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)

You’re very welcome. This “green tint” so puzzled me once (because if I just look at the lens, there really is no green “tint”) that I made a youtube video. :partying_face: Much easier to use the ceiling light bulb reflection test. Note that a flat angle you’ll see purple; only when you look straight down on the lens that the green reflection is seen.

I just looked at my result and can’t help but have a chuckle. I appreciate your sharp mind. It’s as if you had looked at my cheat sheet; but actually it makes me feel very good and somewhat assured that my humble home test results are no BS. :partying_face:

1. UCL AR Purple: will set this at 100%
2. UCL AR Blue: 99%
3. Convoy AR Green: *98.5* (3.5 just to exactly match your “ish” lol)
4. Wurkkos Non AR 95%

When it comes to light transmission, for some odd reason the table is turned, oddly in reverse order of Duv listing. The UCL purple was horrible with Duv change but now is the star of the show (btw previously I measured about 98.4% transmission vs no lens using ceiling bounce +*HERE). I did 6 runs with the above test; hotspot on sensor this time (vs ceiling bounce), and the results surprised me with how consistent they were, considering the Wurkkos TS21 is a lousy light to test because of its numerous stepdowns.

Thank you--I'm a bit surprised myself with the guess. 95% is really good for uncoated; I've seen estimates like 92-93% thrown around on the internet. Probably depends on glass quality; some have more or less iron impurities (which manifest in green tint if you look at the lateral side of the lens). I would guess 4-5% for purple based on some random tests I've seen but can't recall, and figured that green would be a bit lower on the lumen count, reasoning below...

For the same amount of radiometric output, green contributes a lot more to the lumen count than purple, which is formed by blue and red at the tails of the luminosity function. The tails of this function exhibit very fast exponential-like decay, so the sensitivity difference between green and purple would be huge, on different orders of magnitude. Since green AR throws away green--the color contributing most to the lumen count--from the spectrum, you would expect more lumen loss.

It is a sad reality that given the same radiometric output, making tint more magenta necessarily induces a loss in lumens. (This is one reason why some emitters with higher flux bins tend to be greener, while lower flux bins have a better chance of being magenta.) Thus, when tint is concerned, a few % output loss should not really be considered as an inefficiency at all.

Beamshots of Convoy AR Green versus UCL AR Purple in the 2 lights, one with 219b 4500K and one with 219c 5000K.
Showing what a 0.0020 Duv gain does to the color of the beam.

Wurkkos OEM (“no lens”) versus Wurkkos with Convoy AR Green. A small 0.0003 Duv difference. The color difference is there, but very slight. I would not want to take a blind test to determine which is which :slight_smile: .

I am reposting a similar Convoy AR Green vs No Lens comparison, except with a LED that’s much further from the BBL line, the 219b 4500K in S21D. Again, the color difference is there. Slight maybe, but there.

For me, this comparison versus “No Lens” shows Convoy AR Green is fantastic but is not going to perform magic and turns a Cree into Nichia for example. Mainly, it doesn’t do any harm to Duv like the UCL AR Purple.

A couple months ago we had the above exchange - the question was about behavior of 8A Buck versus 12A FET, the 2 versions of Convoy S21D. Thanks to @bilakos10’s review of the FET 12A HERE, we now have an approximate answer, and it looks like stephenk hit it on the nose. :+1: :slight_smile:

The 12A FET driver indeed has a brighter max start ~2800 lm, but then stepdowns quickly and rather drastically, to 350 lm, 12% of max after just 2 minutes.

The yellow letter X superimposed on bilakos10’s 12A FET graph represent numbers from the 8A Buck version of the light. They are approximate ceiling bounce numbers, where I assumed 100% is 2000 lumens. Even though the 8A Buck is not as bright at start, it has a much more gradual and gentle stepdown. In addition, it sustains brightness much better, settling in at ~750 lm.

Looking at the above runtime graph reminds me of this puzzle, a question I always have about my Convoy: if the thermal limit is set at 55°, why do I measure 60° at the head during my runtime tests?. In fact during my runtime comparisons, the Convoy always measures the hottest among the lights I have. Most lights stays at 40-50° at the head, versus 60°.

Here is my attempt at explaining this discrepancy. Essentially it is true what Simon says, there is a thermal limit of 55° and it does trigger stepdown. However the stepdown is limited to a power level that is preset at factory. This level is fixed, regardless of what the temperature is. This would explain why my Anduril lights are all over the place with brightness output, and my Convoy is a flatline with hardly any up or down regulation.

Not complaining at all. The flat output is one reason why I love my Convoy’s. Anyway, I am not sure if my “theory” is correct at all. Please take the “explanation” with a table-size grain of salt and correct me as needed.

Flashlight porn to spice up this somewhat dry topic. I have dual citizenship lol so you’ll have to put up with my love for Emisar D4V2. They are like little jewels that fit perfect, PERFECT, in the hands. Enjoy your Convoy but do get an Emisar one of these days to know what the hype is about :innocent: .

My question would be which light will be able to sustain a higher output on identical leds, the S12D with the 8A buck driver or a D4K/D4 with the 8A boost driver.

Might be close but with similar calibration I think the S21D would have the edge due to the more basic temp regulation algorithm.

Interesting discussion. I made some comparison of S21D versus D4V2 previously, but only with the NON-boost driver version of my D4V2.

I am not sure if the sustained brightness is related to the driver, or more to the stepdown algorithm that Convoy uses (basically a fixed lower limit that a Convoy light will not go lower, regardles of temp)? In other words, if D4V2 with boost driver has the same thermal regulation algorithm as the non-Boost, wouldn’t it step down just as much?

Looking at S21D FET vs S21D Buck above, I wonder the same thing. Is the drastic runtime difference due to difference in driver, or difference in thermal regulation algorithm? I didn’t think flashlight could be so complicated and interesting. :slight_smile:

Which exact driver (boost 8A?) is in the D4V2?

Right, back to the green vs purple AR glass. This is my testing with Convoy S21Es on 40% mode, hotspot measurement. Opple Lightmaster 3 Pro.
2700k green AR 2681k 97.9Ra +0.0010 DUV
5700k green AR 5147k 98.1Ra –0.0013 DUV
5700k purple AR 5157k 98.3Ra +0.0018 DUV

The lens switch is between 2 & 3 right? If so, that’s about a 30 point raise in the Duv? We are getting results that are very close. Mine has been about a 20 point raise from green to purple (not purple to green lol - I’m confused, and can be confusing).

Sorry for the confusion. My D4V2’s are all regular, NON-boost version. AFAIK this means CC 9A for the 219b D4V2, and CC 5A for the E21a D4V2.

Yes, the lens switch (both Convoy lenses) is 2 to 3. I took a few measurements and would say there is a +/- 0.0006 margin of error of hotspot aiming. There is certainly a significant difference between the Purple and Green, though both are perfectly neutral when not compared side by side. The purple had marginally higher lux, but the difference was not significant.

Thank you. I did another round of testing with new AR purple and Non-AR lenses that fit S21D (I bought just for testing). Results again follow the same trend:
1. From our findings, AR Green drops Duv about 20-30 points from AR Purple.
2. It may also decrease Duv from non-AR lens and from no-lens, but to a much lesser degree. I call this “active-rosying” lol.
3. AR Green is not as good at light transmission vs Purple, but we are talking about 3-4 percent - not relevant from Nichia hobbyists’ perspective, and for example absolutely nothing compared to stepdown loss. I do use AR Purple when I am trying to hit 1000m throw for my TS30S for example, but that’s about the only time I care.

I have not had so much fun with flashlight since I got the Opple. And I am glad you got one too to confirm and thanks for doing the test.

Are we the first people who quantify Duv change of AR Green vs AR purple? :+1: :slight_smile: That would be cool.

We aren’t the first, as Simon was already talked into green AR as default for S2+ 519A. Not sure why green AR wasn’t default for S21E 519A.

During the lens comparison I noticed Duv would drift downward the longer the LED is on. As if it’s part of warming up. Because I have nothing better to do I ran another test to see how much and how fast this drift is.

These are numbers from S21D with Nichia 219b 4500k. Brightness is 2 down from max, 10% (?). The light is on tripod as usual and in this setup I could see x,y coordinates drifting slowly 1 or 2 points at a time (4th significant figure).

Start: Duv –0.0084 CCT 4455
5 min: Duv –0.0088 CCT 4466
10 min: Duv –0.0091 CCT 4470
15 min: Duv –0.0093 CCT 4471
20 min: Duv –0.0094 CCT 4473
25 min: Duv –0.0095 CCT 4475
30 min: Duv –0.0096 CCT 4478

So Duv dropped about 12 points while the LED is heating up, and was still drifting slowly when I got bore and terminated test. I’ve seen the same thing with my 519a LED. (EDIT @ 25 min)

Is there a reason for the sudden discontinuity at 25 min?

Thanks for your eagle eye attention. The reason was probably someone interrupting me during my lunch break. :slight_smile:

25 minute: x 0.3581, y 0.3430
Duv –0.0095 CCT 4475

Makes sense, thank you for the clarification!

The drop in duv and rise in CCT seems to suggest that more blue is leaking through the phosphor--I guess the phosphor has slightly worse performance when hot.

Another related (?) phenomenon I observed yesterday: I was messing with a Convoy S2+ 519A triple running on a 4-mode 5A driver and protected 5A battery. I've had the light for a month and used the turbo lightly, no issues. Yesterday I ran it a bit too hot, and it tripped the cell's protection circuit! I guess that heat somehow also causes higher current draw (through lowering forward voltage), which might also cause more blue to leak out as the phosphor comes closer to being saturated.