I kind of thought that, but it’s hard to go through 600+ messages. Wish that BLF had a thread search.
could use 1 cr123a and a spacer in a pinch
I think the Title of this very thread ought to be a clue 
World’s Smallest 18650 Flashlight
Many (larger) 1x 18650 lights have drivers with voltage range of 2.8-6V which permit use of 2x CR123A…. PDs, 1st responders, and military usually use CR123 primaries instead of 18650 li-ions.
.
1x 18650 or 2x 16340 (2.8-8.4V) capable drivers are not as common… Which makes a lot of sense to me. If you are going li-ion instead of primary, best to use the larger 18650s with superior run time. Only advantage of 2x 16340 is that it may be brighter (and hotter), but that depends on what kind of regulation is used. And run time will suck.
.
That said…. given the heat issues this tiny light has, anything over 4.2v would not be wise IMO. I see no reason to fit a driver that exceeds this.
I did some ceiling bounce lumen estimates on my (NW) DQG18650. There were measurents before from _the_ and relic38, and those were undoubtly better performed than mine, but here are the numbers anyway, actually consistent with their numbers and somewhere in between:
low: 12.5 lumens
med: 157 lumens
high: initially 470 lumen, within a second already 420 lumen, settling after two minutes at 335 lumens and from there a very slow decline.
What I believe is happening on high setting is the following: the very rapid decrease within the second is actually pointing at a unusual high thermal resistance very close the led, directly where it sits on the board, then the still fast decline points at a lot of thermal resistance anywhere through the board to the body. The very slow decline leftover after two minutes is the limited surface area from body to air, so that the body is still warming up slowly.
Design-wise just the limited surface area of the body is not avoidable (it is already made of aluminium and we want this light small) and it is not a big problem anyway. The other thermal issues can be avoided largely by a different design (such as using a small pill anyway with separate led board and driver, it will cost a few mm extra). It made me wonder wether the board the driver is made on and where the led sits on is a metal core one? I noticed from the DQGAAA that that is not the case, and my guess is that it is the same here so that the heat path directly near the led is only through the electrical traces and not through the board at all, which may also be the reason to choose for fujiking the board behind the alu divider in the head.
Any comment on my wandering thoughts?
Actually i can answer the question myself, and with my own picture of the driver ( #574 ) : the driver board contains a large amount of via's and I guess that would be impossible with a metal core one
commenting on my own posts, I must like myself 
Going from 470 to 335 lumens means a decline to 71% of initial output. If you look at Cree's xml data sheet about junction temp/luminous flux that would mean a junction temperature of about 155 deg. C ! Awfully close to the melting point of solder, bad for the led anyway. I hope my measurements were lousy and this is not true 
.
I saw that post Bort!
I don't think it is voltage sag, I used a Efest 18650 IMR battery fully loaded, at 2.0A in the first minutes (not seconds) the voltage drop is 0.3V the most, the voltage stays very near to 4V. It will affect output but nothing near to 30%. And I could repeat the same drastic output decline after cooling down for half an hour without recharging the battery.
I think you’re on to something djozzy. There’s not much to the head.
How could this be modified/improved? Is there space to do anything?
i reread your post and revised my thoughts but you were too quick
:bigsmile:
:bigsmile:
Actually, it has nothing to do with the small size of the flashlight, what happens very near to the led is what causes the lumendrop. I suppose there is not much you can do afterwards, but the design can certainly be improved. The fact is however that a 30% increase of output is nothing impressive enough for the majority of people (apart from us freaks), and the fact that the led does not even get close to its 100000 hours lifetime is also not a concern for most users (same discussion was happening on the bad heatsinking of the Nitecore EA8).
Well I’m loving this light- wish I had bought 2. Nice floody beam, great build, tiny for a 18650 light. everything that was promised….
I just finished my review: Review: DQG Tiny18650 (XM-L2, 18650)
Hi,
I just saw it. Very thorough job! Thanks,
Jim
Yeah thats what I meant… is there room in this existing design to add some kind of heat sinking improvements?
I haven’t look inside the ones I have, but could you do something like pot it with Fujik so that the heat from the emitter+driver PCB could transfer to the outer body better?
Thank you very much!
I am imagining this torch as a SC52 with less modes, poorer build quality, slightly larger with longer runtimes.
Accurate representation?
What?!
No.
I almost spit water on my monitor LOL!
Comparing the DQG Tiny18650 to a ZL SC52 could not be much more apples vs oranges. The DQG is a budget light in every sense of the word. It is simple and gets the job done. ZL SC52 is a top end light. Fit & finish, build quality, modes, thermal management, performance and price are on a whole other level.
Comparing the SC52 and SC600 would make more sense. Obviously, the sc600 costs considerably more than the DQG. For good reason. But the SC600 is the 18650 version of the SC52 if that is what you want.
If you want to compare the DQG tiny18650 to a AA powered light the best parallel would be the DQG AA. Same FnF, same build, 1 less mode on the AA, similar thermal management [weaknesses], lower performance, and lower price - in keeping with it being a AA powered light.
For what it is, the DQG tiny18650 is a cool little light. But I don’t think zebralight is losing any sleep over it stealing their customers… Just as I doubt Porsche worries about VW customers. 