Iām looking forward to seeing beamshot comparisons with the IF25A optic vs. the stock S21D optic.
I measured 2055lm for the 519A in the S21D on 100% (stepped down to 790lm between 105secs and 10mins), and 860lm on 35%, which is pretty good for the lights size and 95CRI. Convoy tend to quote theoretical lumens.
Good suggestion thanks. Not metal but at least lighted.
I sure will. Prelim result is excellent; as expected dimmer spill, but brighter central area. The stock S21D is nicely floody, but a bit too much for my use. The clear optic makes it a little more balanced with some spill + some throw. The Convoy now behaves more like Emisar D4V2.
The frosted optic makes for a nicer smoother white wall shot (probably one reason why itās used, for the sake of perfect usersā reviews 
), but in actual use the clear optic is ābetterā for me because it increases max brightness (albeit in central area only). Any increase in max brightness is always a plus for Nichia lights for me.
At 69 cents IMHO this is a no brainer rec. for owners to try and decide for yourself.
@cannga - I tested my S21D (w/ 219b emitters) with the IF25A optic. For me, this is a major improvement. It gives a truly defined hotspot and makes it much better for general purpose use. I appreciate you doing the research on this and sharing your findings, as Iām very pleased with the results.
Strangely enough, I think I prefer my IF25A with the optic that came with the S21D, so itās a win/win. Are you aware of any other optics that fit these two lights?
On a different note, have you noticed any strangeness on your S21D in terms of brightness? Iāve tested two groups (#1 and #5), and the 100% on both is NOT max brightness. If I cycle through the levels, every now and then it will jump to what I would refer to as āturboā. The easy way to tell, aside from the increased brightness, is that at 100% the light head does not heat up to uncomfortable temperatures even after 10 minutes of running. At the āhiddenā turbo setting it gets rather hot after <1 minute. The light does not remember this setting after turning it off - it reverts to whichever setting it was on before you slipped into āturboā and you have to fiddle with it to get it back there.
I would describe the perceived brightness difference between ā100%ā and āturboā as roughly 2x. Itās very noticeable. It seems to be easiest to get into this mode by holding the button in just before the amount of pressure normally required to cycle modes, then releasing once it switches to āturboā. I have run it at this setting to the point of it getting as hot as my single LED Convoys get (not too hot to touch, but getting uncomfortable) and it has not released the magic smoke.
@undefined - I canāt say that Iāve seen any abnormality with the brightness. Not sure what is going on with yours but just a couple suggestions:
1. Turn memory off to remove that from the equation. Does the flashlight still have problem?
2. I use group 6 with memory off. Maybe gives this a try? It should turn on at max 100% every time and there should not be any brighter level. I will try your maneuver later to see if I could reproduce the ā100 PLUSā brightness. Hope this helps.
Btw credit for the IF25a clear optic goes to poster āVerodin,ā and yeah I am not going back to frosted optic.
I tested the hidden turbo out with my other S21D. I put it in group 12 (100% only). Turn it on, then hold button and it jumps to much higher brightness and stays there. Turn it off and back on and it returns to ānormalā 100%.
Two different S21D, same odd behavior. One is 519a, the other is 219b.
Easy way to reproduce my strange āturboā.
1. Set light to group 12 (the one that has only one power level, 100%).
2. Turn light off and then back on. Verify that half press does not engage other illumination levels.
3. Hold button in, not as far as a normal half press. This may take a few tries, but youāll know when youāve got it as the light will get noticeably brighter. When you release the button, the higher brightness level will remain.
4. When you turn it off and back on again, it will return to normal brightness level.
The 519a does this with almost 100% reliability. The 219b I have in mode 6, it takes a lot of tries in that mode to make this happen. I may try to change it to mode 12 later and see if it behaves the same as the 519a, I just hate changing mode groups on the Convoy driverā¦ 20 clicks is too many, Simon!
If youād like, I can try to record video of this.
Also, finally have both versions (219 and 519) of the S21Dā¦. definitely like the 219b better.
Please note my two S21Dās have the metal switch, not the stock rubber switch. I tried the above maneuver with 519a in Group 12 setting and cannot reproduce the ā100% Plusā mode. Tried the 219b in Group 6 setting and cannot reproduce it either.
Thanks Stephen. I added your data to first page of thread.
Comparison of stock frosted optic vs clear optic from Sofirn IF25a. Shot in RAW with Canon DSLR, WB 5000k. Exposure settings are same for both shots to show brightness difference.
Less spill but a more defined and much brighter hotspot. Previously it was a very floody light, now a bit more balanced between spill and hotspot.
For my use, this is more useful so Iām keeping it.
A:B comparison of frosted vs clear optic
Same exposure and WB settings.
Very useful comparisons, but there is something to add, namely descriptions as to how you experience these various LEDs, in various situations. I mean the actual experience vs. pictures is not the same because we watch pictures on a web page (with more stuff on that page around that picture), in a room which has still more stuff that we can all see at the same time as that beamshot picture, such that actually what you show is not what you will experience when using it because our own white balance depends on the whole picture we see, so unless you watch in a dark room and only see the picture on the computer screen, what you see in reality is not what the picture shows. The more is on screen in the picture (the room) the better it is but it still wonāt be the same so a description on whether āI get a pink feelingā experience occurs for example would be useful.
This is also a problem with beam patterns for cutoff lighting, actually even more so, which I analysed and explained on my website. I havenāt done this in in detail about the white balance, but I did mention on my site the issue with beam shots not showing what you experience due to perspective, the aim of your view (what you look at, and at what distance).
So various things that are obvious in your pictures may not be obvious in reality except in special circumstances. In your pictures the 219b shots I all see as pinkish, very unnatural, but perhaps in reality it works well in many situations, but not all situations! I ordered a D4V2 with 219b 4500K so I can try out different firmwares and try out these LEDs, will likely only arrive in a few weeks at best. I read comments that some people dislike the 219b in some situations, especially when there is snow, and that makes sense, it would give a pink glow then to what you see.
This is likely also my problem with the LH351D, in that the situation with lots of wood colours, things are off. I will do a beamshot session with the 219b 4500k, 219c 4000K, XML 4C, SST20 4000K and LH351D when the D4V2 arrives to see about the situations that I feel are problematic. I estimate that the SST20 4000K and 519A also ca. 4000K will be the best in giving good colours under all circumstances. I will try to get hold of some 519As in 4000K and 4500Kā¦
Oh, something else, lit tail caps. According to Simonās aliexpress pages of his lit tail caps, these take about 9-14 mA with a full battery, dropping off when the voltage drops. This would mean these suck an 18650 empty in a bit more than 10 days. That seems far too much (small LEDs donāt need so much power, e.g. you can run a 3mm standard LED that is rated to 20 mA at 1 mA and it will still be quite visible). In the thread āD.I.Y. Illuminated tailcapā a current is mentioned of 0.29mA, which is a lot better, that would take more than a year to drain an 18650 ca. 3400mAh battery.
Have you noticed such a high drain with those lit caps from Simon?
@swhs, thank you for the thoughtful analysis. You brought up some very good pointsā¦
1. Metal switch: The switch is nicely bright so I suspect it is draining on the high side. I have used the light continuously so canāt tell. When battery voltage is 4.2V, the current for blue light is 9.4 mA, when voltage is 3V then current drain decreases to 2.74 ma. How to calculate how long this will last in a 5000 mAh battery? Any expert please chime in if Iām wrong: Is it 5000mAh/9.4mA = about 500 hours minimum? Current drain goes down with lower voltage so I believe actual drain time will be longer than 500 hours.
2. This is obviously not the place for a lengthy photography discussion, but yes, there is NO replacement for ālive viewā and photography must be used in conjunction, not replacing. Where pictures are useful is when the difference is small, and when you need to confirm your suspicion in ālive view.ā One example is the Nichia E21a: Iāve read all sorts of posts, some hate it, some think itās the second coming. Photography confirms for me that it looks very similar to 219b. Only photography could āfreeze frameā (vs fleeting visual memory) what we see and allow us to share the scene, often enough āmore reliableā than our words of description.
3. Magenta or not? Multiple discussions already and subjectively how nice it is, depends on personal preference, no right or wrong. I would like to emphasize that objectively, the 219b is unmatched in its CRI numbers. IOW itās still the king, especially for the color red (R9). Used daily in nature or in my cluttered living room, it distinguishes subjectsā colors better. For this reason IMHO a 219b 4500k flashlight is a good first start for any light collection. And for various reasons the Convoy S21D with that LED is the very top choice for me.
4. Lastly IMHO, if I want a happy flashlight life, I would stop doing comparison 
. Really, whether live view or photographically, comparison āmagnifiesā differences (partly why itās so important for hobbyists). I never thought my SST20 4000k or LH351D 4000k were green until I compared them with 219b.
Yes, 500 hours.
Thatās mainly why I donāt like the lighted switches: they are quite innefficient.
Ideally, theyād be sub 1mA so that the drain would not be noticed much.
Where are the objective data to show that the SW45K is unmatched in its CRI numbers? The SW45K has been famous for being a subjectively pretty emitter, but it has healthy company for R9080 rating.
Here are some actual objective data courtesy of Maukka:
E21A:
219B SW45K:
Optisolis:
SST20:
You are correct - thanks. E21a is outstanding and I actually mentioned it in my post above. 
I have both Emisar 219b 4500k and Emisar E21a 4500k. I compared them with pictures and use them often Convoy S21D w/ legendary Nichia 219b :) - Review & comparison w/ Nichia 519a, E21a & other LEDs & lights (D4V2, Convoy S21F). - #16 by cannga . Both with my eyes and with photography, they look very similar (E21a a touch more yellow). So at least with the LEDās I have, subjectively IMHO E21a is nearly a 219b twin, meaning top-class CRI, but my opinion is to pass if you already have 219b, and partly why it didnāt get mentioned.
Thanks for the confirmation. :+1:
1 mA might be too dim to see? I have a Wurkkos TS21 that has the same problem with light button drain, but since it has Anduril I could change the light it to lower output setting. Donāt know what the drain is but then itās so dim I canāt see it.
That depends actually.
Sub 1mA is very easy to see in the dark, especially if youāre willing to get a green/blue LED.
One other reason I recommend 219b over E21a @4500k is, at least with my Emisar, the E21a requires frosted optic 10623 which makes it looks dimmer, and too dim for me.
When I replaced that frosted optic with a clear one, brightness is better but then I discovered the beam shot has yellow greenish ring/corona. 219b 4500k with same clear optic still shows beautiful & smooth magenta beam.
