Uber-Light 1000

Intro:

This Instructable is for a high power flashlight that uses a CREE XM-L LED and is powered by 3 D-cell batteries.  Most of the parts are fabricated using a 3D printer and due to the size/shape, it is broken down into several segments.  The wiring for this flashlight is very simple.  It composes of the 3 batteries in series, connected straight to the LED via a rocker switch.  There are no resistors in series due to the fact that there is enough resistance from the batteries (alkaline), wires, and switch.. and because the CREE XM-Ls are beastly LEDs.  Granted, this is a very inefficient way to power these LEDs for various reasons, but it does work and is really bright... battery life might just be a bit lacking... and the light output may be a bit inconsistent from beginning to end of battery life.  The middle segment of the body is sized to be the length of a D-cell battery so one can add extra segments to make it a 4+ battery light.  This will of course require different wiring/electronics... not covered in this Instructable.

---

Instructable Contents:

• 1:  Tools, Materials & Parts
• 2:  Print UL1000 Components
• 3:  Prepare LED module
• 4:  Assemble flashlight head (heat sink/LED module)
• 5:  Prepare positive circuit wiring
• 6:  Assemble flashlight body
• 7:  Prepare/install end cap
• 8:  Appendix A: Extra information

Step 1: Tools, Materials & Parts

Suggested Tools:

• 3D Printer
• Soldering Iron
• Heat gun
• Electric drill w/ 5/32" drill bit
• M6-1.00 tap
• 8-32 tap
• 5mm Allen wrench
• Phillips head screwdriver
• Wire Strippers
• Pocket knife
• Emery cloth or sandpaper
• Small frying pan/skillet
• Tweezers

---

Suggested Materials/Parts:

• ABS printer filament
• Flux core wire solder
• Solder paste
• Heat sink compound
• JB weld (or JB-Kwik if you are impatient like me)
• 14" of black 20ga stranded copper wire
• 8" of red 20ga stranded copper wire
• 2 - 1/4" x 1" lengths heat shrink
• 2 - female terminal ends (spade)
• Small zip tie
• Cree Xlamp XM-L LED (Mouser PN: 941-XMLAWT0000LT50E3)
• LED thermal substrate (Mouser PN: 951-804936)
• LED lens (Mouser PN: 928-FA11902T3WXM)
• Radial heat sink (Mouser PN: 588-SA-LED-151E)
• D-cell positive battery contact (Mouser PN: 534-5250)
• D-cell negative battery contact (Mouser PN: 534-5251)
• Oval rocker switch (equivalent to R13-133)
• 1 - 8-32 x 1/2" machine screw (Phillips head)
• 8 - M6-1.00x12mm socket head cap screws
• 3 - D-cell batteries

Comments

[jockkitto16]

this is a fantastically made instructable! i would make this torch in a flash (pun intended) if i could afford the 3d printer.

very well done!

[randone]

I am wondering if you have to have such a massive heat sink, or a heat sink at all. I have had other electronics that have never even got warm. Do LED lights get hot when they are this powerful?

[macskyver]

They do get warm, and output decreases as they heat up.. (check datasheet for details).

[aristide202]

I agree, that's the biggest problem with LEDs

[keegancdr]

hey so where did you purchase the electronics? im having a little trouble finding the LED and the thermal substrate

[macskyver]

Sorry, I should have been more clear about that... I got most of the parts from http://www.mouser.com/... you should be able to find them easily with the listed part numbers (but let me know if you don't).. Someone also found pre-mounted LEDs on thermal substrates on ebay for about $10, although mounting them is half the fun and excitement...

The trickiest part for me to find was the rocker switch. You can find them on Mouser.com if you search for "R13-133" (and get the datasheet) but you have to buy them in multiples of like 1000 from them... Although not labeled the same, I found the same switches on Amazon.com:

http://www.amazon.com/JT-Products-2703-2J-Illuminated-Rocker/dp/B002UZW77C/ref=sr_1_9?ie=UTF8&qid=1330992687&sr=8-9

I think I also found them on http://www.parts-express.com.. but I ended up ordering from Amazon because I love amazon prime. Unfortunately, I haven't been able to find the plain unlighted versions, but I know they exist..

[FightCube]

I was wondering about the lumen output of this LED, so I'll post what the spec says here:

The XLamp XM-L LED is the
industry’s highest performance,
single-die white lighting-class LED.
The XLamp XM-L is 20% more
efficient than the XLamp XP-G at
the same current, and can deliver
1000 lumens with 100 lumens per
watt efficacy.

Even if you only achieve half of that, 500 lumens is insanely bright.

Have you looked at the forward voltage and current on an oscilloscope with a fresh set of batteries? I wonder if the peak current and voltage exceeds the recommended spec max of 3000mA and 3.35V?

[macskyver]

With relatively fresh batteries, I measured the current with my multimeter and got 2.7 Amps... Using the datasheet as a guide.. that should mean the voltage is about 3.3V... and outputs about 780 lumens... and yes, it is bright...

[lesizz]

How about lithium batteries? They would last longer and make the led brighter. Only question is whether they would power the leds too much and smoke them.

[zkus]

Awesome. Has a nice, badass look with the giant heatsync.

I was wondering though, why not resistors? Wouldn't that increase the batter life (with i've gathered to be a little poor). Maybe a pot to make the brightness adjustable?

[macskyver]

Thanks.

I didn't use resistors because I wanted it to be as bright as possible with no regard for battery life or consistency.. and because it was simple and easy..

[Mike Nelson]

What run-time do you get from the D cells?

@Demonic69 - i just got 3 of these LED's in the mail with some 18650's... I normally make headlamps, but a flashlight version would be fun too.

[macskyver]

I haven't tested the run-time yet... I did just test the current draw and got 2.7 Amps... which I am sure is absolutely horrible on a D-cell battery. That is outside any of the charts in the battery datasheets I have... I am pretty sure the run-time would be measured in minutes instead of hours...

[ac-dc]

With alkaline cells you will get what is known as voltage depression if it was left running, the cell voltage dropping continously would cause less and less current through the LEDs so you would probably still get at least a few hours of runtime because it is a direct drive design, but it won't be very bright the majority of the runtime.

BTW, you had mentioned in the instructable that this isn't very efficient. It is actually the MOST efficient way to drive an LED since there is no power conversion loss, but with the downside being if you had used a lesser LED it might damage it, and the voltage depression issue mentioned above + voltage/current relationship so it won't be consistent brightness.

[Kinnishian]

I feel like a lot of people may have commented already...If the majority of resistance is coming from the internal resistance of the alkalines, that's chill, butt...... I think alkaline batteries really aren't very good at supplying current as compared to their total capacity.

The test would be...if the battery lasts for 1-2 hours at good operation, then you're not abusing it *too* much (0.5-1 "c"), but if it lasts for only <1 hour, then you'll be getting a lot less capacity out of the battery than you could otherwise.

[macskyver]

Yeah, I agree... looking at a duracell alkaline battery datasheet, you can see that on the "Service hours vs. power drain" graph... functionally speaking, the battery choice is fairly poor... I chose D-cells because they are more common and they do work (although not extremely well) in this application. I have a bunch of 18650s heading my way, so I will modify this light to be a bit more usable...

[jolshefsky]

Re your "if you are impatient like me" comment on JB-Kwik, in my experience, JB-Kwik is the go-to epoxy, and JB-Weld is "JB-Slow" if I happen to have a need for hours of set-up time. It's peculiar naming on their part.

Also, JB epoxies claims to be a good thermal conductor, so if you want to attach something to a heat sink, JB epoxies are a good option instead of thermal paste and screws.

[leviathan937]

Just out of curiosity, does the JB brand list the thermal properties of their adhesives anyway?
It would be good to compare them to a dedicated product like Arctic Silver adhesive, or even double-sided thermal tape.

[macskyver]

I didn't know that about JB Weld being a good thermal conductor... That is good to know, thanks.. I swear, that stuff is magical.. Another random use is as a filler putty (like bondo) for things that get powder coated..

[tmack0]

Actually, epoxy in general is a decent thermal conductor, in the 0.3-1 W/MK range, as compared to air at ~0.025 (but still way less than copper at 400), and is usually an electrical insulator. Its commonly used to pot electronics. JB Weld is about average of the lot, but for something as powerful as an XM-L, you would probably want to spend the extra for Artic Alumina. Its similar, and made by the same people that make the more well-known CPU thermal paste "Arctic Silver", but is an electrical insulating 2-part epoxy designed for high thermal flux (ie: gluing hot stuff to heat sinks), so gets better than 1W/MK. It uses ceramics instead of silver to maintain its electrical insulative property. There are quite a few threads on this over in the candlepower forums (people that do this kinda thing often, and in even more excess).

Things to watch for: keep the application of it thin as possible as the stuff its gluing together has a MUCH higher conductance, and its expensive so excess is just waste. The thinner it is, the less resistance from the stuff regardless of its coefficient. Also, once set, this is an epoxy, so it will not move, make sure you put the parts together correctly!

[ac-dc]

Epoxy in general is not a decent thermal conductor, nor is JBWeld. There are certain epoxy formulations highly fortified with things like metal or ceramic particles that do a fair job. The only way some people get away with using epoxy is that it's an extremely thin layer applied.

Thermal conduction is not a matter to be taken lightly. Never substitute an epoxy not specifically designed for high thermal conduction if at all possible. Otherwise there is excessive heat and or excessively large/heavy/expensive heatsink needed to compensate for the wrong thermal interface material.

Also potting electronics is not the primary thermal path for anything specified to need heatsinking (just wanted to clarify this in case anyone got the wrong idea).

Lastly, JBWeld cures to a granular brittle structure, it is not the right formulation for anything with a different coefficient of expansion or subject to possible impact like a flashlight that might be dropped.

I agree Arctic Adhesive is a good choice, or for larger area applications it is more economical to buy a tube of one of the various brands sold for use on peltiers.

[rmoore28]

I think your design is very cool especially with the glow in the dark ends. Would you please write me at [e-mail] and let me throw a couple of ideas at you? Richard

[clazman]

Very nice!

I would only suggest a small change.

To use BHCS's to replace the SHCS's on the business end. They'd present a much cleaner look.

Again, very nice job!

[macskyver]

I agree... a button head cap screw would be cleaner.. good idea.. thanks.

[bhouston]

Is your 3D printer homemade? If so, do you have an instructable for it?

[cc67]

I saw an instructable that can answer you problem
http://www.instructables.com/id/Build-a-Laser-3D-Printer-Stereolithography-at-Ho/
check that out... maybe that can answer you

[macskyver]

My printer is homemade, but unfortunately I don't have an instructable for it.. Fortunately though.. there is tons of information on the printer I built. There are already really good step by step instructions. This link will get you started.. (mine is the Prusa Mendel)

http://reprap.org/wiki/RepRap

I am thinking of making a scaled up CNC router version of the Prusa Mendel, in which case, there will most likely be an instructable for it..

[snowluck2345]

this is cool. I've done some maglite mods and such but this is just an awesome idea. i love the heatsink. it must be able to take a lot of heat. it would be cool to do a sealed light that was rechargeable and to have a small fan inside.

[macskyver]

Thanks. Yeah, the heat sink is pretty massive... and probably bigger than it needs to be.. I don't like running it continuously, because it is harsh on the batteries... so I haven't really tested to see how hot it gets... but there are shorter (and longer) versions of it with the same bolt pattern so people can modify to their heart's content if needed...

It sounds like version 2 will definitely be rechargeable (with 18650s) and water resistant.. A fan would be kinda pointless on a sealed design, but it could be a cool mod for the open design.

[snowluck2345]

yeah, i've been working on a few flashlights and i just confused them when speaking. I was thinking a simple body for one with a fan with a grate and a computer style heatsink and i've designed a sealed light based with heat pipes through it. I'm thinking about how i'll make it and i'm thinking 3d printing, making a plaster mold of it, and casting it in aluminum.

[Demonic69]

Very, very cool!
I wonder how much work it would take to make this waterproof?
If you need a tester for a WP version I'll happily volunteer to try it out at 30m or so :D
Would also be great to see it running on 18650 cells, at 3.7v each I reckon you could get a good runtime!
If i ever get my hands on a 3d printer I'll definitely give this a go, brilliant work.

[macskyver]

Thanks... I don't think it would be too difficult to make this water resistant.. It would probably take the following:

- Sealing plate between heat sink and body head
- O-ring between end segment and end cap
- Silicon around the toggle switch.. and most likely a different switch type
- Silicon/hot glue around the LED/thermal substrate

I like the idea of using 18650 cells... I'll do some thinking on that and get back to you... Also, if you want to design/draft a waterproof mod or 18650 cell mod (preferably in OpenSCAD)... I'll print a body for you. Just make sure to provide room for some control circuitry...