Thank you for this advice, it worked! It is indeed possible to remove scratches from led dome with wet q-tip mounted in micro drill, however must be done slightly and q-tip must be always wet. I had not plastic polish so I used a tooth paste and even with it the dome now looks much better! For the Cyclone I’ll go with new emitter as I already ordered one, but certainly will not throw this one away and will use it in other flashlight.
Looking forward to the updates.
Was it expensive for you to get the extension made?
I was just thinking to myself it would be really nice if the whole middle section was 1 piece/tube, then maybe strip the anodizing and polish it all up or paid the head and tail and have gold/silver.
No no, it was not expensive, extension tube was made for about $10. With more pieces at once the price could be better, but I was glad the man was willing to made me only 1.
I didn’t updated the first post yet, but if someone like the extension tube just machined without anodizing, this could be the price + shipping. Without anodizing, I’m able to get even 1 piece only.
Btw. you posted in earlier post that you like the flashlight with silver tube, as on picture - I still didn’t painted mine because I’m liking it silver too 
that’s really impressive work, especially on the direct mount LED. Did you mill the post hole in the star? The battery holder is a very neat piece of work too, well done!
Well I thought this was the easiest way of how to bond led directly to copper and still be able to use it in c88 I drilled the hole with micro-drill in the star first and then used mini-files to finish the rectangle shape. I wish I had the milling maschine
The battery holder - I’m sure it will be able to give 3.5+ amps as it is, with default springs used, but maybe I would think of upgrade it later with copper parts, or at least with better copper springs to lover overall resistance - does anybody know where to get thin copper plate and gold plated springs in size close to original ones?
How about this copper foil - this is 5mil I used for wrapping P60 drop-ins. And what about these golden springs? Or these might be better. I can't comment on the spring sizes though.
-Garry
Are you now going to do a runtime test on the 6AA setup?
-Garry
Garry I was thinking I’ll do all the tests at once after the flashlight is completed, so far I was able to supress my curiosity, but now as you asked… you convinced me
I still have the scratched emitter mounted, so I’ll wire it with default 3mode nanjg driver and test runtime with 6 eneloops in the holder on high 2.8amp. Just apms&runtime test.
Thank you for ebay-link to copper plates!! I have those springs from CQG and they are very strong, not good to use in batt.holder, but those from manafont looks promising.
I have some of the Manafont ones, but haven't used them for anything. Let me know if I can get you any details on them.
-Garry
Here are first results of the diy 6xAA battery holder. I have to say first the conditions how it was done to avoid confusion
- batteries are new eneloops 2000mAh, charged about 3 weeks ago, it’s their 2. discharge cycle
- I’m using default 3mode nanjg 2.8A driver
- wires were soldered to driver and led, others were connected only with clips, and all connections were too long!!!
driver~~led about 0.5m, batt.holder~~>driver about 2m together with dmm
Clips and long wires are not good here and I didn’t realized that it will cause a problem to reach 2.8amp, so I started just around 2.4A
(when all was ready I was lazy to go and reconnect everything with shorter wires, so went with 2.4amp as start, also must note that the holder is able to give 2.8amp without problem as I shortly tested before)
Important points are:
- total consumed energy was 3366mAh, (1683mAh per battery) than the led started blink
- 60min the current remained more than 2.1amp, which is 750+ lm for xml-u2
- 1:20 >1.5amp, which is 600+ lm
- 1:35 >0.75amp, still 300+ lm
After the led started blink I set medium level and it was running another 30min, than the eneloops were ready to charger
Nice testing. You certainly put a lot of work into this. Thank your for sharing your idea with us it is a good idea since there is very few 6*AA lights out there with proper LED+Driver+reviews/tests. TY
Hmm. . . runtime doesn't seem all that different than 3 Eneloops measured by _the_ in his review. I know _the_'s current draw was only 1.78A though, so perhaps that's the key. You're getting higher output with about the same runtime. I was hoping to see more runtime myself.
Thanks for doing this!
-Garry
Well, if he’d replace driver with something near 1.78A (5 AMC7135’s would be closest, with 1.75 - 1.9A depending on whether they are 350mA or 380mA binned ones), he’d get double the runtime :Р
+1. Whay did you heat the pill with to solder the centre section?
Shadowww is right, at 1.78amp it would be about 2 hours with 6 eneloops.
I see the problem in that the lights are usually tested for max runtime on max.output until led starts blink or the light shut-off, which in case of nanjg driver means the light output varies from max.lumens until let’s say 300lm. Then only runtime value is considered which may be a confusing as in this case. The point of using 6xAA is getting real high output, I would say at least 750+ lumens, for longer time than with 3xAA. As I pointed out in the test, in this case it was 60min real high output.
Then of course, we cannot get more energy from the battery than it has, so even 1 hour output at 2.8amp is not possible with 3xAA. Garry, you will definitelly has more runtime with 6xAA than 3xAA, I’m not sure I explained it good enough.
Gas stove in the kitchen I put a metal plate on the burner and copper heatsink on it to avoid very fast heating.
I'm trying to solder the earth from the driver to a heavy copper pill but this would not be a good idea with that as it would melt the whole driver. Nice job though.
Hows youre progress, jamio?
Hello! It was a while since I was here and that’s because of very little free time during summer. There are so many new interesting posts here!
Anyway, the flashlight is done, I only used it yet and not tested for values. Will do the tests and post all about it.
OT: I see there is new thread watching system now and the old is disabled, I missed the date of migrating, oh nooo :cry:
Hello! Build of this flashlight is completed and I finally made the time to put together how the project went till completion. Let’s see
- I have installed U2 1C led and used the light a while, then changed back to neutral white T6 3C
The U2 1C led was pure white, no blueis, no green and about 15% brighter than T6, but I could not help myself, I like neutral color much more
- with above change, I had to create another star to fit the copper heatsink, as I accidentally tear off one of contact pads. Not that much work though, I only had to make a new hole in the star so led is directly bonded to copper heatsink. Here is the new one:
The heatsink with installed driver:
- changed front glass for coated one from intl-outdoor
It is a little bigger diameter than the original so now there is a small gap between bezel and head, I put o-ring there, looks okay.
- changed tailcap for green one
- painted extension tube black
- applied clear coat spray to the inner side of front bezel
I noticed after removing the bezel that it is creating rings in beam (visible on wall), a thin layer of clear coat fixed it perfectly.
- added +one 7135 chip to driver, 3150mA
I’m sure the copper heatsink could handle much more, I guess output with 4 amps would be very interesting, but as I plan use Cyclone only with AAs yet, I don’t want to kill eneloops. Also the flashlight is not overheating itself and it is ok to use it on max for whole runtime, as seen on test below.
.
- programmed the nanjg driver for 6 modes 1 blink mode, modes are (in lumens) 1,10,65,210,480,940 strobe pattern.
As I don’t have an IS for exact measurements, I ‘calibrated’ the best I was able to measurement space with my Fenix L2D and values are created with it’s calculated lux to lumen constant.
.
.
Okay, so here are few more pics of the finished Cyclone:
Will be testing with this:
.
.
Test results:
I used fresly charged 2000mAh eneloops, with eneloop XX runtime results would be 25% better.
I did only runtime, lumens and temperature and not measured amps as I was not able to do it all at once. But measured 3.15 amps at start and 350mA at the end. At the end eneloops were about 1V.
Temperature is measured on head. Heat is very well transfered across whole body, tail was always about 5C colder than head.
.
Measured output on max: 940lm
.
Throw (at)1m, (calculated from throw (at)5m): 39500lux
.
Modes with calculated runtimes:
I like to have more choices than default three so went with six, change of brightness is well visible between each mode
moonlight low 10ma 300+ h
10lm 30ma 126h
65lm 190ma 20h
210lm 660ma 5.8h
480lm 1550ma 2.5h
940lm 3150ma 1.2h
.
Now rutime test on max 3,15 amp, ambient temp. 22C:
.
I’ll shoot some beamshots outside and add comparison shot between U2 1C and T6 3C later.
And one more thing - I want to thank everyone for help, advices, specially drjones and tido for nanjg programming stuff, sixty545 for led-calc, match for testing data - it was all very helpful! I would not be able to do this without help here on BLF!
.
.
Edit: I had a chance to test with Keeppower 18650 3400mAh
throw on max. 3.15amp at 5m was 1700 lux, so it’s 42500 lux vs. 39500 lux with 6xAA