Hacker News Reader

Comment by alister | Hacker News Reader

Questions:

(a) Does this study sound plausible?

(b) Would staring at the correct red color on a monitor work? Or would that not work since monitor colors are generated by mixing R, G, and B pixels and therefore not a pure wavelength?

View on HN

As to question (b): the wavelength is not the same.

The red primaries in most color spaces, including wide-ish stuff like DCI-P3 is around 615. Even Rec 2020 is just 630.

https://clarkvision.com/articles/color-spaces/

Note that it's fairly common to see chromaticity diagrams labelled with wavelengths around the edges, so you can check multiple sources if you want.

I didn't read the original article in sufficient depth to tell you if the researchers even know what 615nm would do, but it's not 670nm anyhow ;-).

Incidentally, the linked article also includes measured spectral responses for a few wide gamut screens, some of which have at least some response near 670nm. The most extreme on that page was the Del 3007WFP-HC LED, which peaked at 653, and has a wide peak hitting well over 700nm too (but below 650 too). A 2019 model Samsung Q80R qled TV (perhaps more common), had a peak at 632 nm, and at 670nm is about 20% as bright as at its peak.

Another question:

(d) How much does the precise wavelength matter? The study used 670nm but also mentions the range 650-1000nm. Lots of red-light gadgets are available on Amazon but how do you verify the wavelength? I was thinking of an optical spectrum analyzer[1] but at $28,000 they are shockingly expensive. Even with raw LEDs bought from a reliable electronics distributor like Digi-Key, it would be nice to be able to measure it somehow.

[1] https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=52...

You can get a 1nm resolution "science grade" spectrometer for under £1k - I used to design them [1]. You can also buy used ocean optics kit for very reasonable prices on eBay (a few hundred for a USB2000 if you wait).

Or make your own? It'll probably be good enough https://publiclab.org/w/spectrometry

ThorLabs also sell cheaper spectrometers by the way. You don't need a spectrum analyser with seven significant figures! https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=34...

You can also get light testing devices used for film/industrial illumination https://www.sekonic.com/color-meter/c700 (about 1.5-2k)

Or even single chip spectrometers with broad 20nm resolution from Hamamatsu (about £400 on group gets). https://groupgets.com/manufacturers/hamamatsu-photonics/prod...

[1] https://is-instruments.com/products/miniature-spectrometer/

You might find this of interest - it seems to be a cottage-industry spectrometer selling for £67: "The i-Phos can see wavelengths from approximately 420 - 980nm and their relative (though not their absolute) intensities." http://chriswesley.org/spectrometer.htm

I suspect you do not need to use 670nm exactly, however I suspect you need to be above 650nm. Note that many red LEDs are at a lower wavelength (~630nm). So you need to search for deep red or by wavelength.

You can use a OSA, but yes they are expensive. Spectrometers are typically cheaper, but still expensive (you could go to a university optics group and ask them if they could measure it for you). That said if you buy from a reputable source you should get the right wavelength.

If you are trying this out be careful with the brightness.

Instead of trying to emit 670nm, why not just use a wide spectrum source and a 670nm bandpass filter?

Like the sun and your closed eyelids. Light transmission goes up with wavelength https://www.spiedigitallibrary.org/journals/journal-of-biome... (see figure 5.) To speculate, perhaps we have evolved to benefit from closing our eyes in the sun.

That's a great connection. If it's that simple (and explained by human evolution), that would be awesome.

I got one of these $25 kits in a physics class that gives you the eyeball version of the fancy digital spectrometer: https://shop.sciencefirst.com/starlab/kits/5800-cardboard-sp...

We like these (or rather, the whole line):

https://www.avantes.com/products/spectrometers/compactline/i...

You can get them used for below 1000$, eg this listing:

https://www.ebay.com/itm/Avantes-Compactline-Avaspec-Mini-/3...

Shouldn't it be quite cheap to determine the frequency using a prism, using the angle of refraction? I think you should be able to use known missing frequencies in sunlight to calibrate your set up. Back in a minute going for duckduckgo

So it looks like you need a bit more than a prism to see Fraunhofer lines, but there seem to be online description of how to do it with a with a prism and a CD, maybe you don't even need a prism, there seem to be 'make your own spectroscope' tutorials than mention building one with cereal box, and maybe a lens of some kind

The D line is probably more easily seen in a colored flame made by adding salt water to an alcohol-burner flame:

https://en.wikipedia.org/wiki/Emission_spectrum#/media/File:...

that sounds like a project to do with my kids, thanks : )

Most LED have their wavelength in their datasheet.

Just select one that has the one you are looking for, it's going to be cheaper and easier than getting an expensive or inaccurate spectrometer.

For instance, this classic one is at 660nm:

https://www.mouser.com/ProductDetail/Lumex/SSL-LX5093SRD-D

(a) not a medical person so can't answer this

(b) unlikely, monitors work by essentially filtering out undesired colors from white light. I don't believe that >650nm light is within the color gammut of monitors. That's quite a long wavelength and close to infrared. To give you an indication, the Helium Neon lasers which used to be very common in schools etc for laser demonstrations have a wavelength of 632nm. Even if the monitor could display this, the brightness would likely not be strong enough.

There are several different approaches to monitors, OLED RGB sub pixels individually emit light. https://en.wikipedia.org/wiki/OLED. That said, your wider point stands.

C is the correct question to see, if this actually works

I have one of those aviator LED flashlights that has the red LED for chart reading at night without destroying your night vision.

I'm going to give this a try!

Please be careful, and stop as soon as you notice any discomfort.

Might be a good idea to stare at the light reflected from a white sheet of paper. The wavelength won’t change but the intensity (and FOV coverage ) certainly will

How about putting a diffuser (like a sheet of paper) in front of it?

Of course! The red LED is actually fairly dim, it's designed for pilots trying to read charts while flying at night!

I got it for use with a telescope for looking operating a telescope without losing night vision.

This is the hex value for 670 nanometre red.

#ff0000

Per https://academo.org/demos/wavelength-to-colour-relationship/

This only makes sense on a precisely calibrated display. On everything else, there's no way to tell what wavelength you actually end up with when you fill your screen with a color.

A perfectly calibrated display might still not be able to display pure wavelengths. At the very least if it's calibrated properly it won't normally do so at #ff0000 as pure 670nm is well outside the sRGB colour space.

yes but when i drag the slider i see that 645-700nm is also ff0000.

i don't know how precise it should be..

I imagine the brightness of a monitor is not enough for there to be any effect but maybe I'm wrong?

I would assume one needs a very good laser+DLP projection system to make this work.

Reply on news.ycombinator.com