I use a hot air reflow gun slowly raising the temperature, if it 1°/sec I don’t know 
Maybe I am not careful enough, but I don’t have the same issue with other emitters.
Thats simple, if it takes near 200 sec it is ok, if less than two minutes this is too fast.
Not only about samsung, we heard many issues with xhp 50.2 for same reason.
Is it proven that the XHP50.2 issues are caused by not following the correct reflow timings? I did not know that. I even have not seen any evidence of not following the correct reflow timing having an effect on any led performance. I’m not saying that there is no effect, but at least I have never seen it demonstrated.
I have a XHP50.2 dying in a flashlight with a Kaidomain 3A boost driver, and I have a XHP50.2 living a happy life at 5+ amps in a driver-less 2x18650 flashlight. Both leds were reflowed by me at the correct temperature but nowhere near the correct timing. Instead of a reflow effect, my feeling is that that KD driver causes some sort of peaks that especially the XHP50.2 is sensitive to, while in direct drive there is no such sudden load. But it is just a feeling based on just two leds, I have no hard evidence.
I cant make any long-time conclutions, I just dont have enought equipment for tests. I would prefer to ask maukka and Lexel.
Received the uni-t ut210e clampmeter couple days ago.
Measured the tailcap current again today. Used a 22 gauge silicon wire, it was probably a little bit too long though.
Measured 7 amps, that makes more sense. Still a little bit on the lower side but for that I blame the cell I used and the longer wire.
I use 12Ga Turnigy wire for my loop with the Uni-T clamp meter. I think it’s 6” long, soldered gold plated banana clips on each end.
I use this, before use everytime I sand the contact surfaces:
Glutton for punishment! 
A little late to the party but I had a threesome and put my D’s all over the place:
Samsung LH351D’s of course, 4000K 90CRI.
Jaxman E2L (left)
FET+N+1 Bistro HD
~3300lm at startup
Thorfire TK15 (right)
SMO reflector gives the 351D a little bit of a hotspot and no more egg yolk from XPL2. Still some slight rings like any SMO but the tint is even.
Side note: Thorfire must be using some of the highest melting point solder on the planet…
Ultratac K18 Mini (top)
Pulled the XPG2 out of this 10180 light
Makes ~12lm on low and ~125lm on high
It’s no CRX light but I really like this light now. It has TWO ways I can lock it out, mechanical and eswitch and a really nice floody beam. I’m going to remove the Manker LAD from my keychain now and see how this holds up. I’ll trade the lower output for not having my pants randomly lighting up.
Contactcr’s wife…. ” He was so happy to see me when I got home from work he positively GLOWED!” 
To check out the new 4040+5050 Q8 ledboard that led4power has made and kindly sent to me without charge (including 4 Luxeon V leds :+1: ), I modded my last unmodded Q8 with Luxeon V leds.
The dimensions of the l4p Q8 board are exactly the same of the stock board so it was no surprise that it fitted without problem. Only difference is that with the double footprint there is no space for the two small holes on the side, so the board can not be screwed down by the two M2.5 screws. Instead the board should be fixed like I did here, sandwiched between reflector and shelf (as shown and discussed before in the interest list thread for this board).
It was quite a pain to ream all 4 plastic centerpieces to fit the Luxeon V leds, but I got it done and the leds are centered pretty good.
I made an effort to reduce the already small resistances of the Q8 further, most significantly I bypassed the tail springs and cleaned all electrical contact surfaces. I replaced the stock ledwires with slightly longer ones (still 18 AWG).
The beam that I get is very nice and smooth without obvious tint shifts, I think even a bit nicer than the stock beam, but I have no stock Q8 left to directly compare them. (If you are really nitpicking, on a white wall the corona is slightly yellow and around the corona is a faint blueish ring. This however is very remote from anything you will notice in actual use)
The highest output with the stock XP-L leds that I measured was 6200 (djozz-) lumen. But that was not with fully bypassed springs but with phosfor bronze springs, I estimate 6500 lumen with bypassed springs. This Luxeon V modded Q8 does 7500 lumen (30 seconds, 30Q batteries). At start-up it is 8000 lumen and it takes minutes to get under 7000 lumen.
7500 lumen, great beam profile, nice tint: this mod is a succes, and the l4p Q8 board works as planned.
Nice mod djozz and a nice improvement in output to boot. :+1:
Some brass goodness today. Another upgrade to the CRX 18350 brass triple Nichia.
I installed a new brass & 14mm glass topped switch cover with green tritium vial.
Previously the light had a rubber boot.
Length - 72 mm
Width - 22 mm
Weight - 104g (with cell)
Max around – 2100lm OTF
Aluminium jacket.
You are to much CRX. 
I wasn’t happy with the rubber boot as it was just a little bit harder to access the tail switch since I put the magnets in.
Much better now, I think that’s about all I can do to this light. Next… :laughing:
Here’s mine, far beyond CRX’s mod
I just can do a LED swap on Novatac 120T :partying_face:
Cool, looks brighter. Did you have to file an MCPCB into a cross shape or use another method?
I modded a floodlight by swapping out the original cool white LED with the (supposedly) 90 CRI, 5000k CCT I bought from alibaba for use as an improvised desk lamp.
Modding in action. The original LED (bottom right) is 20w where the new one (wired) is 30w. I’m actually under-driving the one I used for replacement, but that’s okay. 20w is plenty bright for an improvised desk lamp.
Some cactus shot under the modded light. Does this looks like 90 CRI to you?
The LED used for modding. I had 4 of them in my possession. One used in this project, three to go.
I’m using normal 16-17mm MCPCB and solder short wires to the driver contact.