Convoy S21D w/ legendary Nichia 219b :) - Review & comparison w/ Nichia 519a, E21a & other lights with Nichia LEDs incl. Emisar D4V2, Convoy S21F. (Summary & measurements on P. 1)

In comparing Duv number, I’ve noted that if the change is in the fourth number after decimal point, for example
“No Lens”: –0.012 7 versus “Convoy Green-reflection”: –0.012 4
it is not easily noticeable on the Duv graph.

What this means is, at least per my test, unless I develop Superman’s vision, I am not sure that I could see a difference between for example No Lens versus Convoy Green Reflection versus UCL Blue. I plan to take beamshots against the wall of truth to confirm if this is true.

OTOH UCL Purple versus Convoy Green Reflection is a bigger difference (it’s on the third number after decimal point) and likely will be seen on Duv graph.

Visually if you compare some 519A lights with different lenses, the difference in DUV is visually noticeable. However if you then add other emitters into the mix such as SSTs and LH351D, then you realise that even the “worst” 519A/lens combination is very neutral and perfectly acceptable.

It's difficult to see a duv difference when both outputs are way into the magenta region--the percent difference between the two is tiny compared to their deviation from the BBL. (Imagine trying to discern between 1 and 10 lumens, and between 1001 and 1010 lumens. The absolute difference is the same, but the percentage difference is huge!)

I also suspect that the green AR lens would make a larger duv difference numerically with a light source that starts off as neutral or slightly green. If the emitter starts out magenta, there is little green to begin with, and not much removal can happen.

It would be lovely if the test could be re-done with emitters that are neutral or slightly green.

Djozz’s quick test

Thank you Haukkeli for linking the test! Looking at -0.003 decrease in duv from swapping purple to green AR. That's a LOT of green reduction if the emitter starts off close to neutral. Also, the +8 increase in R9 is really something.

I would actually expect a more drastic difference for Convoy lenses--their purple reflection, at least in my sample of Convoy lenses, appears much more intense than the lens shown in the D80 photo.

Thanks everyone for your input and @Haukeli great job for finding that post of Djozz’s test - thank you. It’s great to see someone else’s numbers to know I am not entirely lost lol. At any rate after looking at Djozz’s results, I have just rerun my test not just with 219b but also with 519a in my S21D using 60 Bead TIR optic, and could now confirm with near certainty my result :innocent: .

For Convoy S21D with Nichia 219b 4500k:
1. No lens: –0.0127
2. Convoy AR Green-reflection lens : –0.0124 CCT 4471
3. Wurkkos Non-AR: –0.0123
4. UCL AR Blue: –0.0123
5. UCL AR Purple: –0.0102 (the Purple Lens’s Duv is “20 points” (0.0020) more positive) CCT 4365

Note that we know changing lenses affect Duv, but not frequently mentioned is it also shifts CCT about 50-100 points.

Where my result is different from Djozz’s - I found essentially no significant difference in Duv among set-up 1 to 4 above.

Where my result is similar to Djozz’s - The UCL Purple AR turns Duv approximately 0.0020 more positive (note the decimal point) and drops CCT roughly 50-100 K versus the group of 4 above. Not a huge change numerically, but definitely there, and does bring beam into more yellow-green territory.

Thank you @cannga for re-doing the test! Would you share the numerical results for the 519A? (I assume the graphic you shared is for 219B given how low the duv is.)

I swapped out a purple AR coated lens for a non-coated one in my 519A 5700K triple and noticed a huge difference right away. In addition to eliminating much green, the tint shift is a lot less, due to the fact that an AR lens does not alter tint equally in all directions. It alters tint the most near the center of the beam, but barely does anything to tint near the periphery of the beam.

I just ordered some green AR lenses and really hope they will bring the duv down more and eliminate angular tint shift completely.

You’re very welcome. Data from S21D with 519a 4500K and 60 Bead TIR below, very similar behavior to 219b 4500K. Again the UCL Acrylic Lens with Purple reflection is an outlier and the Duv shift is again around 0.002, CCT drops around 50-100. I could provide x,y coordinates if you’re interested.

S21D 519a 4500K with 60 Bead TIR
1. Convoy Green Reflection AR –0.0038 CCT~4041
2. No Lens –0.0037
3. Wurkkos Non-AR –0.0034
4. UCL Blue AR –0.0032
5. UCL Purple AR –0.0018 CCT~3985

Above bolded part in your post very interesting; I didn’t know that. Was about to write that any effect from an AR lens would not affect tint shift since it alters the whole field. Always good to read a second time before responding :person_facepalming: :innocent: . But question: if AR doesn’t alter peripheral field, it wouldn’t help with the peripheral tint shift in this Nichia would it?
I actually found the tint shift in my 519a 4500K not that bad especially w/ wide & beaded TIR and in actual use (vs white wall beamshot). Compared to my beamshot is it worse in your 5700K version?

1 Thank

Thank you so much for the data! I must say that I am very surprised by how little the green AR on Convoy lens changes the tint.

Regarding your question: I think green AR helps with tint shift precisely because it does not affect the periphery much. In a typical TIR setup, the beam tends to be greener at the center and pinkish at the boundary. A good green AR lens should take out much green from the center but not remove green from the boundary, thereby eliminating tint shift.

The 519As actually start out without too much shift (way less than the 219Cs that were in my light before), and however much shift that does occur cannot be accurately captured by a camera, so I couldn't make a reliable comparison to your beamshots, unfortunately. My 5700K triple is under medium spot frosted (Carclo 10508) TIRs, and with a non-coated glass the tint shift is ok enough that you probably won't notice unless you know it's there. With a purple AR, however, the beam turns to fried egg and the tint shift is very obvious on a white wall and in real world use.

It’s numerically small but is noticeable as other users have noted (stephenk above). I am glad that my result is in the range of Djozz’s test. His change between green and purple was about .003, mine is .002.

For those new to this you could very roughly see what the change in beam color is by enlarging the pic below and look at the color field that each of the dot is in. It does become a little more yellow-green with the purple AR lens.

I plan to take A vs B picture of actual scene (wall of truth in my living room :slight_smile: ) for further comparison between green and purple.

another data point:


The data jon_slider linked is very interesting, in contrast to cannga's data, as the purple AR appears to have less of an effect compared to green AR. Perhaps variance between batches is too much to allow a definite conclusion.

It’s confusing because of the way he presented data, but actually there’s no contradiction. (Jon thanks for the good find.)

I’ll list results that are common to all 3 tests:
1. Green more negative than purple: no contradiction here. Reddit poster’s result is –0.001, Djozz’s is –0.003, and mine is right in the middle –0.002. Coincidental or not they are actually remarkably close.
2. All 3 values are NOT that significant as far as effect on beam color (EDIT: wrong lol, as shown later, yes even 0.002 is a big difference) is concerned. Mainly because they are only third digit after decimal point and of low values (1,2,3).
3. In fact it seems so far ALL of the values listed for non-AR lens, no lens, green lens, purple lens, etc., are in the 0.001-0.003 range of each other (low value, third from decimal). Meaning this type of AR related tint mod is not going to change your beam color significantly.

IMvHO, outside of beamshot /white wall comparison, the best way to gauge Duv effect is to graph them on the BBL curve (NOT listing percentage like reddit’s poster). An example is pic below, where you could see a Duv change of 0.002 doesn’t alter beam color that much. I am actually very curious what the beamshot will show.

Thank you for the summary and sharing your perspective! I will add some of my thoughts...

I completely agree with Claim 1, and don't have enough info to comment much on Claims 2 and 3. I suspect that some really bad purple lenses I have incur a more than +0.003 shift, but don't have the equipment to tell. I might also add that the threshold for significance might depend on how tint-snobby someone is, and I personally find a change on the order of +/-0.003 very noticeable even without comparison, if the emitter is close to neutral.

I agree that there is no contradiction, but do maintain that there is inconsistency between batches. I am comparing the absolute-differences |Green - None| and |Purple - None|, and for the reddit data the green absolute change is greater, while from your tests the purple change is greater.

I do agree with you that plotting in color space is an excellent way to gauge duv, but I think the choice of 219B is suboptimal--I find the existing tint deviation of the 219B too large, on the order of 0.01. On that note, I would love to see plots of your 519A tests in color space!

I will respectfully argue that percentages are a very helpful way to gauge duv in practical terms. On a related note, I claim that the implicit question "is +0.003 duv significant?" is ill-posed without reference to the initial duv before the change. If an emitter starts off close to neutral, +0.003 would be a huge percentage of the existing deviation, and I feel many tint snobs will spot the difference. However, if an emitter starts off pure pink, +0.003 constitutes a very small percentage of existing deviation and would not result in a very noticeable difference: lots of pink plus a bit of green is still lots of pink.

I will go back to a previous analogy: is +10 lumens significant? Yes, if your light is currently off, which corresponds to an infinite-percentage increase. Arguably no, if your light is already at 1000lumens--I doubt many would notice the 1% increase.

Thank you for the thoughtful discussion. I will try to clarify and hope I don’t lose everyone. :person_facepalming: :slight_smile:

Sorry but your analogy of Duv with lumen doesn’t hold well, and I mention this only because it is related to the point I was trying to get across about reddit poster’s listing % change of Duv (no one does this I don’t think). Yes with lumen, percentage calculation is ok, because lumen is an amount. A LED with 1000 lumens has 100% more brightness than a LED with 500 lumens, etc… However, Duv is a number along a line on a curve graphed by x,y coordinates. There is no LED with negative lumen rating, but all LED’s have positive and negative Duv for this same reason.

I understand, but this is not quite true. A 0.003 change is the same change whether you are at 0.0001 Duv or 0.120 Duv. The pic below is my 219b 4500K with Duv –0.012, on which I’ve added additional Duv points. I hope it would make more sense than my writing.
One point is valid: I stand corrected and should have written: a 0.002 is a relatively small change, since “small” is so subjective, and small or not, a change in beam color has been noticed by various reviewers (see stephenk’s note above). Hope this helps to clarify & cheers.

This is an excellent point! My lumens analogy with duv is not an accurate one. Subjectively, however, I still find that the same duv difference is more pronounced when a source is close to neutral. I've played with the color space calculators from Waveform Lighting before, but thank you for introducing it anyways; it's an excellent resource.

After playing more with the calculators, I've created the following experiment for everyone to decide for themselves. It's not perfect, as the computer monitor does a poor job of recreating colors, but hopefully there is something to be observed. In the image below, all 4 sources are at 4500K CCT. Compare the top two squares, and then the bottom two. Try to decide: which pair has the greater difference in duv?

The duvs differences are about equal. (0.0098 for top, 0.01 for bottom)

@QR, I appreciate that you put a lot of thought (and “work”) into every post :+1: .

For anyone new to this and wondering what on earth we are talking about, if you like to try, Waveform Lighting calculator web site is HERE. It would get you “deeper” into the hobby and believe or not, A LOT OF FUN. Guaranteed lol.

The numbers come from this inexpensive device “Opple Light Master Version 3 (aka Pro)”. It’s the more expensive $40 item - do NOT get the cheaper Version 2.

Just purchased the Opple Lightmaster 3 Pro. Look what you just made me do!

@stephenk lol coming from you I’ll take it as a compliment :+1: :slight_smile: .

stephenk has over 2000 posts...

and I just noticed his username.

Are you a fan of Stephen King perchance?