About lenses and anti-reflective coatings: a call for stepping up the current technology

Great idea! I almost always remove the AR coatings from lenses because of the negative impact on Duv and tint.

Forget about Edmunds-coated lens in a production flashlight.
But there are many manufacturers that can provide quality multi-layer coatings, also in China.
I agree with you that it would be a worthwhile improvement.

If we are to believe in the faithfulness of the anti-reflection coatings' graph, I cannot agree with the suitability of VIS-NIR or UV-VIS. They would work, I know, but they would induce some other sort of Duv and CRI distortions. Too many up and downs in the VIS-NIR or UV-VIS curves between 400 and 700nm, this is the reason. VIS-EXT and VIS-0, on the other hand, look more faithful, staying at ≈0.5% and below between 400 and 700nm.

Faithfulness matters.

Edmunds coated? Does Edmunds posseses the patents of the above shown coatings? Patents expire, doesn't it?

Now, onto the “there are many manufacturers that can provide quality multi-layer coatings, also in China” thing, I agree.

I was also thinking in multi-layer coatings before I stumbled onto the Edmunds article. So, I've taken some additional time to check the following:

Anti-reflective coating: Guide to Solve Glare Problem on your Product @ WeeTect

I am no expert in this stuff, but if the above graph is indicative of the effectiveness of a multilayer λ/4 MgF2 coating, it certainly looks good. With three layers at 450, 550 and 650nm the result should be great. Four layers looks very neat. With five λ/4 MgF2 layers at the proper stepping the result would be gorgeous.

Also googled BBAR anti-reflective coating. In Anti-Reflective Coatings – VAR, BBAR, AR Anti-reflective coatings of vacuum optics @ Torr Scientific Ltd. there is a noteworthy coating, the FS-BBAR-397-727.

Average <0.5% reflection over the visible range at 0 degree is a kind of high-standard that many manufacturers reach.
I don’t know how many layers does it take to reach it but I know that each layer is a different material, so MgF2 doesn’t repeat (if it is present at all).

BTW a coating optimized for incidence angles a little higher than 0 degrees would work better with flashlights - because (manufacturing precision aside) 100% of light hits at least one lens surface at angle.

I mean: light coming from straight under the lens passes through the lens straight. But there is so little of it coming exactly straight from the single point that it’s 0% of the total output. If light goes straight towards the lens from any other point - it doesn’t refract on the
first surface but refracts on the second.

With strong aspheric lenses the incidence angles may be quite large even when the lens is focused.
When a light is out of focus, the average angle is large.

With TIRs it’s quite similar.

With windows it’s a bit different, the light going straight ahead from the entire led surface will come out straight as well as some that was bounced off the optics.

But we can’t hope for manufacturers to design custom coatings for us…even if some actually could do this.
And coatings designed to work best at moderate angles are quite rare.

I agree those two coatings fall into a clear 2nd tier on that plot. I say that because I just assume that better performance = more cost or difficulty. Also, in a perfect world where LEDs didn’t have a tendency towards being green shifted and still having a blue spike, I wouldn’t say that… but they do. I think the placement of the peaks in these coatings could help reduce the peaks in the output spectrum that deviate from the ideal BB spectrum

Great post and replies.

so can we really tell a difference between old and new coating with a naked eye?

In my experience better performance quite often does not necessarily involves higher cost. For me, great performance and affordable cost is a very real thing. Things change and we evolve, its all in the amount of light we let in. ;-)

Concerning what you say about the peaks in led spectral power distributions, to reduce green and blue peaks the λ/4 MgF2 coating must not be set straight at these frequencies, i.e. not “tuned” for 455nm or 550nm. Maybe a multilayer coating set at 410, 500, 605 and 670nm, or sort of. Heck, it's almost 10 AM here and I'm yet to go to bed, so we'll see later LoL.

With a really good coating, pretty sure.

BTW, this is not necessarily a goal worth pursuing but when you look at a lens at angle, you see a colourful residue reflection. That’s how you tell that a lens is coated.

Some lenses have the reflection very weak and white which makes it pretty much invisible.
I remember Dale saying that when he showed his lights with UCL2 lenses, he would often get fingerprints on the lens…because people would feel an urge to check whether the lens was there.

I question lens thickness too. Every AR coated lens from 64mm diameter and up are 3mm thick. I would prefer 1.5mm or less.

I understood the UCL and UCLp lens from https://flashlightlens.com/ were superior in that they had less impact on duv and CRI - I think TK mentioned this a while back. Anyone know if that's still accurate?

They currently sell UCL (98-99%) and UCLv3 (98-99%) in glass, and UCLp (97%) in Acrylic. The v3 is AR coated Borofloat

Yes - the UCL and UCLp I've used in the past are super clear, but you can still catch a tinted color in a reflection off the lens.

I better understand now.

The VIS-number thing. The number is the angle for which it is optimized. VIS-45 would be optimal then, doesn't it?

I've taken a peek at the Flashlightlens.com website. That UCL™v3 seems great, any tests?

There's a ∅22.6 × 1.75mm available size, that conforms to the lens size listed for the Convoy S21A (∅22.8 × 1.5mm). However, although paying nearly €4 for a lens is a thing I could probably allow to happen, the shipping cost :facepalm: to where I live is a whole different story.

Yes, the number is the incidence angle at which the coating works best.

If you’re concerned about zoomie flood more than throw than yes, 45 would be about as good as it gets. Same with mule. If it’s zoomie with moderately strong (or weak) lens but you care about throw, the actual angles are going to be close to 0 degrees. Overall, something in between should work best.
If it’s a reflector light, beam is going to hit the lens (actually: the window) at near-0 angle. Spill - still less than 45 unless you have a very shallow reflector. I think for reflector lights the optimal option would be near 0. Considering that most coatings on the market are designed to work at exactly 0 - that’s not perfect but close.

Not aware of any tests on UCL's in general. Not sure if/where there are AR lens tests here on BLF. With all the crazy testing equipment we got here, I can't recall seeing any AR lens tests to show the changes in duv and CRI.

I've only tested for lumens and throw. I've seen maybe 2% to 5% bumps. 5% more typical from regular non AR glass.

0K but, does it really matters? Where are those VIS-x lenses, after all.

So there is a myriad of AR-coatings out there. 0K, maybe not a myriad but a nice bunch.

The question is what can we aim to get a chance to see manufactured. Since I guess λ/4 MgF2 is public domain, we could probably get someone interested in manufacturing :???: 4-layer λ/4 MgF2 coated lenses. As I see it that coating covers ±40nm very well, and so I would specify the 4 layers to be optimized for 430, 510, 590 and 670nm. That would deliver balanced performing lenses. Or it should, doesn't it?

That’s a good question. I’m truly not an optics expert and my intuition is relatively (on my personal scale) weak in this area. I don’t see how layering works honestly. I could only imagine its extremeley irradic and sort of an empirical method of discovery. What I mean is that my assumption is that the behavior of two layers is not an additive function of the two individuals, but rather something significantly different.

I’m going to buy some UCLv3 lenses and see how they look. I’ve only heard great things.

Irradic? Do you mean erratic?

Reality is created from above to below (as above so below). It all starts with an idea, a thought pattern which is energized and eventually materializes if all its requirements are met. Thus, when something “is not known” it is a lot better to stay in a positive mindset. It's extremely important if the aim is to succeed.

I didn't create the λ/4 MgF2 coating or any of the other ones. I think multiple layers of such coating side by side results in a “flat” reflectivity curve at ≈1.3% reflectance, which is the lowest reflectance value delivered by the coating. It may be argued that ≈1.3% isn't the best thing, but everything between 400 and 700nm at ≈1.3% is excellent consistency; to preserve Duv and CRI consistency matters. For this reason I was speaking about the VIS coatings above, and also about the FS-BBAR-397-727 from Torr Scientific Ltd:

Now that I look at it more carefully, if the graph is accurate I'm not that sure. Performs best “in the green”, around 515nm. Despite performing better overall, green is not what we want.

I have noticed myself that AR causes change in output tint, never knew it will cause such a shift in duv.
Very useful info, lets get rid of AR coating and use sapphire lenses :money_mouth_face:

Anti-reflective and lens loss are new domains and I’m trying to understand / calibrate my lightbox.

Just so finished making my own lightbox and I’m using a shutter diaphragm to simplify the flashlight’s different diameter insertion. But as I’m calibrating and checking against some lights, the lens’ loss varied from 8.8 to 14.9%

Some pics:

I find the loss quite high and too much disparity.

Is it normal for such losses? As I was taking the pictures, I saw that at any incident angle the shutter diaphragm wouldn’t show - too much reflectiveness. Without a backlight and some fumbling, I did get that one picture.

And as for the variations of loss, although I take care to position the light dead center and open the diaphragm just a 1mm or so around the light’s window diameter; yet I get varying losses from Convoy S2+ (4500ºK) to Sofirn’s SP33 (V2 - CC driver) (5000ºK).

Yes. Erratic. Thanks for the correction.