RGB LEDs for color enlargers

The full writeup is here: https://tinker.koraks.nl/photography/why-rgb-leds-suck-for-a-color-ra4-enlarger/
But let me introduce it and explain it a little bit below.

From time to time, I come across threads reporting on the use of RGB LEDs for color enlargers, which is something I've been involved in myself over the past few years as well. I did some experiments, built some stuff and for a few years now I've been using DIY RGB LED lights sources for all of my printing, both B&W and color. At some point I was satisfied enough with the performance and usability to pick up the courage and sell off the last dichroic head I had kept around to fall back on (but never did).

Anyway, there's not all that many people that turn out to DIY something like this, and when they do, I see mention of using RGB led strips, RGB COB LEDs or even WS2812 addressable LEDs. While I understand the attractiveness of those solutions (it seems so simple), I consider them fatally flawed. Today, I compiled my arguments against such an approach, or, if you will, why discrete LEDs are in my opinion really the only way to get this job done. It's all too much to replicate here, so if you're interested, give it a read on my blog: https://tinker.koraks.nl/photography/why-rgb-leds-suck-for-a-color-ra4-enlarger/

On that page, I share my thoughts on the fundamental problems associated with corner-cutting by using integrated RGB products, but I also show which combination of wavelengths works well (based on my own experimentation and daily use) and which combination is promising if and when the technology is available. I also discuss the challenge of using the same light source for color and B&W (particularly variable contrast) and why it's necessary to keep this dual purpose in mind when designing a suitable light source.

I don't expect this to generate all that much discussion, but I'm mostly sharing this here because when I started out designing a color LED head, I literally spent weeks and weeks searching for information in various places to figure out what I needed and what to avoid. Not to mention the hundreds of Euro's I spent on sourcing components and the countless hours engineering various solutions and testing them, also in the darkroom. Hopefully the page I wrote about it might save someone some time if they're about to create their own color printing contraption.

{Moderator's addition: koraks has updated this post in a further post - #89 in this thread, which can be found here: https://www.photrio.com/forum/threads/rgb-leds-for-color-enlargers.193779/page-4#post-2584666}

That’s a solid writeup. And some interesting results and findings.

Your link needs to be front and center since that is the meat of the post. Nearly missed it.

What did you end up doing?

And any photos of the process and equipment would be welcome.

It is a bit of a paradox, but your writeup is a really good thing to read for people who are camera "scanning" colour negatives, and struggling with the digital processing of the results.
I'm wondering whether this should be a Resource.

T

Helge said:

Your link needs to be front and center since that is the meat of the post. Nearly missed it.

Click to expand...

Thanks @Helge, I fixed that.

Helge said:

What did you end up doing?

Click to expand...

Well, it's a work in progress, but the current state of affairs is a light source made up of discrete LEDs, so distinct red, green and blue ones, of the wavelengths I mentioned after the initial analysis: 450nm blue, 525nm green and 660nm red. Due to the power levels required for both color and B&W, the only viable option was to actually use tiny SMD led beads soldered to an aluminium substrate PCB. This can be done at home with a little effort and some not too outlandishly expensive equipment, and indeed I did. So it's definitely feasible, and it's the setup I currently use.

As I wrote it, I developed an itch to try the alternative color setup of 480nm blue, 550nm green (if this can be found in the marketplace; I don't know yet) and 690nm red. Maybe, one day...

Helge said:

And any photos of the process and equipment would be welcome.

Click to expand...

I'd have to dig and make some new ones. I'm planning to do a writeup on my present color head at some point. Mostly for amusements' sake, because it's all rather overcomplicated. There are more straightforward ways to get the same job done. Except for the light source; that one is hard to simplify.

MattKing said:

It is a bit of a paradox, but your writeup is a really good thing to read for people who are camera "scanning" colour negatives, and struggling with the digital processing of the results.

Click to expand...

That's an interesting thought, Matt! I hadn't looked at it that way, but yes, perhaps there's something of use for camera scanning in there as well. I think that's actually a much more forgiving endeavor and I would be quite optimistic of getting that done, and done well, even with sub-optimal wavelengths. The reason is that you don't have to accommodate the sensitivity of the paper, and that's what's making it complicated. When camera scanning, you 'only' have to hit a somewhat suitable peak of the dyes. I'd have to look into that and go over some datasheets to develop an opinion on it. I never did any work in this area.
If you feel this should be a resource, please feel free to move it around etc.; you're more than welcome. I'm planning to keep the original blog post just where it is for a long time, and it's not in a place where it's dependent on the whims of a commercial platform in terms of URL changes etc.

MattKing said:

It is a bit of a paradox, but your writeup is a really good thing to read for people who are camera "scanning" colour negatives, and struggling with the digital processing of the results.
I'm wondering whether this should be a Resource.

Click to expand...

I don’t know about paradox. It’s very much the same thing that we are aiming to do in both processes.

Conceptually, it's similar, but most of what makes a color enlarger light source particular is actually the fundamental imperfections of RA4 paper. I really think digital is much more forgiving in this respect and that it's way easier to get clean channel separation when using a digital camera for 'scanning' than when optically enlarging onto RA4. Then again, I'm not a scanning expert and I'd gladly hear more about this side of the story if anyone wants to share.

The reason that this is of value to those who are never likely to ever see a colour enlarger is the fact that the film is essentially designed with the characteristics ("imperfections") of the RA4 paper and the peculiarities of tungsten light sources in mind.
The laser based additive light sources were also designed with the same considerations in mind.
The rationale behind why the orange mask is also a good thing provides similar valuable context.
The paradox arises because the way to understand how to move forward digitally (with camera scanning) is to understand what older technological choices were made and why they were made.

About 25-30% of the way down:
"Now, most of the RA4 paper that’s consumed today comes close to any dichroic exposure systems."
You might want to add a negative in that sentence, if I understand properly the context and intent.

Apart from this slight editorial remark, do you have nay similar views on the optimum wavelengths for B/W multigrade papers? For Foma, supposedly 2-emulsion? And for Ilford, supposedly 3-emulsion, any adverse effects from using just two wavelengths?

bernard_L said:

About 25-30% of the way down:
"Now, most of the RA4 paper that’s consumed today comes close to any dichroic exposure systems."
You might want to add a negative in that sentence, if I understand properly the context and intent.

Click to expand...

Wow, good catch...I missed that! Thanks, and you're absolutely correct.

bernard_L said:

Apart from this slight editorial remark, do you have nay similar views on the optimum wavelengths for B/W multigrade papers? For Foma, supposedly 2-emulsion? And for Ilford, supposedly 3-emulsion, any adverse effects from using just two wavelengths?

Click to expand...

Not yet, but I'll think about it. For now, I can only parrot what I read before, which is that apparently the same blue that works for color is also appropriate for VC B&W. I think I read somewhere (but I would have to dig around to re-find it) that regular 465nm blue doesn't allow for optimal contrast (grade 5-ish) on some or even most papers. As to Ilford and 3-emulsion papers: that's one I'd have to think about. I doubt it would be a bad thing to have only 2 colors for such papers, also given the existence of Ilford's 'own' 500 exposure system which relies on just blue + green.

MattKing said:

The paradox arises because the way to understand how to move forward digitally (with camera scanning) is to understand what older technological choices were made and why they were made.

Click to expand...

To an extent, and yes, I do follow your reasoning. But I think it would be appropriate to focus on the right aspects. When digitizing color film, you have basically three things interacting in terms of color reproduction: the light source, the dye clouds in the negative and the digital sensor. All three have spectral properties, and indeed those of the dye clouds are likely tailored to some extent to the response of RA4 paper (and/or the other way around..!) In that sense, you have a point. But frankly, if I were to tackle this one, I'd start with the properties of the dye clouds and the problems of crossover in digital sensors. The latter I think are decently controlled due to the proliferation of high-end digital imaging systems. I expect a lot of the color crossover problems associated with CMOS or CCD sensors to be either engineered out of it or circumvented with filters, digital post processing etc. So I think the real limitation, or rather, determining factor here, is the transmission of the dye clouds in the film. As I said before, I think some digging in datasheets would be appropriate on this front. And those aren't always forthcoming; for instance, the datasheet on Kodak Ektar does show a plot of spectral dye density (last page, top right), but it does not differentiate between the three dye colors. I'm pretty sure I remember they did do that in the past, but maybe I'm wrong...
Anyway, in response to what you said: sort of yes, but there's also much more to digitization question.

Edit: I just did some more digging and it seems that Kodak does present dye-specific density curves of their films for the Vision3 films. These are of course designed to be scanned and aimed at a public that's technically curious to the bone, so perhaps that's why they're more forthcoming with data in those datasheets. Still, you have to sort of reverse engineer the actual curves from stuff like the last plot in the Vision3 50D datasheet (last page, right-hand side).

Excellent article! Anyone intending to make a LED lamp for RA-4 printing should read it -- to avoid the pitfalls you discovered.
I made my own LED lamp by copying Mal Paso's design here, and he in turn used the same LEDs as JB Harlin as reported here. Below are my remarks about B&W and color.
My LED lamp uses the following LEDs available in star-form from ledsupply.com. You can find documentation on them here.


What I call "peak" above is called "dominant wavelength" (DW) in Cree's document. I suspect that DW is the centroid (weighted center) of the spectrum, which is not quite the same as the peak.

For black and white:
This combination of LEDs works well for both Ilford and Foma RC papers. A blue-only exposure yields the same contrast as Ilford's #5 filter. JB Harlin discovered that conventional blue LEDs could not reach maximum contrast, and that Royal Blue is needed.

For color:
I have never printed RA-4, but from the article, I can see that red and green would be less likely to cause crossover if their wavelengths were a bit longer. Royal Blue is probably okay, assuming its peak is on the short-end of 450-465. Perhaps we could ensure a short wavelength by buying specific bins.

Mark Overton

A few years ago I got this reply from Ilford about LEDs.

The maximum green sensitivity of our Multigrade materials is at approximately 530nm, and you should aim for an LED source emitting within about 5nm either side of this.

The blue sensitivity of the material is actually quite broad, and thus the LED wavelength is less critical, provided it is not so long that it is starting to hit the green emulsion.

As a starting point, I would suggest you try an LED emitting somewhere within about 10nm either side of 450nm.

It's rather premature for LED's to comprise any kind of serious replacement for a traditional halogen colorhead. As a potential replacement for cold lights in VC paper printing, yes. ... but color paper? I'm skeptical. I applaud experiments; but given all the significant investment in time and money I've already made in high output additive RGB enlargers based on narrow-band dichroic filtration and complex feedback circuitry, I intend to use what I've already got and not go down some new questionable rabbit hole. And even ordinary CMY subtractive halogen colorheads work so well for RA4 printing, and are quite affordable for up to 4x5 film applications, that I don't know what the incentive is to reinvent the wheel. Of course, if you enjoy tinkering, why not? But if you want serious performance with color paper, you need more accurate and more realistically adjustable filtration than current LED's will give you. It's not like better LED room lighting, where the blend fools you eye into thinking there's a balanced spectrum. Color paper dye curves respond best when you hit each nail right smack on the head.

albada said:

I made my own LED lamp by copying Mal Paso's design here, and he in turn used the same LEDs as JB Harlin as reported here.

Click to expand...

Yes, I'm aware of both projects. Very inspiring, and great work.

albada said:

JB Harlin discovered that conventional blue LEDs could not reach maximum contrast, and that Royal Blue is needed.

Click to expand...

Then that was probably the source for my statement on it as well. I knew I'd read it somewhere; it was probably there.

albada said:

Perhaps we could ensure a short wavelength by buying specific bins.

Click to expand...

I'm actually optimistic enough to believe that won't be necessary. For a hobbyist, it would also complicate matters a bit as you'd have to rig up a way to test small variances between individual LEDs. I suppose you could figure something out based on a diffraction grating. Interesting thought, actually...

Kilgallb said:

A few years ago I got this reply from Ilford about LEDs.

Click to expand...

Many thanks for this addition; that's actually very nice, as it confirms the LED selection at least for Multigrade papers! I also wonder to what extent it would be an issue if we shifted the green to the longer side a bit and the blue as well. For the green, it shouldn't matter except for an efficiency penalty, but the blue might become problematic if we shift towards something like 480nm.

DREW WILEY said:

And even ordinary CMY subtractive halogen colorheads work so well for RA4 printing, and are quite affordable for up to 4x5 film applications, that I don't know what the incentive is to reinvent the wheel.

Click to expand...

Drew, thanks for chiming in! Actually, the supply situation in my neck of the woods wasn't as optimistic as you make it out to be. I just couldn't find a suitable head, let alone at a halfway decent price and/or one that was likely to work or at least could be fixed if it didn't. After rooting around for over a year, I chose what was actually the easy way out...go figure!

DREW WILEY said:

additive RGB enlargers based on narrow-band dichroic filtration and complex feedback circuitry

Click to expand...

If you allow me to pick your brain for a bit - how did you determine the required peak wavelength for R, G and B as well as the permissible peak width (or the necessary attenuation at neighboring wavelengths)? What criterion did you use?
Btw, one of the advantages of LEDs is that the complex feedback circuitry is not as necessary as it is with a halogen light source with its inherent tendencies towards variation due to aging etc. It can still be employed, and it is actually easier to implement with LEDs because one could adjust the PWM duty cycle based on a mixing chamber measurement. I did not implement this, but for those who are concerned with absolute consistency over extended periods of heavy use, it might be an interesting avenue.

RGB are quite tightly defined spectral points with respect to not only human vision itself, but also engineering standards, involving 450, 550, and 650nm. With continuous spectrum "black body" sources like sunlight and tungsten lighting, various types of filtration can target these rather tightly. How tight one wants to go depends on a number of factors a bit too technical to get into at the moment. But if you're significantly off-center with any of these three due to a discontinuous illumination source; and one or more of these can't be tightly remediated through additional optical filtration, it's going to affect color accuracy. I'm not saying the result will be uninteresting, but somewhat unpredictable. So sure, you can print color using extant color LED's, but won't get the same kind of hue purity and saturation as with already proven devices (along with other tight process controls too). That situation will no doubt improve over time due to the demand for it.

The second issue is sheer candlepower. It's hard to punch a big print with colored LED's. But not everyone needs that. In a few minutes I'm going out back to enlarge a rather big RA4 print from an 8x10 color neg, and it's only going to take about 12 seconds at f/16; and that's not even my most powerful colorhead. But it's actually one of my most manageable ones because I've stripped it of ALL electronics, even the power supply. It's hard-wired. The less electronics, the better. Less to go wrong. I already have two other ones with all kinds of fancy feedback circuitry, and appreciate what they can do, but not when some electrical component decides to throw a tantrum and needs to be psychoanalyzed. I also like the fact of only two quite affordable, very reliable, and readily available bulbs involved.

With multiples of LED's, there's going to be more of a risk of the specific type going obsolete due to the relative adolescence of the technology itself. They might not last like you think they will. But pioneering new applications comes with risk regardless. Just be glad you aren't testing out some new cliff-jumping bat suit instead!

DREW WILEY said:

RGB are quite tightly defined spectral points with respect to not only human vision itself, but also engineering standards, involving 450, 550, and 650nm. With continuous spectrum "black body" sources like sunlight and tungsten lighting, various types of filtration can target these rather tightly. How tight one wants to go depends on a number of factors a bit too technical to get into at the moment. But if you're significantly off-center with any of these three due to a discontinuous illumination source; and one or more of these can't be tightly remediated through additional optical filtration, it's going to affect color accuracy. I'm not saying the result will be uninteresting, but somewhat unpredictable. So sure, you can print color using extant color LED's, but won't get the same kind of hue purity and saturation as with already proven devices (along with other tight process controls too). That situation will no doubt improve over time due to the demand for it.

The second issue is sheer candlepower. It's hard to punch a big print with colored LED's. But not everyone needs that. In a few minutes I'm going out back to enlarge a rather big RA4 print from an 8x10 color neg, and it's only going to take about 12 seconds at f/16; and that's not even my most powerful colorhead. But it's actually one of my most manageable ones because I've stripped it of ALL electronics, even the power supply. It's hard-wired. The less electronics, the better. Less to go wrong. I already have two other ones with all kinds of fancy feedback circuitry, and appreciate what they can do, but not when some electrical component decides to throw a tantrum and needs to be psychoanalyzed. I also like the fact of only two quite affordable, very reliable, and readily available bulbs involved.

With multiples of LED's, there's going to be more of a risk of the specific type going obsolete due to the relative adolescence of the technology itself. They might not last like you think they will. But pioneering new applications comes with risk regardless. Just be glad you aren't testing out some new cliff-jumping bat suit instead!

Click to expand...

Drew,
This is Very well said. I have almost nothing to add except this. RGB LEDs used for minilab printing are darn near in contact with the paper. You're gonna need some really freaking bright LEDs to match dichro color head lights.

And secondly RGB LEDs used to print minilab prints are printing corrected files, whether analog scans or digital files.

For color prints I still haven't found anything that out performs a conventional subtractive dichro head.

Best Regards Mike

Well ... a simultaneous additive dichro head can potentially outperform a subtractive one; but I can attest that those are a helluva headache to properly design.

DREW WILEY said:

RGB are quite tightly defined spectral points with respect to not only human vision itself, but also engineering standards, involving 450, 550, and 650nm.

Click to expand...

Which specific engineering standards do you refer to when you say that R, G and B are defined as 450, 550 and 650nm? The thing is, if you search in the academic literature on for instance blue-emitting semiconductor devices, you end up with anything ranging from 400nm to 500nm being called 'blue'. We've seen in a previous post in this thread that Ilford's green is actually 530nm. If you search the IES website (they have a convenient index of definitions), I find no signs of definitions of 'red', 'blue' or 'green', so apparently they don't have any definitions for these colors unless they're purposefully kept secret. If you go by Kodak Wratten filters to keep it closer to photography, there's nearly a dozen of them that are somehow labeled as 'red' and they all have different filter spectra and cutoff wavelengths.

Mind you, in the end I don't care one bit about engineering standards; I'd just like to know how you determined what the suitable wavelengths were for RA4 paper and how narrowly you hit them. If you'd rather not say, then feel free to indicate this.


Concerning this:

DREW WILEY said:

The second issue is sheer candlepower. It's hard to punch a big print with colored LED's.

Click to expand...

mshchem said:

You're gonna need some really freaking bright LEDs to match dichro color head lights.

Click to expand...

Yes, but really freaking bright LEDs are common today.

mshchem said:

And secondly RGB LEDs used to print minilab prints are printing corrected files, whether analog scans or digital files.

Click to expand...

If their peak bandwidth were a poor match for the paper's sensitivity, no matter how you corrected files, you'd never get good hue purity,. This is one of the reasons why I believe it's practically feasible to go the RGB LED route - it is already being done, and has been for years. Of course, it's conceivable that the entire industry got behind a technology that prevented them from being able to print spectrally pure hues, but I somehow have trouble believing that. There's only so much of a performance penalty that an industry can accept in return for better economics.

I presume these are just hypotheses. How did you test this? Where are the prints?

• The spectral emission of the red LEDs precludes decent color reproduction.
• The power balance between red, green and blue does not match the real-world requirements of color enlarging; the red is lacking.
• The power balance is also extremely unfavorable for B&W reproduction when using variable contrast papers; the green is lacking.
• Apart from the red and green being problematic, the blue seems to be off as well.

Click to expand...

ic-racer said:

I presume these are just hypotheses.

Click to expand...

Did you actually read the post you quoted them from?

ic-racer said:

How did you test this?

Click to expand...

Apart from the theoretical work, by printing color charts as well as real-world scenes for the color part, and by enlarging B&W negatives for the B&W part. I also used the LED system in various versions side by side with a dichroic enlarger when I still had one. At some point I got rid of the dichroic one because the LED version I arrived at made prints of equal quality and proved to be more versatile and easier to use.

ic-racer said:

Where are the prints?

Click to expand...

Discarded after having run these tests some 2 years ago. I might have some in a box somewhere, but no guarantees. When I did those tests, I was not maintaining a blog and just trying to get to a working color head as fast as I could. I'm now sharing my findings so others might benefit from it.

Feel free to run your own tests to verify or disprove my findings.

Koraks - I expect high quality results. "Quality" is of course a sliding scale, and what is acceptable to one party might not be to another. For quite awhile, jillions of snapshots were being made with automated LED printing systems. Everyone complained; but when you've got garbage going in, garbage is going to come out. But overall, many found it to be a step backwards from even cheapo old chemical one hour labs. The far higher quality option, laser color printers, are far more expensive to acquire and operate, but at least the various models are decently on target with RGB. It really does make a difference.

The basic problems of LED's are well known in architectural and studio lighting, as well as color matching applications. It's come a long ways, and will further improve. But I'm very suspicious of claims it's anywhere near being a serious option to halogen colorheads. A lot goes into any kind of objective comparison. It's not enough to make a few prints and say, looks good to me; or to take any ole colorhead with a worn out dichroic filter set. For decades, as a distinct part of my own career, I worked with industrial spectrophotometers and trained color matchers.

Just knowing how to judge color by eye takes a lot of experience and the right tools, and even at my older age now, I'm still getting better at it. It not just a physiological exercise, but requires a lot of knowledge of the psychology of color too, since our brains habitually re-interpret things we see into what we want to see instead. That like back when manufacturers did ads for film claiming "natural looking colors", knowing darn well people looked at sklntones first, and generally overlooked how fishy the background color was.

Even here, I have multiple light source options to judge color with : the best German color matching tubes one can buy at my retouching station, the best balanced fluorescent tubes overhead, the latest best-balanced LED display lights of appropriate Kelvin temperature, a bank of varible temp LED copy lights, all independently switchable, tungsten options too; and best of all, right out the door, our natural softbox daylight diffused through gentle coastal fog. I pretty much standardize my own color printing to 5000K (previously to 4200K); but if a client has their own specific requirements, I could go there and take lux and color temp meter readings, and print for their own architectural lighting standard if necessary. That rarely happens; and if have to accommodate something as miserable as CFL lighting, I just do the best I can.

And as far as "really freaking bright" LED's. Well, we're not talking about flashlights here. Just before I retired I started selling LED "inspection lights" which shed a broad flat beam against a surface, for example, a big custom table that's receiving a high sheen finish. Even the tiniest scratch or bit of dust will show up. Think of it as a $400 fancy flashlight with a $150 battery in it. But my own eyes are too sensitive to even be in a room where that kind of thing is being demonstrated. There's not practical use for that kind of thing in an enlarger. You still have to figure out all the filtration issues, and how to control them. Even the better little blue diodes in
LED strips being sold for sake of color matching need supplementary violet filters on each of them to correct a particular imbalance.

Drew, thanks, but I don't think we're getting anywhere with this. For some reason you ignore my questions; there's probably a good reason for this that surpasses my brain power.

No ... That's not the point. This is still somewhat an era of experimenting and jerry-rigging with LED possibilities in the darkroom. The lighting industry is the big driver, and namely in its own higher quality applications. But because the hunt for truer RGB performance underlies much of this, even with respect to our current complaints, that is going to steadily drive the R&D sector itself a particular direction. Sure, cool energy efficient light sources are desirable, but so is engineering simplicity and a proven track record. That takes time. In the meantime, continue to experiment and share the results. It all adds up to something beneficial overall.
I've tripped over plenty of my own chuckholes and logs on the road designing my own gear. One either enjoys that kind of shop experimentation or does not. I do.

Alright, so in that light of development and experimentation, would you mind commenting on my questions about appropriate prak wavelengths or filter cutoffs?

The driving forces behind the industry, the prevalence of led in interior lighting etc are all very interesting, but not really core to the issue. Yes, it influences what is available in the market place, but from my perspective I take that as a given.

The difficulties of assessing color especially in an amateur situation without high end equipment is highly relevant, but an unfortunate reality of doing this as a hobby. Note btw that the color deficiencies that I highlighted in the article did not require objective quantification because they were downright glaring even on casual inspection of prints.

And finally, yes, quality standards are not absolute. Yours are probably different than mine or the next person's. Hence my article should not be read as the absolute truth in how to design a LED color head. The somewhat careful reader will actually recognize this. It's also a blog post, not an article submitted to a peer reviewed journal. My interest in the topic is from a hobbyist perspective. I actually do enjoy the tinkering, much like you. I think we both suffer from the affliction of being interested in a broad subject matter.

For a few days, I've been thinking about the first graph in your article, which shows the sensitivity curves for R-G-B.

Blue is the challenge due to its large overlap with green. Measuring with a ruler on the monitor, at 450 nm, the B/G ratio is 2.8. But at the blue-peak at 486 nm, the ratio is 4.0. So I think your idea of employing a blue LED at about 486 nm is promising, because for a given exposure of blue, it will expose green the least. BTW, I got that 486 nm with a ruler as well. That max ratio of 4.0 should yield the purest hue.

The graph does not show the threshold (inertia?) of each color. If green has a higher threshold than blue, then an exposure at a ratio of 4.0 (486 nm) could produce a pure yellow.

Cree makes a cyan LED that is very close to 486 nm. Here's the link to it in digikey. It's 1-amp, so it's "freaking bright." This LED is a member of Cree's XE-G family, and here's the link. This family includes a 660 nm "photo red", and several greens. We could probably make a good RA-4 enlarger using only LEDs from this one family.

That's exactly why certain blue LED's have supplementary violet filtration atop them - to truncate the green contamination. But it's hard to predict how that kind of supplemental filtration holds up versus the kind of dichroic filtration found in typical colorheads. At least the heat issue is less; and excess heat actually shifts the spectral response of dichroic filters somewhat. But those also perform in relation to the incidence of the primary light rays.

Unfortunately, that Cree blue-green example looks like a poor candidate for a balanced color enlarging source. It might be nice for VC paper, however.

Matching light sources to printing paper spectral peaks is a somewhat involved topic in its own right. Not all RA4 papers are quite the same in that respect. Then you have to add the variable of how the dyes in color neg film itself not only interact with the enlarging light passing through them, but themselves differ quite a bit. No time now to even attempt to explain it, except to state that it can take a lot of experimentation to discover what works best for you, your own taste, and the specific components currently available.

Ideally, you don't want any crossover of the spectral curves of your primary light sources themselves. You get a bad enough problem with that in the dye curves of most color neg films themselves, and don't want to compound it.

True additive blue and true additive green do not mix to produce yellow, but block each other and produce black. It's not like working with tempera paint in kindergarten. But additive systems, if well designed, avoid the inevitable white light spillover of subtractive colorheads, which somewhat affects all the dye layers in the paper at once, and hence produce a bit of "mud" as I term it. Additive color is theoretically capable of cleaner color, but not if the primary RGB responses are displaced from their correct nm positions, and not unless the is full adjustment control of the intensity of each with respect to one another.