I did the same mod, this result:
With fresh KP 3400's:
Before tail cap spring braiding, lumens: 1,713 - 1,625, 442 kcd measured at 4.34 meters
After lumens: 1,717 at start, 1,645 at 30 secs, throw: 448 kcd measured at 4.34 meters
It's not much but it is a bump. I had already copper braided the + spring but didn't notice any increase at the time.
Thanks for the numbers Tom.
For me, I guess it’s not worth it. 
Bit late to the game.
Received an XM-L2 on a 16mm direct copper bonded star today and have managed to dedome it and get it installed into my original TN31 (which had the “old” XM-L.) I tried the acetone method with cheap nail polish remover…didn’t seem to have done much after over an hour. So used a hot air gun and removed the dome with a needle.
Everything works fine, but I mounted the star to the reflector first, applied thermal paste, and then put the reflector with star back into the light. My hope was that the reflector would squish all the thermal paste to get a thin layer. However, I’m a little concerned that the reflector actually rests against a ledge in the light, as opposed to pushing up against the start with force… can anyone deny/confirm this?
Great work btw, amazed at how much this light can be upgraded! When I feel secure about the thermal path I’ll do the resistor mod. :bigsmile:
I see this is your first post, welcome!!
Unfortunately, you sort of downgraded the light with respect to the star. The stock star is the most impressive stock star around -- you can't buy one as good. Going from the heavy gauge, direct thermal transfer, large diameter, screwed down star to a 16mm is a major step down. Since the stock star is thicker than your 16mm replacement, maybe that's why you see or suspect some play there? I've mod'ed the XML U2 generation TN31's and always remove the old LED and reflow on a new XM-L2 U2 -- that would avoid this potential problem. I can't be sure that's what's happening, but I suspect it is. You may have to sandwich a copper disc in under the star.
Also, your procedure sounds odd - normally we mount the star to the pill top first with the thermal grease, than solder the leads, than assemble it with the reflector afterwards - you should still be able to center the emitter in the reflector.
Thanks Tom! It’s a great forum here, I just now have a reason to sign up.
I’ll see if I can remove the XM-L2 from the 16mm heatsink and reflow it on to the original one. Holding it now I see the original star is pretty beefy. I used my odd method because of course the screw holes don’t align with the 16mm star, and I worried I couldn’t get the star central enough by eye. (I’m still using the plastic centering ring though.
Thanks again for your help! 
fyi - gas is my preferred dedoming method - 100% with it, lots of XM-L2's done, some XP-G2's. My thread here: https://budgetlightforum.com/t/-/17374
Also, the resistor mods work fantastically with this TN31 driver! Bridging the resistors have seen throw measurements made as high as 490 kcd, de-domed. If you want to go milder, than adding resistors will do the trick, just compute the conversion from ohm to amps, and add resistors accordingly.
Welcome to the forum Deeksie.
Tom E is one of the best TN31 modders out there. So you are getting some great advice. Make sure that the stock base does not have kapton tape under it. Some of the first generation models did not have it removed. The bases were manufactured with it to protect against oxidation, but the manufacture failed to remove it prior to assembly.
Well…managed to reflow the XM-L2 onto the big copper star, but I think I’ve over done it with the heat as it’s pretty dim now when direct driven from an 18650. Double checked the connections but the same. Will order another xm-l2.
Oh boy, wasn't sure of your experience... I use solder paste, and actually I use a frying pan on a stove top but monitor the temp closely. I use a temp gun and watch for the solder paste to melt and flow, then maybe an additional 20 deg C at most. If I stopped as soon as I saw the flow, had some problems. However, I've been using a cheap solder paste and just ordered better stuff pre-loaded in a syringe. You have to keep it refrigerated and then it's only good for 6 months.
Also, I now check for continuity and grounds to be sure the +, - and thermal pad are still isolated after the reflow. I do the same sort of test - direct with a 18650 for just a second or so as well.
+1 on the tip about the kapton tape underneath the star -- ThruNite is notorious for QC problems on these lights. I've seen a couple kapton tapes, also examples of poor LED reflows.
A close up picture might help us help you. Too much solder can restrict current to the emitter. Everyone has their own approach. I just use rosin-core solder and a heat gun aimed by hand at the bottom of the base. Very clean and fast. You can quickly adjust the temp my how much of the hot air you let hit the base. I like to heat the base without the emitter on it to first tin the pads. I try to not get the base so hot that the rosin boils. Just trying to get the pads tinned. That way I can get the exact amount of solder on the pads that I want. Once satisfied. I set the emitter on the pad and heat. The emitter will lift for a second as the rosin boils off and then will suck down into place.
No worries, it’s all good experience for me.
I have a XM-L2, 21mm copper sinkpad, and Arctic Silver 5 sitting upstairs, just waiting for leaded solder paste to arrive. The original star isn’t in great shape now. I’m afraid if I try to reflow it again the pads will lift. Thus the sinkpad…
I did some calcs, and I don’t think the sinkpad will push the temps up more than 4-5 degrees assuming the aluminium it sits on is about 2mm thick. Hoping this will be OK if I resistor mod it to say 5A. I didn’t find any kaptop tape in my light. Also if the standard TN31 can function with a kaptop tape insulation layer and dodgy reflows hopefully the sinkpad won’t mess things up too much.
I also practiced reflowing with the old XM-L out of the light, and it worked fine, I just need to watch the temperature as you say.
It’s not dead yet. :bigsmile:
Thanks for the help!
ok let’s do some resistor math…
the stock resistors are “R100” 0.100 ohms, and “R082” 0.082 ohms.
in parallel:
so effective resistance is: 0.045 ohms
now if you add another 0.082 in parallel, the total resistance becomes: 0.029
so can i just install a single 0.03 resistor? the current through these sense resistors is very small right?
Well, You can do the math, pretty simple actually here: http://www.1728.org/resistrs.htm
But, what we've been doing now is just bridging the resistors and you get about 5.5A - 6.5A (I measured 5.5, others measured up to 6.5A) to the LED, so about 1,700 lumens with a de-domed LED, and close to or about 500 kcd. You will see much higher effective amps at the battery carrier though, maybe 9-10A, 3A-3.4A measured. Think we've been beefing up the wires, along with other minor tweaks of using electrically conductive treatments on connections and contact surfaces, etc.
yep
my question is if it’s ok to use just one resistor (of the desired value) instead of having several in parallel.
the current through this sense resistor is very small, right?
Ohhh - not sure if it's true 'current sensing' or 'current limiting', sorry...
TomE, have you found any tape on the back of XML2 TN31 stars?I have the newer version and am getting board with it now also having a TK61.Without doing any other mods., would it be safe to just jump over the resistors?-Rick
I'm no Tom E, but I urge you to check under your base. The shelf under mine had a nipple in the center and just as small dot of thermal paste at that point. So basically, the base was only making contact at the nipple and at the 2 screws. Once you are confident you have a strong thermal connection, short away. If your lucky, you will get over 6 amps current at the emitter. More likely, you'll be in the 5's.
Best wishes
I’d be happy with 5 amps, but your right.I had better pull off the star and check the thermal path.Guess I had better get some strap wrenches.
Oh you mean kapton tape? Yes - but this was a while back, on only one or two. You'd think they took care of those assembly issues more recently. I'd recommend for sure to pull the star as ImA4Wheelr said - who knows what's goin on with these lights - replace the thermal grease with AS5 or better. I've seen every possibly combo of glue on the threads of these things - some lights none at all, some lights on every threaded connection, it's a total lottery. The sticking out nipple is a new one for a TN31, but not surprising. You can work on the MCPCB area without opening the threaded connection at the LED shelf and work on it from above like vinh does, but it's more difficult.
You should be able to bridge the resistors, or remove them and solder up the connection - just be sure of the clearance for the cover. To get max amps, like 6.2A and above to the LED, replace the LED wires with 18 gauge. Think I measured 5.5A to the LED with stock wires, and 6.2A with 20 AWG. Also to get best runtimes from the batteries, wire up the springs in the carrier and clean up contact points. Heat sinking is actually not great in these lights, so you should limit runtimes on highest mode. However the stock large copper MCPCB is excellent.
Just fyi, but ThruNite is well known for assembly quality issues... I see no evidence they have improved.
Word. My TN30 battery tube has rough and jagged threads which demolish any o-ring I try to put on.