I have been working on a triple 3-up XP-G2 Maglite. I figured I would share the build process. Keep in mind I have not done any research as far as what others have done for similar projects. I wanted to go into this fresh and iron out any engineering issues from scratch. So if anything seems odd, out of the ordinary or outright wrong, that could be one reason.
Update: This project won third place in the LEDSupply.com LED project contest. I purchased the LEDs, drivers and optics for this project from their store.
Blog article: http://www.ledsupply.com/blog/7-cool-diy-led-projects/#more-146
The foundation of this build is a dark green 3-D mag. This was basically chosen based on the color.
My plan is to drive each 3-up at 1000mA. I am using (3) Luxdrive 1000mA Buckpucks, one for each cluster. Buckpucks require a slightly higher input voltage than the forward voltage of the leds. The forward voltage I would be dealing with is about, let's say 9.5 volts. Out of all the options I surveyed I chose three Li-ion in series, giving me 11.1 volts. Finding a lithium combination that fits in a mag tube and has reasonable capacity was the challenge. I ended up using two parallel rows of three 1800mAh 17670 cells in series. The resulting package has 3600mAh and 11.1 volts. This battery setup should be capable of delivering the current without straining. The cells are spec'd as being capable of 1.5C. 1.5*3600mA=5400mA=5.4Amps. I'm going to be drawing about 28.5 watts=(9.5 volts * 3 Amps). The current draw for the batteries will be about 2.6Amps=(28.5watts/ 11.1v), well below the available 5.4Amps. Also worth noting is the fact that these are protected cells which is always a good idea when running multi-cell configurations.
Fitting these cells into a 3-D body is no easy task. Three D cell batteries measure about 186mm while three 17670 measure approximately 201mm. This leaves me with about 15mm to make up. Nothing a little material removal couldn't fix. I started by doing the obvious, putting the tail cap spring inside the tail cap. This is easily accomplished once the wide end of the spring is heated with a torch until it's slightly red.
What this does is remove the springiness from that section of the spring allowing it to be reformed. Trying to do this mod without heating the spring is not fun to try and for me unsuccessful, so out came the torch. The heating does destroy the plating on the spring so i sanded it down to clean steal and will have to keep it lightly oiled. The end of the spring that contacts the batteries was left unheated and therefore the plating and springiness on that end is good to go. I also had to do a little grinding to the tail cap itself. I shortened the threads by about 3/32". There is still plenty of meat on the threads.
With the tail cap modified I still needed to make up about 5mm. That left me with only one option, modifying the switch housing. So that is what I did. Much of the mag switch is superficial. I was able saw off a large portion of the rear of the switch while still maintaining the integrity and normal operation. At the same time I cut off the front protrusion to give me the space I will need in the head.
Back:
Front:
With the Switch shortened I had enough space to cram the batteries with just enough play so I know they are not getting crushed. I now needed to make a contact area for the positive end on the batteries. The cells would be side by side so it would require a contact area accommodating this. I cut a piece of brass sheet to fit and soldered it to the positive part of the switch. The black piece protruding though the brass helps to stop the cells from rotating inside the body.
I needed a way of keeping the two rows of cells oriented side by side. My solution was two corresponding wooden dowels to take up the space and keep everything aligned. I cut the dowel to size and put a sleeve of clear heat shrink tubing over them to give them a better finish.
Now I needed to make a negative end contact that would allow the spring to bear against it while also resisting the rotation during screwing on the tail cap. This was one of the more perplexing aspects of this build. I ended up with a brass plate that has interlocking pegs that tie into the cell/dowel arrangement. I soldered the brass tubing pegs to the brass sheet. I also soldered two "dots" to contact the recessed negative ends of the cells. It was a relief to have the battery situation solved. I can work out any bugs later as they arise, but for now it is very functional.
The three buckpucks easily fit into the body in front of the switch and do not protrude from the end of the body. I arranged the drivers and taped them together while orienting the wires in an organized fashion. I'm hoping they will not get too hot, I may have to revisit this area.
I soldered the input conductors directly to the switch, not an easy task in those tight quarters. I padded the drivers with a little soft foam.
I had been thinking of ways to lock the head from turning. If the head is turned when the wires are all connected they will twist and nothing good will happen. I didn't want to do anything external so I went with a set screw on the inside toward the front of the body. Simple enough and highly effective while remaining unnoticeable from the outside.
Ok, so enough of the boring stuff, let's get to the business end of this thing. I purchased three 3-up Luxdrive stars. Said stars have attached some XP-G2 (6500k). According to the specs, each star should put out 1014 lumens @ 1 Amp. This should give me a total of over 3000 lumens. To top off the stars I have Carclo narrow clear lenses. This thing is going to look pretty cool even when it's off.
Now you may be wondering how I am going to keep these things cool. I have already mentioned that I am going to dump almost 30watts into the poor old Mag. I have made a heatsink out of copper. The heatsink disc is comprised of two 1/4" layers of copper soldered together to form a 1/2" think chunk. The diameter is the same as the inside of the Mag head (about 1 7/8"). The heat sink is going to be thermally coupled to he Mag head with thermal paste. I have to imagine this with be suitable for the minor usage I will put this light through. It came out pretty good with the use of a disc sander. The silver spot on top is a threaded hole that was in the copper, I filled it with solder.
I had a tough time deciding the orientation of the stars. It was a toss up between a radiation symbol or the Triforce. I went with the triforce (Because Zelda is awesome). I can't imagine orientation would make much of a difference at all in the operation of the light. I adhered the stars with thermal epoxy. Both surfaces were lapped in order to used the thinnest layer of adhesive as possible. I will route the wire through the middle of each star. Should look pretty clean well all is said and done.
Soldered up the leds. Did it twice because in typical fashion I forgot to feed the wires through the bezel before soldering. So after that minor setback I had them re-soldered and ready to go.
So that is basically it. Just had to put on a glass lens and tighten the face. Tightening the lens compresses the whole lens assembly.
I was thinking the other night that I will have quick easy access to the Carclo lenses if I want to change the beam.
Popped an anti-roll thing on the end and took it's time for a test drive.
The beam is basically a giant smooth flood. That's fine with me, it will be a good area light. Here are some shots comparing it with a Fenix PD22, PD32UE, and TK22.
PD22 (200 lm):
TK22 (650 lm):
PD32UE (740 lm):
AND THE NEW BEAST:
So I hope you enjoyed my build, please comment. Thanks.
Update: This project won third place in the LEDSupply.com LED project contest. I purchased the LEDs, drivers and optics for this project from their store.
Blog article: http://www.ledsupply.com/blog/7-cool-diy-led-projects/#more-146
The foundation of this build is a dark green 3-D mag. This was basically chosen based on the color.
My plan is to drive each 3-up at 1000mA. I am using (3) Luxdrive 1000mA Buckpucks, one for each cluster. Buckpucks require a slightly higher input voltage than the forward voltage of the leds. The forward voltage I would be dealing with is about, let's say 9.5 volts. Out of all the options I surveyed I chose three Li-ion in series, giving me 11.1 volts. Finding a lithium combination that fits in a mag tube and has reasonable capacity was the challenge. I ended up using two parallel rows of three 1800mAh 17670 cells in series. The resulting package has 3600mAh and 11.1 volts. This battery setup should be capable of delivering the current without straining. The cells are spec'd as being capable of 1.5C. 1.5*3600mA=5400mA=5.4Amps. I'm going to be drawing about 28.5 watts=(9.5 volts * 3 Amps). The current draw for the batteries will be about 2.6Amps=(28.5watts/ 11.1v), well below the available 5.4Amps. Also worth noting is the fact that these are protected cells which is always a good idea when running multi-cell configurations.
Fitting these cells into a 3-D body is no easy task. Three D cell batteries measure about 186mm while three 17670 measure approximately 201mm. This leaves me with about 15mm to make up. Nothing a little material removal couldn't fix. I started by doing the obvious, putting the tail cap spring inside the tail cap. This is easily accomplished once the wide end of the spring is heated with a torch until it's slightly red.
What this does is remove the springiness from that section of the spring allowing it to be reformed. Trying to do this mod without heating the spring is not fun to try and for me unsuccessful, so out came the torch. The heating does destroy the plating on the spring so i sanded it down to clean steal and will have to keep it lightly oiled. The end of the spring that contacts the batteries was left unheated and therefore the plating and springiness on that end is good to go. I also had to do a little grinding to the tail cap itself. I shortened the threads by about 3/32". There is still plenty of meat on the threads.
With the tail cap modified I still needed to make up about 5mm. That left me with only one option, modifying the switch housing. So that is what I did. Much of the mag switch is superficial. I was able saw off a large portion of the rear of the switch while still maintaining the integrity and normal operation. At the same time I cut off the front protrusion to give me the space I will need in the head.
Back:
Front:
With the Switch shortened I had enough space to cram the batteries with just enough play so I know they are not getting crushed. I now needed to make a contact area for the positive end on the batteries. The cells would be side by side so it would require a contact area accommodating this. I cut a piece of brass sheet to fit and soldered it to the positive part of the switch. The black piece protruding though the brass helps to stop the cells from rotating inside the body.
I needed a way of keeping the two rows of cells oriented side by side. My solution was two corresponding wooden dowels to take up the space and keep everything aligned. I cut the dowel to size and put a sleeve of clear heat shrink tubing over them to give them a better finish.
Now I needed to make a negative end contact that would allow the spring to bear against it while also resisting the rotation during screwing on the tail cap. This was one of the more perplexing aspects of this build. I ended up with a brass plate that has interlocking pegs that tie into the cell/dowel arrangement. I soldered the brass tubing pegs to the brass sheet. I also soldered two "dots" to contact the recessed negative ends of the cells. It was a relief to have the battery situation solved. I can work out any bugs later as they arise, but for now it is very functional.
The three buckpucks easily fit into the body in front of the switch and do not protrude from the end of the body. I arranged the drivers and taped them together while orienting the wires in an organized fashion. I'm hoping they will not get too hot, I may have to revisit this area.
I soldered the input conductors directly to the switch, not an easy task in those tight quarters. I padded the drivers with a little soft foam.
I had been thinking of ways to lock the head from turning. If the head is turned when the wires are all connected they will twist and nothing good will happen. I didn't want to do anything external so I went with a set screw on the inside toward the front of the body. Simple enough and highly effective while remaining unnoticeable from the outside.
Ok, so enough of the boring stuff, let's get to the business end of this thing. I purchased three 3-up Luxdrive stars. Said stars have attached some XP-G2 (6500k). According to the specs, each star should put out 1014 lumens @ 1 Amp. This should give me a total of over 3000 lumens. To top off the stars I have Carclo narrow clear lenses. This thing is going to look pretty cool even when it's off.
Now you may be wondering how I am going to keep these things cool. I have already mentioned that I am going to dump almost 30watts into the poor old Mag. I have made a heatsink out of copper. The heatsink disc is comprised of two 1/4" layers of copper soldered together to form a 1/2" think chunk. The diameter is the same as the inside of the Mag head (about 1 7/8"). The heat sink is going to be thermally coupled to he Mag head with thermal paste. I have to imagine this with be suitable for the minor usage I will put this light through. It came out pretty good with the use of a disc sander. The silver spot on top is a threaded hole that was in the copper, I filled it with solder.
I had a tough time deciding the orientation of the stars. It was a toss up between a radiation symbol or the Triforce. I went with the triforce (Because Zelda is awesome). I can't imagine orientation would make much of a difference at all in the operation of the light. I adhered the stars with thermal epoxy. Both surfaces were lapped in order to used the thinnest layer of adhesive as possible. I will route the wire through the middle of each star. Should look pretty clean well all is said and done.
Soldered up the leds. Did it twice because in typical fashion I forgot to feed the wires through the bezel before soldering. So after that minor setback I had them re-soldered and ready to go.
So that is basically it. Just had to put on a glass lens and tighten the face. Tightening the lens compresses the whole lens assembly.
I was thinking the other night that I will have quick easy access to the Carclo lenses if I want to change the beam.
Popped an anti-roll thing on the end and took it's time for a test drive.
The beam is basically a giant smooth flood. That's fine with me, it will be a good area light. Here are some shots comparing it with a Fenix PD22, PD32UE, and TK22.
PD22 (200 lm):
TK22 (650 lm):
PD32UE (740 lm):
AND THE NEW BEAST:
So I hope you enjoyed my build, please comment. Thanks.
Last edited:
