ENEDED

It doesn't work that way though - you only see 4.35v with no load. They drop like a rock under high draw. I know this from the past, but more recently re-tested - same results, worse than say an HE2 or 30Q, etc.

Also, it won’t take much more to fry the XP-G2… it can usually go up to 5A but isn’t good for much more than that. With some power-on fluctuations and such, gotta be a bit careful not to be too close to the ceiling on those.

Interesting work there Tom, thanks for taking the time and going to all the effort.

with regards to the contact “bumps”, is there a way to mod the whole driver board into something like this from Courui d01? ? No bumps, just flat PCB cut to this or similar shape

that should relieve most of the pressure on threads and even lower the resistance by increasing contact area

If you can get the bumps off somehow (sand them down?) you can just use a washer in their place.

for instance a single large piece of a thin copper sheet like this 99.9 Pure Copper CU Metal Sheet Plate 100x100x1mm for Handicraft Aerospace for sale online | eBay or this
http://www.ebay.com/itm/1pcs-0-5-x-100-x-100MM-99-9-Pure-Copper-Cu-Metal-Sheet-Foil-/291644627765?hash=item43e75fdb35:g:AN8AAOSwEetWAkC7

Possible I suppose. I looked closer at the thread - don't think there's damage but the anodizing seems to be coming off on the first thread - think that's why it looked worse than it is. Still, it's a challenge every time to get the threading started.

Decided I'm gonna go ahead and piggyback in a 20 mm driver, and go with the U4 1C I was using, but de-domed. I added a couple more trace jumpers, and added 2 screws to secure down the LED shelf. Since the darn thing is so loose, it's a PIA to assemble the light. I'll just add thin layer of thermal grease on the overlap shelf, then the 2 screws will hold it tight all together.

Ohh - the flat edge in the body that the LED shelf rests on is exactly 5.0 mm wide Smile.

Been side tracked a bit on a L2 mod, and driver issues.

I ended up using the dedomed U4 1C in the L2, after finding a broken bond wire in a under-performing dedomed V6 1A for the L2, so I dedomed a U4 1A for the BOSS1 - took a few hours in gas but came out look'n perfect. First tested on one EFEST 26650 cell, and got almost 5A, then tried with all 4 cells before assembling the reflector/head, and all I got was a quick flash, then nothing. Looks like I popped the LED - it's dead now. I can't make out a break in the bond wires, but can't be sure - I really need a microscope, hard to tell with the 20X-30X cheap magnifiers I got.

I always though the high Vf of these new XM-L2's would protect them from going over the point of popping, but using 4 high power cells in parallel, combined with no driver was probably too much, or, the dedomed XM-L2's can't take as much amps - not sure.

I'm out of FET's right now to build a driver, and low on parts - put in orders with MtnE and Mouser. So, guess my next mod will be to add a driver and try another possible sacrificial dedomed LED.

The tube threads aren't as bad as I thought now - they are holding up, and the driver retaining ring is holding up so far, but I'm being really careful with it. I got spoiled on the new Convoy L2 threads, but the L2 also has a very thin driver retaining ring, but out of brass, not aluminum.

Have you tried focusing the xp-g2 to improve the throw?

No - I wasn't getting much power from it anyway - didn't think it was worth it. From a close range of 5-6 feet or so, the beam looked well focused, but further out, it developed a significant corona in the pattern, so I think it could have used focusing. Still, it's one of the highest kcd #'s I've measured, only beaten by a couple of Shocker mods, best being the one I built for the BLF competition last year.

I need to spend more time on hot dedoming, and go thru my batch of XP-G2's - I got one performing well at about 5A from gas dedoming.

maybe you had just bad luck with faulty emitters- both the xm-l2 U4 1A and xp-g2. Cant wait to see the next results, I believe this light has more potential than it has already shown.

I totally agree. So, figured I'd try a XPL I have here, bought from FT - a claimed XP-L V6 0D (bought 3). I hot de-domed one, came out clean, bench tested to reach 5A so I'm happy it's got full power.

First tested on one EFEST 26650 cell and got 5.15A tail (@switch), 575 kcd measured at 5m.

Then taking some risk, put in the 4 cells and got a 5.95A tail reading at the switch, and got 635 kcd (taken at 5m), 1594 meters

I'm a lot more impressed now, also knowing most likely if I took the reading at 10+ meters, it could go as high as 10% more, meaning as high as 700 kcd, and this is without an AR lens, and did no tweaking for focus, though it looks very impressive.

This is the highest throw reading I've ever taken on my meter.

Outdoors against the 280 kcd Convoy L2, it's a noticeable bump up.

It is a beast. Waiting for Chris at flashlightlens.com to get back to me. Might just go ahead and order a 88.5 or 94 mm that he stocks - the 94 could be trimmed/sanded down, or the 88.5 would probably fit ok.

At this point, I'd have no problem recommending this light for it's sheer throw potential. There are issues, like the threads which I have some concerns about, but you just have to be careful with changing cells.

Tom do you have a modded Courui? Do you think I’ll be able to see a visible difference between ~400kcd and ~650kcd? Or will the difference really only be found with measuring equipment?

Impressive numbers! Could you try measuring the throw without the lens? To see how much it lowers the output. Such readings would be closer to those with an actual AR lens

400 kcd vs 6xx kcd? Boy, no clue - more light on a target even at 100 yds, so you can do the calculations exactly how much light is thrown at what distance. The range is based on 1/4 lux (moonlight level).

I would need some range to compare and can't do that at my house. I got a modded COURUI but slightly modded, not much for throw. I got a modded K40 though, dedomed, does high 300's I think. Also got a TN31 but it's been down for a while - not sure if it works.

I'm not sure how to do the test without a lens reliably -- have to think bout that - might need some sort of spacer to keep the reflector tight. I usually get 5-6% bump on pure lumens, double that for kcd. The glass lens looks like nothing special.

400 will sorta show that there’s trees across the Mississippi River. 650 will show you what shirt a guy is wearing on the other side of that great River.

This was shot with my TK-61 right after I got it and with my first mod attempt. I couldn’t get it completely done before we went on this trip, so it was only about 430Kcd, or in that neighborhood. The range to the trees across the river is 1158 yds.

While you can sorta make out the trees, at 650Kcd you’d be able to see a man standing there fairly clearly. I don’t have such a shot from a light I later got up to 665Kcd, regretably.

Some pics of the holes drilled/tapped to secure down the LED shelf. I used brass flat head screws, what I had that fits. On that ledge the screw holes are in, I used a light covering of thermal grease in the final assembly.

This is a view of the driver (20mm) I'm planning on using, and where I'd place it. On this area of the board, I could connect up 3-4 batt- pads and wire them to the new driver's top edge, which is nice and wide. I think it's a pretty good setup for the batt- signal. For the LED+, I'd use the existing 20 AWG wire, then wire batt+ separately to the driver to power the MCU - I typically use 26 AWG wire because you don't need anything heavy for that.

Notice above the longer jumper just above the OSHPark board - I added that from the what I did originally. On the left end of that jumper, I scraped off the silk screening to expose the copper trace, so I could solder to it. Figure any improvement over traces has to be a good thing. Was thinking of doing these jumpers on the traces on the opposite side (batt+ side), but I think they would interfere with the cells when tightening the battery tube on.

Ugly, non professional look, yet functional and stable :slight_smile:
(I am using normal one face cladded PCB ground away the copper I don’t need by air tools and soldered whatever necessary to close the circuit. There is a locating Alu-bronze pin to brass tube which is embedded 20mm into the body (with a spring around the tube to hold the pin)
With XHP70 N4 1C on copper MCPCB from KD, powered by 18x 7135 HZ-3902 driver, modified 2s2p 26650, what I get:
4000+ lumens (~ 3800 lumens after 20 seconds), just approx. 15% lower than K60 by comparison.
~160+ kcd (centre point “flower shape” shows similar lux compared to TK61), throw better than K60 even the lens is un coated and chipped off at one edge.

During modification, one of my XHP70 was “dedomed” (fail) by tightening the reflector base onto the LED :_(

The XHP70 base plate surface has to be at least 2mm below the reflector flat base in order to avoid round dark center, as discussed here
However this will reduce the lumens by approx. 15%.



The PCB is held by the 2.4mm Alu-Bronze pin (welding rod) and spring which is wrap around a brass tube embedded into the battery cartridge. The negative short springs are from intl-outdoors.


XHP70 is 2+mm below the reflector surface


Compared to TK61 on the left

Nice work! If you wire bypass those springs you’ll get even more. Stack some chips or go with an FET driver. With the equivalent of 4 XM-L2 dies on board, at only 1.7A available for each one (here we go again, probably figuring this wrong) there is room for more.

My SR-90 is running a High CRI XHP-70 from a Zener modified FET at some 5600 lumens. The normal XHP-70 was doing over 6300 before I pulled it in favor of tint. Sheer power helps to close up the donut hole if you don’t have a reflector like the Lum 5-90, which my Intimidator does.

Edit: Very creative with that pcb manipulation, I’d have never thought of that! Does the large driver board have a Zener mod for the MCU? Is it built in or just modified? Looks like a very viable option. Thanks for sharing… nice fat spot light for search and rescue type work, should have pretty decent runtime with the right cells loaded up. Sweet!

Thanks. Great suggestions. I believe the amp I got is < 6A although is 18x 7135 (=6300 mA) due to high resistance.

Before increasing to higher Amp, the next thing I need to add is a better Heatsink. The brightness decrease rapidly after switching on the power, I believe is due to temperature increment. The original Alu pill is not sinking the heat fast enough.

Makes me wonder if the Illuminations Machines Lum 5-90 would fit in there and maybe give some space to raise the shelf, use copper and replace the shelf completely with a thicker copper version. Might be worth looking into. While those are only about $5 each, they make you buy like 5 at a time. A group buy here would easily manage the purchase requirements.

Also, the Lum 5-90 isn’t as robust as that original reflector, it can be trimmed with scissors pretty easily. :wink: