Another 'what's wrong with this picture?' thread (5x XML SRK- wait, what?)

Flashlight Modding and DIY Parts
16

With the XREs the necessary wire clearance was built in to the design of those LEDs and their boards, with XMLs it had to have this relief cut for clearance. As it is in that pic it's technically deep enough, but just barely (I'm using 22AWG silicone wire on everything from now on, the teflon stuff is just too hard to work with, it likes to rip off solder pads and generally be a pain in the ass). I'll probably take another cut of something like .060-.080" in that same pattern before switching to the smaller bit and removing the bridges between holes (the one that cuts fast enough to not drive me crazy is too big to fit inbetween).

17

Ohhh what can be done with proper tools, I yearn to be handy!!

18

Proper tools? LOL Not hardly, just a $130 Chinese SKIL drill press, and an assortment of Dremel carbide bits! One of these days I will pick up a small cross slide vise, as it is now I have to slide the parts around with my bare hands. Knock on wood, haven't lost a finger yet, but the parts do sometimes get snagged by the bit and flung to the other side of the room.

19

Nice machine work and pictures. I like the last picture. Great idea one how to precision cut the reliefs.

20

Plans are to cut out the middle of the stock threaded aluminum plate/pill thing, leaving just the threads and a small ~1/8" wide shelf around the ID, and make two copper plates to sandwich it... top plate will be bigger than the thread OD to seat on the lip inside the head, bottom one will be smaller than the threads so it can pass through. They'll be clamped, then soldered together. Should be just as good at handling the heat yet still be removable. It'll be like a big flat C8 pill, basically.

But I don't have any suitable copper plate on hand currently, I have to get some before much of anything else can happen.

Or... I could use aluminum, and TIG it...

21

Thats fantastic work there comfychair. Very creative. Do you have a problem with the chuck falling out of the taper in the spindle?

22

No it's a cheapie, the spindle isn't removable and has a JT33 male taper on the bottom. The chuck is garbage, jaws aren't parallel... major PITA to get the tools chucked up so they spin true. I'm intending to get me one-a them thar fancy ball bearing keyless jobbies, those make life so much nicer. :bigsmile:

23

This makes your effort even better. You are seriously showing us all how top class work can be achieved with the wrong tools, like a couple off other people here. Thanks. It makes us all think a little harder.

24

25

I dont know whats wrong with the picture. Put me out of suspense and tell me. All I see is a lot off ingenious hard work in a fantastic mod.

26

Does this help any? :p

27

Nup. I’ll stick to what I said. Have you got a plan off attack to polish the reflectors?

28

Wait, so 5 LED SkyRay Kings aren't out of the ordinary? :_(

Reflector polishing isn't hard, just horribly tedious and boring. I can't even attempt it if I'm the least bit sleepy or I get half-hypnotized and start to nod off which is not good when you're holding a Dremel.

I didn't mess up the plating on accident, there was horrible casting flash all over, it was a mess to begin with.

29

Yep very unusual. But I could see nothing inordinary with what you have achieved. Top job. :wink:

30

honestly, there’s space for at least 7 in there, where’s your sense of adventure? :smiley:

31

After starting down about elebenty-nine different paths with the, um, 'driver', then turning around and going back to the start, I finally settled on this:

Since they're all going to be in parallel, no need for the 3-section layout of the original, right? And the AO D436 has plenty of overhead for everything four cells can put out, I think...

(<--LOL)

So I double-stacked them anyway because hey, why not?

I've bench tested it and got 3.40 amps.

Here's what the driver does, as best I understand it (i.e., not at all). The MCU-like thing controls the single 7136, which doesn't see any of the output load at all, it's just used as a sort of relay to control the gate on the D436, as that FET needs more current to switch than the MCU-thing can provide (if this is not how it works, then I am completely lost and can't see why they didn't just delete the 7136s altogether and save a few pennies). The set of three sense resistors (original setup was 3 resistors and one '000' zero-ohm thing per D436, new setup is 3 '000's and no resistors) limit the current the D436 passes. Shorting across the zero ohm resistor pack does not increase the current, but connecting LED- direct to BAT- does, so I have no clue what part of the circuit is doing the regulatin' here. Who cares. If it does better than 3 amps to each of the 5 XMLs, that'll be... yeah. A lot.

The solid copper wire I added is for heat transfer, and also to tie together some of the unused pads to give more space to attach the LED wires so I don't end up trying to solder five wires onto one little pad. For the positive side I will probably drill thru the board at the 3 original pads and solder wires from the battery side of the board rather than rely on the almost certainly undersized vias in the PCB.

What's still left to do is: weld the threaded ring onto the heatsink base (I hope I'm not out of argon, it's been months since I touched the thing), drill wire passthroughs in the heatsink, and polish the damn reflector.

(I found this same 5x reflector at DX for I think 9 bucks, if anybody's curious- don't have the link but it's easy to find, in their reflector category)

32

Lots of awesomeness going on in this thread!!

33

This is cool, I think the R0 resistors are holding this back. You should have destroyed one emitter with four 18650s pushing through that circuit. As I suggested in my thread, try reducing the resistance at that point to get a jump on the current.

34

But jumpering the resistor pack doesn't increase the current beyond 3.40A! I do not understand that! It gets SIGNIFICANTLY brighter if the LED- is connected to the ground ring instead of at the D436's drain!

Where could the resistance be coming from? All I did was sever the 3 traces that originally went to the three 7136... wait. Fake edit time.

FAKE EDIT: I think I see it... I have the MCU's PWM output tied to the 7136's DRV pin, instead of the Vdd. Why did I do that? And why didn't it already catch fire when tested?

35

Ooops :stuck_out_tongue: