Gentlemen,
The following is an account of what happens when I get bored and have a box full of emitters to play with. Since the Cree XML became the new Belle of the Ball, a lot of torches are getting modded to accept one, three, or in one case seven of 'em (underdriven). Curiosity got the better of me one night, so I set about seeing what I could do. I wanted a torch that could meet the following objectives:
- 6 xml's driven @ 2.5-3 amps
- Built on a 3D maglite (because I had one handy, no other reason)
- No Li-ion (don't want any "poofs" with a multi-cell setup)
- Must not be a 30sec "wow" light (i.e. good thermal management for >20min)
Let me preface this post by saying that this isn't a completed project (to me, anyways...). Consider this more of a proof of concept and a working design.
Here's how I went about it:
After doing some basic quick calcs, I realized that there's no way the stock maglite head would reasonably meet the above requirements, so a little time was spend designing a new one. I started with a 4"x4"x6" hunk of pure aluminum, mainly because I had a scrap piece on hand. Plenty of time at the lathe produced a rough, but functional prototype head. Looks like this here: (no picking on the not-so-smooth finish)
and here it is as compared with the stock head off of a 3D maglite:
What's hard to see is that where the emitters are mounted is over an inch thick, as compared to most factory pills that run a couple of mils. Not knowing how radical I'd have to go with the fins I left them relatively shallow.
So...now it's time to stuff it full of XML's and get to testing. For consistant and accurate thermal testing I used a regulated bench power supply and a Fluke 724 for temp monitoring. These two shots are at 5min and 10min respectively:
The thermocouple is mounted inside the center lead hole between all the emitters. That, combined with the thermal conductivity of pure aluminum should give a pretty accurate idea of base temperature. Interestingly enough at about the 15min mark the temps were at 125F and at 20min was at 130. After a half an hour temps were a bit over 135, which is were I stopped the test. Note that this was with a full 3 amps going to each led for the entire duration. The cells I plan on using won't be able to sustain this, and the head will be attached to a little over a foot of maglite tube so thermals should stay in check.
The following series of pictures shows what 6 Cree XML leds look like at various current levels. Those levels are 500ma, 1a, 2a, 3a, and just under 4 amps. There's been talk about a sweet spot around 2.5a for this emitter, and in testing I saw nothing to disprove that. The difference between 1a and 2a is substantial, but 2a to 3a less so. Output slows rapidly after that. The difference betwen 3a and 4a was much less than the difference between 2.5 and 3. Between 4a and 5a I could disern no noticable difference. I didn't use a lightbox to tell, only a visual evaluation...and honestly, for me personally it doesn't matter a bit if there's a small lux gain if I can't tell by looking at it. I don't in daily practice use my torches to illuminate lux meters.
The "amps" meter shows the current in each of these shots:
So, now come the part were we need to get this somehow mounted and powered so it's actually useful as a torch. Considering the softness of pure aluminum, threading was out but that's ok, since my threading ability on a lathe is lacking. Tight tolerances on the other hand I can do, so I made it a press fit with a set screw. The negative lead in the series is mounted to the head itself, while the positive is fed down and mounted on a plastic dowel that fits perfectly into a stock incan mag bulb holder. That's right, it's a slip-on with no mods required to the host other than removing the head and bulb. Here's a shot of the post and set screw:
Powering all of that xml goodness is another trick, sort of... Being a minimalist and a subsciber of the KISS method of engineering I used 12 NiZn AA cells in series for 22.25v fully charged and run all of it direct drive. This simplifies a few things, first and formost saving cost and complexity of a driver of which few out there could handle this, and also prevents the stock mag switch from having to see more than 3.2amps. Win-Win! I do miss out on modes and that ever useful SOS mode (blah), but who am I kidding...this isn't a EDC. The neat thing about these cells are their ability to provide high current at a higher than normal voltage. As Steve Jobs would say, here's where it gets "magical". The system provides a self regulation effect of sorts in that as the emitters warm up the vf increases thereby limiting current and heat. What I've noticed is @ 3.2-3.4a at start-up, and then after about 5 minutes current drops to around 2.5-2.8a and holds there for about the next 10minutes then slowly starts to fall off. As of now, I've yet to do a full runtime test but will update when completed.
Well, enough about the boring details...Let's see how it all turned out: Here's the before and after shots:
And a nicer one using the wife's fancy napkin as a background: (kinda looks like a M24 german hand grenade...not intentional...)
As you can see, I used Carlco TIR lens's. They are the narrow one's spec'd for mc-e emitters. I did try using the narrow ones for the xpg emitters, but the beam looked worse. Why did I use these? They were cheap and I had a bunch on hand (notice how I work now?). Let's see how they perform with some beamshots. I'll be comparing it to the Zebralight SC60 and the ever popular MF xml drop-in in a solarforce host. Each set will consist of a control shot, the SC60, the MF drop-in, and then the new creation in that order. (it'll need a name eventually...maxi-match? - almost sounds like a feminine product...).
Please note that there was a full moon out and the pollen count is off the charts in South Carolina this time of year, so distance shots didn't turn out so well. I'll update them when it finally clears up.
Here's the first shots at the big oak tree in my back yard. All exposures set to 2sec:
Distance @ 35yds
Control:
SC60:
MF XML drop-in:
The big bugger...
Yup...definately not an up close work light.... The reflection off of anything closer than 20yards is actually quite bothersome to look at. Kinda neat! Let's look at some distance shots now. Opposite to my backyard is a golf course, so I snuck over there for some more shots. Again, all the tree pollen in the air kind of ruins the shots and makes everything look real hazy, but this will still give you an idea. I had some 400yd beamshots, but after reviewing them they were useless due to the pollen, so here's the midrange shots:
Tree line is @130yds:
Control shot:
SC60:
MF xml Drop-in:
The new BLF delight "Match-edition": (scratch that...too cheezy...)
I don't have a I.S. yet to measure total lumen output, but lux readings at 1 meter are @ 48klux, making it my second best thrower. Again, kinda neat. Does it meet the requirements laid out in the opening paragraph? I'd say closer than I expected. I'm not sure how to wrap up this build write-up, since I'm really not done yet. I guess really all that's left is just final polishing on the head and making a better battery holder. Oh yea, I did bore the body out to accept 4 aa cells in a carrier, then realized I could just stick with 3AA 4 deep by removing the spring and boring out the tailcap. In a nod to Don, it's all held together by electrical tape and popsicle sticks .
This took longer than expected to write up, so I'm sure I missed something. If you have any questions, please ask away!
Thanks for reading and humoring me,
-Match
Update:
It's been awhile, but here is a much needed upgrade by the addition of a fantastic driver called the H6Flex. The whole works is now being powered by 4 26650 LifePO4's on a 3D host. The difference is outstanding. It used to be the light only had two modes : Pitch black and Rapture. Now I have a choice of 8 different modes from 35 lumens up to over 5,100lumens...and all current regulated. The UI is really outstanding on this driver and worthy of a write-up all by itself. Suffice to say that this is the one light now that can truly do it all from emergency hurricane light with a runtime over 2 weeks in the low modes to impromptu fire starter in high mode while pointed at combustibles...and it also makes a dandy blunt weapon to boot!