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)

Probably the most favorite light in my small collection is the Sofirn SP36 BLF with Samsung LH351D 4000k, a very well known and one of the most highly regarded LED’s outside of Nichia. (LH351D is the same LED that is used in the famous lantern Sofirn LT1 BLF.) Known for its high CRI, the Samsung to me stands the perfect middle ground between magenta tint of Nichia and yellow tint of SST20 4000k.

Before comparing Nichia vs Samsung, I thought it would be fun/interesting to first compare Samsung LH351D 4000k versus Luminus SST20 4000k, both in Sofirn lights. Both LED’s have CRI/Ra in the 90’s, but I don’t know about R9 (strong red). Judging by how the lavender plant is rendered, I would guess neither is as strong as Nichia’s R9 value of 80 plus in the 219b series (219b is R9080, the highest Ra and R9 rating possible).

I hope the photo shows why I love my Sofirn SP36 w/ Samsung LED. Without doing photography comparison or CRI consideration, if you were to ask me based on real life viewing which is my most favorite tint overall, it would be the LH351D 4000k. In this particular crop, LH351D is less yellowish/greenish than SST20 4000k, and has more faithful reproduction of the bark.

The LED’s tint has little to do with its R9 rating. 95cri 4000k SST-20s have R9 of 97 (with no additional optics or lens over the LED).

Thank you.

SST20 4000k Luminus SST-20-W 4000 K CRI95 color and output test
Looks like CCT 4312, Ra 96, R9 97

Samsung LH 351D

CCT 3735, Ra 98, R9 97

I have not seen an R9 of 90+ on an LH351d before

it is nominally a 9050 LED, whereas the SST-20, 219b, and E21a are 9080 variants

the red bar is shorter on LH351d, than on 219b, sst-20, and E21a:


look also at the DUV for each LED… the SST in this case, is greener tinted (higher DUV, farther above the BBL) than the LH351d (varies by CCT also)

additional food for thought
compare R9, R10, R11 bars (red, yellow, green)

notice the red bar is not shorter than the yellow and green on the 219b, iow it has the most consistent red, yellow, green color rendering capability

the yellow and green are taller than the red on the LH351d, iow it renders yellow and green, better than red

the green bar is taller than the red in the SST-20, iow, it renders green better than red and yellow

photos of plants tend to lack yellow and red targets, they tend to be rich in greens. trivia… plants look really green under low cri light, it does not reveal red pigments at all

Maukka did do a couple charts of the LH351D when he was playing with Lee minus green filters. They turned the LH351D into a CRI monster with like 98 cri and 80+ R9.

here is one of maukkas threads about Lee Filters,

and how they affect DUV…

they show a small increase in R9, not significant, and comes at the expense of a large reduction in output, imo

Lee filters Light transmission
No filter: 100 %
1/8: 88.8 %
1/4: 83.8 %
1/2: 73.7 %

Im not a fan of Lee filters… (they not only rob output, they also create excessive pink tint across the whole beam, when it might only be needed on part of the beam). The filter color is also not stable, it fades, and the coating is fragile, it wipes off easily with alcohol… I would rather use a better LED in the first place

Great discussion guys.

Jon, a couple of questions for you:

  1. The Lee filter - is that the same as the typical green filter that people would use? Its problem with light loss, fading, etc. Is that typical of these filters? I have no interest, just curious.
  2. Good catch, R9 for LH351D LED is not 97 (my mistake), but still in the 90’s, 92. The numbers I quoted is from “u/maukka.” Did I read that right? I don’t care so much about the exact number but am curious: how much is the typical variation in CRI among different bins of a LED? And is that what we are seeing here (different R9 numbers)?

EDIT: Oh I see now Jon. Your low R9 is for the 5000k Samsung LED. Mine is the 4000k version.

I actually have that 5000k LED in my Sofirn D25L headlamp (Can’t control myself and have bought way too many lights. :person_facepalming: :slight_smile: ) It is horribly green and that’s why I haven’t talked about it.

BTW that Sofirn D25L headlamp on Amazon and their mention of the Samsung LED is one reason I got started in this hobby - “hmm the LED’s are different”.

> The Lee filter – is that the same as the typical green filter that people would use?

yes, a minus green filter is a pink photography gel, they came in free swatchbook samples, here is a pic of some I tested… I no longer use filters…

there is a higher grade, more temperature resistant, that maukka used in the post you linked

Zircon 804: light transmission 83.6%

good catch that the R9 difference are for different CCT …
it still seems unusually high R9 for LH351d… I do not attribute that to the Lee Filter, but rather a “good batch” of LEDs… at that time…

The only Lee filters I would use are the Zircon filters. They’re used professionally and don’t fade really easily. The 804 I have in my ROT66 gen 2 has seen 100s of hours of usage on lower to medium outputs, and is still as pink as the fresh 804 I still have rolled up in my storage. Very easy to sandwich a Zircon filter between the lens and optics on TIR lights. I’ve never used them for reflector lights.

Up until this point my favorite high CRI combination has been SST20 with 804 over them. The combination gave more output, more throw, and just as good if not better color rendering and same rosy tint as 219Bs. But now since dedomed 519As are a thing, it’s probably finally time to switch over.

glad to know the Zircon is much more durable than the swatch book gels

dedomed 519a is fun, not as low a DUV as 219b sw45k, but lower DUV than SST-20 and LH351d… a dedomed 519a, might resemble an SST-20 with a Zircon 804 :wink:

We are fortunate to have great 9080 LED choices, in a range of Tints now.

sw45k, E21a, SST-20, 519a, all offer 9080 CRI LEDs, with varying Tints

I still like Tint below the BBL, but Im seeing that an LED does not need to be ultra pink to have great color rendering, even if it looks slightly greener than some sw45k… :wink:


519a 4500k de-domed (24% CCT deduction, more negative BBL) - may look exactly like 219b 3500k?

BTW, I am finding out more & more: photography is excellent at telling how two LED’s are different (for example 219b and 519a), or how they are the same (219b and E21a). BUT… to have the complete “picture,” you have to combine photography with real-live view to be 100% sure. For example, the Emisar has a strong hotspot (vs Convoy’s floody), so for the Emisar picture, only the center part reflects what I see in real life - a beautifully warm tint. The under-exposed periphery appears too saturated.

When is photography superior, IMHO? Example - 219b & 519a: crop of the red hat and back ground leaves no doubt about 519a’s positive Duv tint. 219b versus E21a: I hear comments all over the place re. what E21a looks like. A photography crop shows even w/ staring and repeated staring (shutterbugs call this pixel peeping), I cannot tell them apart. ie For all practical purposes, identical.

> 519a 4500k de-domed (24% CCT deduction, more negative BBL) – may look exactly like 219b 3500k?

the Tint DUV is lower on the dedomed 519a, I like it better than 219b 3500k. Note also the variation between some of my 219b 3500k




I looked up the ROT66 - VERY nice, and powerful. I love my beer-can size SP36 BLF (among my most favorite) and would enjoy having the ROT66 if I hadn’t had so many lights already. Is company still active? Wonder why they stop making it?

Jon you are a great source of information. Mention Nichia and I could see your eyes light up, rubbing hands together . :+1: :slight_smile:

Fireflies is still active and still selling lights, but they’ve been hit hard by the chip shortage, so they can’t produce any more E12R and Nov-MU for example, because they can’t procure the chips needed for the buck drivers in them. ROT66 gen 2 doesn’t use a buck driver, but still might be affected in some way or another.

Lol, yessss…


Does anyone know if these is a pocket clip that works with the S21D?


ask Convoy if one of these works

The R9 for the LH351D is stated to be 75 in the post by AEDe for a group buy. The measurements by Maukka of the LH351D 4000K are done with a filter which I presume lifts the red and yellows by reducing other frequencies, as he states: “So here’s how amazing the Samsung LH351D triple looks with a Lee Zircon 804 minus green filter”.

If the LH351D was that good without a filter everyone would be using it, and there would be no need for the SST20 4000K, Nichia 219B/219C/519A at 4000K etc. I have the FC11 with 4000K LH351D and it’s definitely not performing as in that review.

I did a test for example comparing the rendition of yellow (for example using a big yellow Stanley fatmax spirit level) between:

  1. LEDs of ca. 6000K (not sure exact type, used in an Exposure Strada 1200 bike lamp),
  2. ca. 4000K neutral white (XP-G 5B1),
  3. 4000K neutral white 90 CRI LH-351D (FC11),
  4. 4000K neutral white SST 20 95 CRI (IF25A).

The result is that between each of them there is a clearly visible difference, with an increase of vibrance of the yellow. It looks most washed out with 6000K and most vibrant with the SST20 4000K. This means the yellow bar should be lower with the LH351D 4000K than with the SST20 4000K, and it is (when not using a filter)…

In your picture comparing the LH351D 4000K with the SST20 4000K the tree looks a bit washed out with the LH351D, and in other areas too colours are washed out a bit. So I see issues with rendering of esp. wood/browns there. In my house I have a lot of plywood inside, varnished and natural, and downstairs mahogany window sills with inlaid hardwood plywood, floors everywhere dark wood grain laminate, etc. and since finishing all that, I always get this feeling that everything feels off with the FC11 (LH351D 4000K) whereas everything feels pleasant and natural with the IF25A (SST20 4000K).

I’m not there now but I will try to make some pictures when I get back there.