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)

If the charging port is hidden under some kind of metal cover ( like a mobile metal part with a thread or something) i`m in 100%. If is with a rubber cover like all the rest i`m only 30% in. I`d even like the side switch without usb charging port option.

If I have to guess it would have rubber cover, similar to the side-switch Convoy M21E and M21F pictured below.

The old S21D has a beautiful smooth floody/”diffused” beam that, for me, is the best way to appreciate color rendition of Nichia (as opposed to a throwy beam with bright hotspot). Besides USB-C and side-switch, I hope Simon keep all other aspects (TIR optic, etc.) the same. Can’t wait for its release.

With a sphere there is less problem of calibration:
you just have to multiply the lux by the square area to know lumen output.
knowing the radius of the sphere is a breeze.
It is important to treat the sphere internally with the appropriate paint after having applied a suitable base.

I already have the primer and the paint, this weekend I should work on this sphere hoping that everything goes for the best. fingers crossed.

@Argo, that sounds like a lot of fun - best of luck! I just looked at some integrating sphere videos on youtube. It seems so complicated but that could just be my brain not absorbing science anymore.

Will you be using one of those “inexpensive” light meters from China? They have so much variation from one another. I have 4 including Opple and they read 3-4 different numbers :person_facepalming: :slight_smile: , probably the most inconsistent of light parameters I’ve measured. (This why Texas Ace calibrates the lumen tube with its light meter as one unit.)

yes you are right but I will use a philips e27 with declared number of lumens, cct and CRI, to have a reference on the reading of the luxmeter and light master.
yes Chinese instruments also have a large confidence interval on measurements, so I have to take this into account too, but I’m not a professionist, I don’t care about absolute accuracy, but making measurements all with the same error in order to be able to compare the various flashlights I own.
however the sphere is the best geometric shape to have repeatable and reproducible values ​​with the various flashlights.

(EDIT: Updated with more info and comparison with Convoy newer light S21F 4/23)

Thermal regulation of S21D has been a puzzle and topic of discussion for a long time. We know it’s there because output goes down within a few minutes, but… output then stays at a constant level of about 40% of max (and NOT lower) even temp remains high.

In nearly all other lights, including the newer Convoy light S21F for example, output would drop much lower than 40% and temp regulation clearly active as shown by a zig-zag runtime curve. As temp goes up, output goes down, as temp goes down output goes up, etc., resulting in the zig zag pattern. S21F for example would zig zag to keep temperature for example would stay in the 40’s centigrade. Not so S21D - you would see a straight line as illustrated below. (I hope I’m making sense here LOL.).

The bad thing is it could get really hot as owners have found out. The possibly “good” thing is that sustained output is higher and more stable than any other quad Nichia lights that I have.

Well, the design is very much overdriving the emitter and the heat dissipation capacity of the host is far exceeded…as the current drops the emitter is still supplying waste heat so the saturated host isn’t going to cool down quickly. Starting on the lower current mode never supplies as much heat to begin with, but the host does saturate eventually of course. Aluminum is good at dissipating heat but there’s a point where physics wins so we either shut the light off, provide cooling, or switch to more mass in the host (which only delays the inevitable when such heat is still being supplied at the source).

If your DMM will take a thermocoupler those are fairly inexpensive and can be taped to the host for better readings (they are notoriously inaccurate between the thermocoupler itself and then the meter inaccuracy, but in this case they will at least be consistent even if they’re off by a number of degrees). IR guns are tricky to use correctly and get their rated accuracy. Some black gaffer’s tape or berlin black will go a long way to repeatability as long as the distance is correct for the emitter in the gun.

An important note: As expected, in a larger host with more thermal mass/- less heat+ from LED, the temperature does not get nearly as high as in S21D, and the temp difference is much smaller. For example in my M21B with single Getian GT-FC40 LED (vs quad Nichia in S21D), at 30 min mark, T-35% is about 55-56 C° and T-100% is about 57°C.

For me the most interesting question arising from the graph: If there is temperature regulation, why did brightness not decrease further from ~37%? IMHO, there’s a trade-off involved with this algorithm: On the plus side Convoy lights have great sustained brightness that I love, on the minus side, small hosts with multiple LED’s like S21D could get very hot. (Not complaining, still buying mostly Convoy lights. :+1: )

As always I should note this is a home test done indoor, not some professional lab w/ actual internal T monitoring outdoor, so although I think it’s done carefully and results correlate with actual observations (S21D feels very hot during my walk, etc.), pls take with a table-size grain of salt :slight_smile: .

I don’t know if anyone has found a definitive answer for how this works in Simon’s recent drivers other than simple observation like this. It’s probably something as simple as “if T=x then I=x” without any sophisticated regulation. Simon would have to answer that, and/or he may need to ask his driver manufacturer about the details. But it’s just how this driver works. I don’t want to put words in his mouth but it seems like he’s maintained a perspective that max/turbo is just for brief use (as it was always mostly intended…because heat saturation) so straight run graphs like this don’t equate to real world use for most users. I’d mostly agree with that, and for those that do need to run higher outputs for longer times, well, you let physics dictate the host you buy for the arrangement of emitters and current delivery.

This is a lot like bringing a pot of soup to a rolling boil and then turning it down to simmer vs. bringing it to a point that is just below simmering. The burner is never turning off.

So if there is an alternative to Convoy’s temp regulation algorithm, what does it look like? One excellent example would be Anduril 2’s ATR in my beloved and very pretty Emisar D4V2. D4V2 is a smaller host so I think I will set Tmax to 50° C only, so that I don’t accidentally fry anything. (IIRC Tmax at either 55° or 60° makes D4V2 so hot it cannot be hand held.)

With my Anduril lights I generally have seen a very quick drop off from Turbo (not 10 minutes like Convoy S21D), and to a level much lower than Convoy’s 37%. I would expect the same characteristics for D4V2.

Another test already done but I am too lazy :slight_smile: to post yet: S21D with active cooling. Active cooling means having a fan a couple feet from the flashlight. I did this to confirm that S21D’s temp regulation is not a timed response . This question arose because S21D so consistently settles to the exact same 37% brightness, and on the dot between 9 and 10 minutes after Turbo turn-on. And IMHO the result shows thankfully it’s NOT a timed response.

For Christmas gifts, do you guys recommend 519a 4500k or 219b 4500k for people that aren’t into flashlights?

Will the new S21D make it in time before Christmas?

IMHO 219b 4500k would be a great gift. 519a 4500k has a warm/yellow tone that “normal” people may not appreciate.

China in yet another MAJOR lockdown so I am prepared if arrival of this light is delayed.

I should note that S21D is a floody light, more useful for close range such as work around the house.

For an all around gift light that could be used around the house, and for camping, walks etc., and probably more in line with what “normal” people expect of a flashlight, I would recommend a GT-FC40 4500k light like the M21F wholeheartedly. Not necessarily my favorite M21E because E is more throwy and spill not as bright; F is more “balanced” between throw and spill. IMHO and hope this helps.

Would it be better to go with a hank light?

Not necessarily better, but definitely great and pretty alternate choice.

Multiple factors involved in deciding which light is “better”: host size, throw, output. Can’t help you much though as this is your decision to make.

Personally I prefer gift light to have integrated USB-C charging, and not all Hank lights have this. But again YMMV.

Does the stock lens on Hank Lights have AR coating?

Is it better for the sw45k 219b to have AR or without?


Stock lenses on my Emisar’s do not have AR coating. That said the beam of D4V2 with 219b 4500k is equally as beautiful as S21D’s; the 2 look the same to my eyes. Duv actually measures nearly same.

Basically you can’t go wrong with either light. Some comparison numbers:
S21D ~ –0.0084
D4V2 ~ –0.0087
Basically both measure in the mid 80’s for Duv.

S21D 170 m with 10° Clear, 157 (10° Bead), 144 (30°), 113 (60°)
D4V2 137 m with Carlo 10622, 140 w/ 10621
If you want throw the S21D is better with the 10° clear TIR.

About 1700 lm for both using Texas Ace Lumen Tube.

Sustained Brightness in favor of Convoy (that famous 37% at 10 min) at the expense of heat. I plan to do a comparison graph later when I’m not in lazy mood.

S21D is larger so it has thermal mass advantage related to specific heat, but the smaller D4V2 feels GREAT in hands. The higher thermal mass (without battery S21D 4 oz, D4V2 2.7 oz) is important for sustained brightness and this and the longer throw of S21D with 10° TIR are the 2 main reasons I personally prefer S21D, but YMMV.

Drivers, Cosmetic, Cost, USB charging
Convoy has 8A Buck, D4V2/D4K has Boost option but that costs $12 extra.
Emisar is prettier to my eyes, with multiple choices for finish/color.
Emisar D4K with boost driver I think is about about twice the cost of the $30 Convoy S21D.
New Convoy will have integrated USB-C.

All IMHO/YMMV. You could do what I did and ordered both Emisar and Convoy. :slight_smile: :+1:

For the average person the coldest whitest light is best for the greatest wow effect.

Agreed, unfortunately. I only got a jump start as a result of rather excessive time spent with my other hobbies, photography and video. Still it took some time for me to learn about Duv and green monsters. Never forgot my puzzle with Cree 70.2 in a Thrunite light - wait is it normal for it to be so green :slight_smile: .


Thanks for all the helpful tips guys.

Do you guys know if the S21D lens will fit on a hank light?

I got the stock carclo 10622 and was thinking maybe I should’ve got the 10623 or 10624