I’ve got a TR-J20 coming with the sole purpose of a little mod work. First credit to FlashPilot, A number 1 and LEDMatt for the starting leg work. I plan on filing down the emitter spacer rings, adding a 1 ohm resister that should get me about 4.2 amps at the tail. Each LED should get about 1.2 amps nothing to brag about except that there’s 12 of them. Would changing from the original XM-L T6’s to XM-L2 U2’s really get that much more light? Would holding out for the XP-L’s be a better move? Any input would be appreciated.
Mount the stock XM-L on Noctigon and add a FET driver. You should see around 4A per emitter.
Is there a FET driver with nice modes for 3 cells?
At 4 amps per LED the switch would go poof. 2 amps sounds good, I do want some extended high run time but can the original driver take it? Can the 4 mosfets be stacked like 7135’s? or is a fet driver the best option?
I think running a FET with three cells would see about 3.2A per emitter, but the switch would only be seeing about 13A with emitters wired 3S4P. The switch in a J20 is probably an Omten, which should have no problem.
Having the switch go poof would be a great problem to have. Oh, the bragging rights! If it does, you may want to look into anodizing the threads and convert it into a twisty. If I figure out electroplating, I'll try anodizing next. There are many hosts I'd love to work with, if only they had anodized threads, and most hosts could use better anodization anyway.
FET with Zener mode. RMM has them with five modes, no disco.
Thanks. I thought they only worked with 2 cells. It looks like my BTU Shocker will be getting a driver upgrade and a showerhead cousin.
The stock driver seems to be quite robust and has memory. Maybe try the resistor mod first and see how you like it before going any further (the price is right). Even though the head and floor plate are quite an impressive chunk of thick AL, it can still be turned into a very hot running light in short order if enough amps are sent to the emitters. Also, the emitters are mounted on 14mm MCPCBS. You might impress your light meter, but your eyes wont see a huge difference going from XM-L to the L2 (20% increase is barley perceptible, everything else being equal). If you approach 3A or more per emitter, then the copper MCPCBS’s might be a worthy investment if you can find direct thermal in 14mm… or cut down a bunch of 16mm, which for me wouldnt be worth my time or money. IMO, the glory in this light is in its stock ability to sustain high mode continuously without overheating until the cells are depleted. Any bump in power will hinder that to varying degrees.
As Ouchy mentioned, you can go with a FET driver for mega output, but then you have no regulation. If it gets hot, just drop to a lower mode. Anyone know how the PWM is on the FET driver in the lower modes?
Lower modes? I must try those one of these days! :bigsmile:
The light came today and it is the “CHUNKY MONKEY”. Spent an hour shaving 1.5 mm’s of the height of the spacers. Resistor mod next but other jobs per wife before that. Thanks to all for the input. Did some measurements not even enough room to tribore for 16650’s let alone 18650’s. Rebore the head’s threads and something on this scale http://kaidomain.com/product/details.S022371 but running 3s3p instead of 6s looks nice but pricey.
I know the feeling! J)
Its quite a solid skull-basher, isnt it? :bigsmile: It sounds like you’ve got the tough part worked out. To save some time at the work bench, I was thinking about going with the thinner “wing” type spacers, and clipping the wings off of them. I wonder if there are any reputable 18650’s that would fit a tri-bore if the plastic sleeves were removed from the cells? You probably already know this, but someone on this forum already modded the kai light you linked for 6S. As I recall, the results were rather spectacular.
I fixed the spacers. I took a grinding wheel and just stuck a spacer to my finger and rubbed it on the side of the wheel until it was the right height. The scalloped out the fuzzy left overs inside the spacer. Original spacer still plugs into the hole with no fancy trimming or cursing.
I’ve received a few resistors in the mail. Now i’m charting them in the driver. When the battery draw gets to about 6 amps some of the parts, especially the four mosfets get toasty. Would I be okay if I just put older copper pennies on the mosfets and glued them with thermal adhesive. Ultimate goal would be to dope the whole thing with something like Arctic Alumina. Suggestion needed.
19kHz in all modes (well, except high, which is 0kHz). As a point of reference, stock SRK drivers are something like 200Hz.
Im looking forward to your resistor mod test results.
For heat sinking, you’ll need to attach (whatever copper you chose for cooling the mosfets) to the outer housing and mosfet for it to transfer heat to the air. Otherwise you’ll just momentarily delay the copper from becoming heat soaked and accomplish nothing. One method Ive found particularly effective in transferring heat from small components up to 2.5 inches is 12 gauge solid core copper household wire (available wherever house wiring is sold. aka : most hardware stores.). Strip the wire from the insulation and cut to length. You can shape it any way you like. Flatten the contact surfaces of the wire with a hammer before affixing. Ive been using thin viscosity permabond glue with a clamp on each side to maximize the contact area. The clamps stay in place until the glue completely sets up overnight. For this type of installation method, its been far more effective than any other thermal interface material Ive come across (including AA) because it can handle the stress of bent wires under a small amount of tension without failure. You could sand and polish the wire on each contact side to maximize the surface contact area, but Ive found that to be completely unnecessary.
If you search, you’ll discover that most thermal interface materials (TIM’s) are insulators of heat (to varying degrees) when applied thicker than a ten thousandth of an inch, so avoid using them in access and squeeze out as much as possible with the clamp. After you test and touch your properly attached heat sink wires, you’ll be amazed how hot they get and how much heat they can transfer. If you use pennies, go with 1981 or older vintage. The newer ones are mostly zinc.
Thanks comfy! That helps. Im mostly unaffected by PWM, but Ive noticed that a lot of people are.
@FlashPilot
So you'd make the solid wire like a spring so the other end is pushed against the pill when assembled?
I keep burning my fingers while soldering wire tips, so I very much agree that wires are great at transferring heat.
Not necessarily, but that would be a best case scenario if you weren’t able to apply clamping forces to both ends of the wire while the glue dries. It often seems to work out that way when fitting two parts with varying geometries over a small distance. I got the idea after experimenting with high performance copper laptop computer heat sinks and other remote mounted heat sinking solutions. Thats where permabond comes unto its own and seems to works well with non-ferrous metals. Solid copper wire can definitely push heat a long way if its thick enough.