Personally I would say the opposite. I use momentary turbo (temp unregulated) a lot, with the D1 my hand is closer to the head so I can feel if it’s overheating and release the switch to turn it off. With the KR1 I would be worried about cooking the emitter or driver if I’m not careful.
2 Thanks
Thanks for the detailed, thoughtful reply @nicodimus22!
However, what I will say is that I would not recommend getting the SBT90.2 in a small host that doesn’t have much thermal mass or a substantial battery backing it up. It’s a disappointing experience.
I understand what you mean. But the usage I had in mind was not one of using turbo all nor most of the time: the idea is to use it for short bursts up to a few dozens of seconds max each. Sorry for not making that clear in my OP, I will amend it to that effect next.
The L7 sounds like a very nice light, but my issue with it is having no Anduril (sorry for not mentioning it explicitly, but it’s a good part of the reason I wanted a Hanklight in the first place). Perhaps something like what @YBF650 has commented could be arranged (ie, replacing its entire driver by one from Hank, or at least its MCU by piggybacking a daughterboard with an AT1616 on it) could be done, depending on whether the L7’s head has enough space inside for such a feat. I had a look and according to this photo taken from this review by @tactical_grizzly , it doesn’t look too good: the MCU is exposed on the battery side of the driver board, so piggybacking would be problematic, and the MCU is a lowly Sonix 8F570200A, which has only 4KB of flash and 256 bytes of RAM memory, many times smaller than the minimum needed for Anduril even if the rest of it (eg, being 8051 instead of ATTiny based, etc) didn’t already made impossible to port Anduril to it. Too bad as per the graphs in that review I just mentioned, it seems the L7 indeed blows all the others right out of the water.
A lot of food for thought here, and I will definitely consider what you stated if I change my use case to need more than short bursts at turbo.
Very good. Yes, I realize short bursts (like 20-30 seconds max) is all that’s going to be possible with this LED in a small host and a single battery (I just amended my original post to make that clear).
About efficiency, is there any regulated drivers for these lights? I would like to have one if possible, even if sacrificing moonlight (I already EDC a TS10 for that).
And good idea re: being able to use 18650s: I even have a 18650-battery-to-21700-flashlight plastic adapter, I just hope it doesn’t heat-insulate the battery too much and lead it to baking itself inside the light. And I understand I will need to buy high-current 18650s, as my LG MJ1s top at 10A max – and I hear the SBT90.2 can drain up to 25A on turbo – so I might end up spending less and getting better capacities by buying a few 21700s anyway.
I see your point… as long as we’re moving out of pocketable-light territory with the K1, why not go all the way and get something larger but with much better thermal/electric/optical capabilities as well as a nicer price like the 3x21D, right? Problem is, no Anduril (I already amended my OP to make that clear).
and has loads of space in the driver pocket to put in a piggyback running Anduril, or fitting an all-new buck-boost + FET driver in it. I have one with a DM11 driver stuffed in, and it breaks 1Mcd with some minor hotrodding.
Now that’s intriguing! And would of course solve the Anduril issue in spades. A few questions:
- Does Hank sell the DM11 driver all by itself, or would it be necessary to cannibalize a DM11 in order to get one?
- Have you posted pics or whatever from that Hankonvoy (Hank brains on a body by Convoy
) anywhere? - How difficult was it to swap the board? I am reasonably handy and know how to use a soldering iron, but don’t have any experience with SMD (and I guess neither the proper tools like a heating plate for reflowing etc). Do you think it would be viable for me to DIM?
TIRs do suffer from greater losses on light transmission than reflectors, especially large ones. The reflector based lights will be brighter due to less light being lost.
Either way, the most important factors in running a Sbt90.2 in a singl cell host smaller than a WT90 are a good thermal map on the driver and a well regulated driver. The Manker MC13 II SE i did is a good example.
On that same vein, what about the Convoy 4x18C? IIt seems to have even better sustained performance than the 3x21D (sources for these graphs: 4x18C, 3x21D); the 3x21D has more reach at 1816m vs the 4x18c at 1162m, but I wonder if that’s practical (I don’t think my eyes would be able to discern anything even at 1000m without a telescope/binoculars) and there’s value for me in staying with my LG MJ1s (as 4 of them would be more than capable to supply the 25A the SBT90.2 needs).
What do you think, would the 4x18C be a good candidate for receiving a DM11 brain transplant? 
If you decide to go with an 18650, get the newer Molicel P30B, otherwise the best suited 21700 is the P45B. My hottest 18650 is the slightly older P28A, but I’ll be testing the max intensity between both lights a bit later today.
If you want to drive the lights as hard as possible without any regard to capacity, the Lishen LR2170LH is the best cell available, but it is only sold by Neal as the LR2170HP and by Vapcell as a special order item. It was those LR2170LH cells that allowed my 3x21D to hit 1Mcd, though it was very close with P45B at 985kcd.
It might be possible for the 3x21D to reach 1Mcd with P45B by doing some hotrodding to the host, most importantly replacing the tailcap spring holder with a copper plate with BeCu (or bypassed) springs.
Yes, but any of Hank’s linear+FET drivers should work fine. You can buy them as an email request, iirc it was $12 for the linear+fet drivers. I specifically chose the DM11 driver because of the larger board, it isn’t as tightly packed with tiny SMD components and only had them on one side.
You want to request a large size switch PCB to go along with it, K1 or K9.3 sized.
So far I’ve done 3 (M21C-U, H1, 3x21D), but I don’t think I’ve posted the build for the 3x21D on here yet. It was kind of tricky to put together but that was mostly because I had to figure out where each thing went.
Replicating it shouldn’t be too difficult especially if you have access to a hot air rework station, I just had a soldering iron, tweezers and a microscope (optional but convenient).
Regarding the 4x18A, IIRC the driver pocket is the same so the mod should also fit. I would still use Molicels in it though, as the modified FET driver will draw as much current as it is given, and the SBT90 will still keep increasing in brightness past 30A.
The regulation should be the same as any linear+FET anduril light except that it uses the FETs on the original driver to provide an optimized current path, so the maximum brightness will be a lot more than it normally would be. I haven’t been able to measure it but I would put the ballpark at more than 5000lm.
1 Thank
Just tested my SBT90 D1v2 and D1K. Used a UT383-BT lux meter at 410cm distance. The difference was pretty surprising.
Output measured at peak, using 2H momentary turbo.
D1v2 (P28A): 92.6kcd
D1K (P45B): 126.8kcd
1 Thank
Get the D1K. Higher output, better battery life, and marginally better thermal performance.
2 Thanks
Incredible! So, while the D1v2 is a bit (~7%) under, the D1K is impressively over (~27% !) their stated 100Kcd specs. As they’re very close it not exactly the same re: optics, and probably ditto re: driver, I wonder whether the only explanation for that isn’t the P45B in the D1K being capable of supplying way more current (45A vs 35A) than the P28A. Arithmetic seems to support this as 126.8/96.8-1= 31% ~ 29% =45/35-1…
If that hypothesis is correct, then feeding a P28A to the D1K would make it produce about the same candelas as the D1v2, no?
2 Thanks
It is in fact entirely from the battery. The reflectors are identical, and the drivers are almost identical too (21700 driver has a wider contact ring on the outside), being a simple FET in turbo. Keep in mind though that brightness does not increase linearly.
Yes putting a P28A in the D1K would cut the performance to D1v2 levels, and putting a LR2170LH would improve it further (small gains in performance compared to P45B in all the lights I’ve tested it).
3 Thanks
Thanks for these pics; I love to see what’s inside stuff, flashlights included!
You mean human vision registering brightness in a log2(N) way, right? Yes, I’m aware of that regarding lumens; wasn’t aware it also applied to candelas, but OFC it does as AFAICS candelas is just concentrated lumens (lumens/cm2).
Thanks for the confirmation. I just went to Google and searched for the LR2170LH, and was led right back here to BLF and this post of yours: Solved: Solving the mystery of the Nealsgadgets "LR2170HP" 21700 ultra high current cells
Interesting to note that the LR2170LH is capable of supplying 50A vs 45A for the P45B (so only ~11%+) while storing only 3000mAh vs 4500mAh of power (therefore ~ 33% -); as human perception of battery duration is quite possibly linear, I wonder whether the added CDR is worth the lost capacity…
Thanks for the advice @wolfgirl42! In terms of small host options I’ve basically discarded the D1, and I think the D1K is indeed the best option for my use case.
1 Thank
While I couldn’t tell you for sure how much the LR2170LH can actually supply, I know for sure it is more than the P45B.
The official rating is only 30A, but 3 of them and a good driver are able to push an SBT90 to it’s absolute limit, and that’s accounting for losses in the tailsprings and wires. It’s probably getting something along the lines of 30A to the emitter, which is where the SBT90.2 current/output curve starts to plateau.
1 Thank
Identical driver to the D1v2, and the same optics as well. I have a few KR1s but none with the SBT90. They’re a nice host but the 18650 battery limits the potential output, and I wouldn’t trust myself not to overheat the emitter in momentary turbo, because of how much further the hand is from the heat.
1 Thank
Of course, I forgot the KR1 is a 18650 light 
then it’s out just like the D1, and only the D1K stands in terms of really good small-host SBT90.2 Hank options.
The only two other options with a Hank brain are then the K1 (as Hank itself calls it, a “medium”) or a large one like the Frankenlight/Hankonvoy 3x21D/4x18C with a DM11 driver board like the one you made.
It’s interesting to note that going the Hankonvoy route would not only yield a (way) more capable light, but also be less expensive, as it would cost just $75 for the 4x18C plus the $12 you mentioned for the DM11 board… round it up to $100 and it’s still quite a bit less expensive than the $140 Hank asks for his K1 (which AFAIK is his best-performing SBT90.2 light).
Of course, a lot of elbow grease is required to assemble the Hankonvoy… which makes me wonder why some enterprising soul isn’t offering it yet for us less-capable dummies (hint, hint! )
If you are doing the mod to a 4x18 you would also need an additional “28A” FET driver, which adds a bit more costs. It wouldn’t throw as much as the K1 either, having a smaller reflector. Maybe just 300-400kcd compared to 500-600k in the K1 or 900k in the 3x21D.
If you get this mod I would definitely recommend the large 3x21D host, the beam is very useful as far as a searchlight goes, visibly being useful up to 600-700m in the city with the naked eye, and still being visible at over 1km with a bit of magnification.
In my case the answer is unfortunately high shipping costs and low time availability for doing heavy modding. Though I would be glad to teach someone how to do it, so that they can have a very unique light. I think if you can do an emitter swap, you can do this too. It just takes a bit more time, and maybe some swearing too.
If you really, really, wanted me to put one together, I might be able to do it but the price might give you some sticker shock, and that wouldn’t include shipping (never can be sure of prices with the peruvian postal service either)
1 Thank
I have ended up with 4 of the “HP” cells from Hank. Capacity tests show about 2700 to 2800 mAh. The P45B around 4350 average.
The output, based on testing of lights like the Valkyrie and TS30S have shown ~500 or more lumens increase with those lights using the (Lishen?) HP cells. For a few seconds… But this from lights with 5000 to 17,000 lumens. Not something that I can see. So I use the P45B cells. In fact, in the N103S I have been using the Q26800 cells. They drop output even more, but as I use lower levels much more than the MAX output, the added run time is well worth it. FWIW.
IOW, the HP cells (mine are 4 different types, so one never knows what they will get when ordering), especially for the price of ~$15 per, is not worth it. For me anyway.
1 Thank
Thanks for chiming in! 
Yeah, I wasn’t planning on buying any cells from Hank (basically because I haven’t seen any being offered ), but now I will actively avoid them…
Whaaat? $15 for an unknown, supposedly “High Performance” cell (supposing that’s what “HP” means, and not “Horse Prodder” or similar 
) definitely doesn’t sound like a good idea, better to buy known quantities like the P45B at the most tradicional stores like 18650batterystore for less than $10 each… even if the idea of a flashlight coming empty all the way from China makes me sad. 
1 Thank
He hasn’t sold batteries for several years. Caused too many shipping problems.
The cells are actually the Lishen LR2170LH that I am referring to. You can special order them from Vapcell for around $2 per cell. The performance of them is unbeatable, but they do run out pretty quickly.
2 Thanks