match. Working in Europe, also most colleagues (men and women) noticed, and made fun of, companies/ stores / restaurants that were too unaware or cheap to notice the problem or fix it.

Light is the prime inputs to human sensing their world! It amazes me people can be so unaware of lighting conditions / problems, when that is how they do a vast majority of these sensing of the world.

OK… I know plenty of non-engineering people who can’t tell or bother either.

And it’s not simply being aware or not. We aren’t all born with the same senses. I can spot extremely minute color variations on tests but I have never been able to tell small differences in smell no matter how much I try.

What the heck? How is gender or where you live anything to do with it?

Even ignoring the higher prevalence of colour blindness in males, it's pretty widely understood that females have better colour perception.

https://pubmed.ncbi.nlm.nih.gov/21675035/ https://www.nature.com/articles/s41599-019-0341-7

Wouldn't mixing bulbs at different colors improve the color rendering? Unless they're the same underlying substrates in different ratios (diode+phosphor), I guess, but I don't think they usually do that.

Ultimatly, what matters is how the light spectrum looks. CRI is just an attempt at reducing that to a single number. Even then, many «daylight» scenes have very different light spectra.

While daylight definitely varies, I believe the spectra for the sun at noon at sea level is reasonably even and would be the main goal (as opposed to mimicking daylight in the morning or towards the arctic).

CRI doesn’t measure vibrant red very well and human skin has that red.

Good comment, thanks. Yeah you are right. I tried to simplify this as much as possible while still providing the gist of "various white light sources can have a different coverage of the light spectrum".

One thing about CRI is reflection though. When you use an LED with a high CRI then when it illumates your furniture, objects, people etc. it will look like as if the sun shines on it (in terms of colors).

But what I would like to have is indirect light. E.g. point the lights to the ceiling and have the ceiling illumate the rest of the room. Not only is this less efficient in terms of electricty (but there isn't really a way around it) but it AFAIK also impacts the CRI depending on what material the ceiling/mat is made of.

Any insights here?

High quality glass diffusors so at least the light is more indirect?

This made me think how little I actually think deeply about things.

I read an OP question, and was like, meh there must be something out there.

Then I read your answer, and was like, ohhhh this makes total sense. Of course you can't have equivalent of THE SUN for $50 and consuming couple of watts per day.

Do these lights also stimulate vitamine D production?

UVB is what's required, I believe. 290 to 315 nm.

Right, sunlight contains UVA and UVB and the UVB is what creates vitamin D. An indoor light that outputs UVB could be dangerous, the equivalent of laying outside in the sun all day, every day. It would also quickly degrade many materials in your house. UVA would trigger photosensitive eyeglasses turning them dark. Sperti makes a vitamin D lamp and it has a five minute timer and comes with UV blocking goggles (not sure what intensity it is relative to the sun) which suggests significant risk. So you may not want any forms of UV inside your home. Many windows these days have glass formulated to block UV to protect your rugs, drapes, and furniture. I'd question whether anyone would want the IR spectrum from an indoor sun-light also. I think what most people would want would be a high CRI light that has enough blue to trigger the brains time keeper. A bright indoor daylight-like light will make everything look very stark though (black shadows, objects brightly lit so you see all the fingerprints, scratches, and dust everywhere.) Your eyesight may degrade faster if you are constantly in brightly lit areas also. You may find that your autistic friends stop visiting you (or ask you to turn the lights off.) A lot of people who built A-frame houses with floor to ceiling windows at the end of the house found out that all that light entering 1) makes the house really hot 2) interferes with reading, watching TV, and using computers 3) fades their furniture and rugs. Even without the UV, bright indoor light has some issues, though it certainly can brighten dark spaces.

Reply on news.ycombinator.com