What did you mod today?

Thanks a lot JaredM.
The board was non-DTP, copper, right?

Vf is quite low….but it may be partially caused by heat.

This LED falls right in the ambiguous range where I have no idea if it ultimately throws better or worse than sliced SST-20 (and is definitely better than non-sliced SST-20 or E21A). :smiley:

Efficiency is a bit low at low currents.

Virence boards are non-DTP but have an extreme thin dielectric layer so they behave like DTP.

Virence has shifted from nanoceramic alu to more regular non-DTP copper later.
I’m not sure about that specific board.

Neither is quite as good as DTP but they are close.

Hello, here’s my humble mod I’ve made time ago. Astrolux (Mateminco) S41 with XP-G3. The flashlight came with damaged LEDs right from the seller. Domes fell off the LEDs after first use, so I’ve changed them. I wanted to make it a pocket-rocket. Thank you guys for inspiration, so many usefull informations are here.

Beamshot before and after

I had spare Astrolux lighted switch, that came blank withou LEDs and resistors on it. So I’ve tried to put that dedomed bluish XP-G3 LEDs on it.


Great work ma3oska!

:blush: Thank you djozz, really appreciate it.
Can you tell me please if that lighted tail switch with XP-G3 LEDs have more current consumption than standard lighted tail switch?

Duv is definitely positive and cct is probably a bit higher than 4000k to my eye. Angular tint shift of the bare emitter is minimal. Despite being objectively green, it is somehow pleasant and Rg appears good. I tested it briefly with an SMO reflector (technically out of an ET T25C2 but from KD) focused by hand and again, angular tint shift was really not obvious and the corona was barely greener than the hotspot, and really just at the edge. I’ll try more today.

Heat was definitely NOT an issue here. In fact, I couldn’t believe how well this thing held steady output after turn on at high amps. I ran the emitter at 2.25A for a few minutes before collecting data seen here. I also checked Vf bone cold and it was only three to five mV higher at 350mA. The Virence quad 757 board is of the ceramic/Al variety. The CZ has an isolated three pad footprint, and on the 757 board it actually reflowed into place first try no hassle. Even better, I think that it’s centered enough to work in a 106xx spot optic. I only have the wide angle carclo on hand currently, and only one emitter. If you reference the datasheet, my results line up fairly well, but again I over perform in the higher current range. Efficacy is right on at binning current, but flux about 5% high, so that suggests Tj is significantly lower than 85C.

I don’t know if it stood out or not, but the output data fit a near perfect 2nd order polynomial, so estimating output at any current should be easy. I’m going to go check my other test data and see how well they fit a parabola. It’ll be nice to just type in a current and get flux.

:+1: thanks for the info

Should be close, they both run well under normal current where heat is no issue so lm/W should be similar.

I’ve never trusted the generic Astrolux switches with high current flashlights so I’ve replaced them with Omtens. But I do like how the Astrolux lighted switches use double springs.

Nice work Ma3oska.

Thank you NeutralFan.
You mean that generic Astrolux switch has effect to high current? I thought that the only problem with high current flashlight is in springs. So if you do springs bypass or have double spring it’s enough.

Sunwayman V11r mod to LH351d 3500k High CRI:

The LH351d produces an excellent beam, better than stock, no obviously visible rings nor shadows in the spill, and the large hotspot is very useful:

The color rendering is Much better than stock, reds look Red:

Outputs
stock
0.02 to 199 lumens on cr123
0.03 to 440 lumens on 16340
0.04 to 145 lumens on Eneloop AA

mod to LH351d 3500k
0.02 to 155 lumens on cr123
0.03 to 385 lumens on 16340
0.02 to 119 lumens on Eneloop AA

fwiw, because the stock reflector has a large hole at the base, intended for the XM-L, the LH351d is an excellent replacement, and makes a much nicer beam than a 219b. The mcpcb must be replaced to accomodate 3535 LEDs, I like to use a 12mm board, here is one source

Nice job Jon!

FYI, I can’t decide if I want to emitter swap my OG RRT01, it’s one of the few lights that I haven’t modded at this point.

thank you :slight_smile:

If you do decide to mod the 2012 RRT-01, you Can use 219b IF you change to the reflector that moderator007 found. It produces an excellent beam. This post has all the parts options.

and how did you know I have a few 219b’s stashed away?! HA!

> how did you know

because you have good taste! :slight_smile:

Got an old style Thrunite T10T that I wanted to mod. Sanded the edges of a 12mm mtn DTP board to fit on the shelf, put a 3000k 95cri SST-20 on it, added a 15mm mtn fet+1 that needed a home from the parts box, guppydrive dual from a while back.

Much better than the stock cold blue led. Roughly 3.1A on rested, maybe 3-4 years old fandyfire 14500 on turbo. Cleaned the switch and everything inside and out with deoxit gold and bypassed driver spring, lubed up the o-ring. Titanium threads can be a real pain if you don't keep them lubed. Light weight and handles very well IMO. MCPCB picture is probably before the last sanding, I eventually got it flush with the shelf without too much work, it basically press fit in after taking the diameter down just a hair.

Old host lives again.

The mtn dtp board is probably twice as thick as the stock aluminum, so I got rid of the stock reflector spacer, kapton taped the board and reflector, and the reflector opening sits perfectly snug around the led base without a spacer. Useful beam with lots of flood and presses board down onto the shelf nice and tight, you can tell when it heats up quickly on turbo. Not going to be using turbo that much though. Nice EDC or bed-side light that fits perfectly in the hand.

Edit: founds some more pictures in case anyone else has this host.

I tried an LH351D at first seen with the black stock spacer around it. I didn't like it in this reflector as much as the SST-20, plus I figured out the larger 12mm pcb would fit with a little persuasion. I tried the smaller sinkpad first too, and the stock spacer, and it just wasn't right. Filing it to fit the spacer was a pain. Comparison below of the three mcpcb's and a teardown of the tailswitch, which I didn't bypass because it would be nearly impossible without melting the plastic for a minimal gain. Cleaned it up and lubed/reassembled.

That is a very nice upgrade mountainair26!

great refit!

Thanks! After more than a few years of learning from this forum and countless failed mods and destroyed drivers, my average on simple mods is on the up. I can't seem to give up on an old host.

One more old mod from a few weeks ago with another old, forgotten host:

  • Jaxman X1 flood version
  • MTN fet 20mm, old
  • sliced and diced 3v XHP50.2 from sofirn store back when they still had sub-6000k in stock. Sliced with a washer and lubed razor, cut the sides around the LES, and sanded it very carefully with 1000 grit, 1500, 2500 and finally 5000, then cleaned it very well with alcohol and electronics cleaner. Sanding may not have been nevessary, but I wanted the top to be flat, and since this was my first slice attempt, it was uneven prior to sanding. I think it was around 5700k originally, now something around 5k, but just neutral enough for my liking since I don't really like anything that's blue or cold white. It makes the cold white xhp50.2 in the sofirn SD05 look cookie-monster blue by comparison on a white wall.

I got this old version of the X1 on ebay years ago in a "for parts" sale since the previous owner destroyed the driver in some mod attempt. Ever since, I've been practicing/destroying parts in it for years. It has a brass pill instead of an integrated shelf. In it's final version that now finally actually works, I soldered the driver in more cleanly than in this pic. I had pretty much destroyed the driver cavity through tons of stupid experiments. This is the final mod for it and I'm happy with the result.

Original failed "mod" with dedomed XML2 U4 1A I did a few years back before shelving this host and nearly throwing it away a couple of times:

I think the efest 26650 just blew the bond wires direct drive, because I've dedomed this led before with no problems on an 18650. Who knows, because I can't remember the mod details or what went wrong. You can tell from the over-use of ceramique 2 that I had no idea what I was doing when I modded it before, still don't. But learning.

Anyway, I removed all the old paste, torched away all crappy soldering, cleaned it up, and put a fresh mcpcb in there soldered down after polishing the pill a little bit. I also put thin tinned copper strip on the threads of the pill to increase conductivity and heat transfer. Screwing it in was a bit tougher, but not impossible and it's still removable. I got the copper strips from a reputable Ebay seller intended for soldering/spot welding battery packs, but I've used it for this and on an S2+ pill with success. I only use a weak keeppower 26650 with it now so I don't blow up the led, but it's got something like 9-10A at the tail on turbo. Not bad for something I almost threw away several times in frustration. Lots of flood, and at least 400m of throw, uniform tint with no green or shifting.

End results below with comparison to M26C (sst-90.2 delens), KD copy of convoy L2 with XHP35.2 and convoy 2.3a driver, and 2 cell genuine L2 with sliced/diced XHP70.2 and 4a buck driver.

Left: X1 with sliced XHP50.2, Right: L2 copy with XHP35.2

From top to bottom: M26Cvn - SST-90.2 De-lens, X1 with sliced XHP50.2, KD L2 copy with XHP35.2, 2-cell L2 with sliced XHP70.2 (taller, so appears larger)

26650 light comparison, the X1 is noticeably smaller and chunkier, I like this old host now instead of wanting to spike it on the ground and throw it away

Thanks for everyone sharing their ideas for me to steal and knowledge to learn from!