Uniquefire UF-T20 mod: 3.05A xpg2-3C dedomed on Sinkpad

So, I have a UF-838 that I was going to put a de-domed XP-G2 in with a 3.04A Qlite / NLITE driver. But it has a hollow pill and I haven’t had the motivation to build a heat sink, and it has a rather long focal distance so it’d probably lose a lot more lumens than necessary when focused. My estimate is that it’d max out at about 60 kcd, based on holding the optics in front of another de-domed XP-G2 I have and measuring it.

I’m thinking it might be a better idea to just buy a UF-T20 to mod instead, and keep the UF-838 as a flooder. In wide mode, its lens is only like 1mm away from the emitter, so it has an extremely wide beam. Great as a “flash” for photography, but not very useful for anything else.

The goal here is to make a pocketable aspheric thrower for entertainment purposes. I don’t need 200+ kcd of throw, but I wouldn’t mind getting as much as possible out of something the size of a C8 or smaller.

Any thoughts on the best approach here? I currently have the UF-838 and de-domed XP-G2 on a 20mm star and a replacement 17mm driver and a dummy plate to mount the driver in a 20mm hole if necessary. Just not sure if I should grab a UF-T20 too, or just find and shape some copper to jam into the hollow pill and use the parts I already have.

Hi

I’ve got a uf-t20 with dedomed xp-g2. It has additional copper in pill and spring mods.I just put in a fet driver by adding a fet to a 105c like comfy chair describes in one of his posts (but not same fet, but similar specs). Now I’m thinking I’d like to see what it is like with an XML dedomed. But I’m concerned with this driver it could blow the led with a inr 20c. Could it blow the led or am I worrying for nothing?

Thanks.

The XM-L2 has a lower voltage at a given voltage so it will draw more amps, if all is done well, as it seems, you could get 7A or so. I assume a direct thermal path copper board, clamped or soldered well onto the pill? Then it should work alright. The light will get hot, but it will handle the heat, at least for the first few minutes.

I would put in an xm-l2, but only have xml on hand. Yes sinkpad, but attached with fujik. So even though the driver is capable of 10amps+ (what I have measured with basically the same driver on an srk), and the battery is supposed to be able to do 15amps, something stops it going to these levels? Sorry if that’s a dumb question. I’ve tried to read up and understand, but not quite there yet.

The XM-L has a slightly lower voltage than the XM-L2 but that should not matter much. The current is limited because a)the XM-L2's voltage at 8A is already 4.2V, add up the various small resistances in the rest of the system and you will only end up with lower current that that, b)even the best high current battery will not be able to maintain its 4.2V voltage at higher currents (see the various HKJ battery reviews). So no worries, it will be even a challenge to get it over 6A.

Btw, try to make the fujik layer as thin as you can get it, getting as much as possible metal-metal contact.

This is something you made? Do you have a thread about it? Have you tested the actual ouput results using this collar vs a Wavien?

What’s stopping it is the Vf, while in a multi emitter/multi-cell setup the FET drivers can do up to the maximum rating of the FET / cell combo however running that high the Vf is the limiting factor in a single cell setup. Most of us see sound 5-6.5A (out of the best 4.2v INR cells like 25R’s and VTC5’s) to a single XM-L2 and about 5-5.5A to an XP-G2. You will need to use a buck driver with higher input voltage to get above that, it’s just not possible with a linear setup cause the cell can’t support the increasing Vf.

From 4.2v in (using 4 Parallel 25R’s in an SRK) the highest I’ve been able to obtain is 21A to 3p XM-L2’s (which averages out to nearly 7A per emitter plus a little burn off). I do not believe anyone has been able to get a linear FET driver to do above ~7A per 3v emitter tho there are several buck drivers that can run a 3v emitter at >8A. I have a few 6v MT-G2 lights that run the same FET driver at 11.3A and also have a series light (3s XP-g2’s and 3s cells) that runs the same FET driver at 15A.

You could switch to 4.35v cells but the real world gains would be probably <.5A more. You could also switch to 26650’s or 32650’s but the real world gains would be even less.

So how did you achieve 1/10 mm accuracy? Did you use spacers?

I used pieces of tape as spacers.

My Uf-T20 measured at 170Kcd .

vestureofblood, I did not finish the X7 project, maybe one day :slight_smile: . The collar was custom cut from aluminium at a CNC machine shop.

What about the angle of the collar? How is that done, is it special for the led, or constant but based on the width of the collar or based on the necronomicon and the guy from army of darkness?

“Army of Darkness guy” LOL Bruce Campbell

It’s time to revive this thread again :bigsmile:
I’ve modded mine a while ago with a de-domed xml2, but not on copper.
I just made a couple of tailcap measurements; with a trustfire flames battery I got ~2.2A. Then I tried a keeppower imr 2500, and got 3.5A.
I have some kind of luxmeter on my multimeter, that can use for rough comparisons. With the trustfire I got ~550, but with the keeppower I got ~250.
I had to check if I accidentally had changed to low, but no. Turns out it can’t cope with the heat.
So I’m thinking I’ll have to reflow the dedomed xml2 to one of the copper stars I bought from banggood a while ago and solder the star directly to the pill.
Do I have to order solder paste or can I use regular solder as well for this?

Might as well do the 100th post of this thread myself :-)

If you do yourself a favor, get some solder paste, this stuff is cheap and worked well for many people. I'm surprised that the XM-L2 does not like 3A, it should be alright, also on a common aluminium board. Was the board clamped down to the pill somehow? (screwed, or with a thin layer of thermal adhesive)

Ok, it’s worth waiting for that stuff to arrive from China then…
I used heatsink plaster, some stars-922 probably ordered from ft. It was held in place with alligator clips during the curing time. The current might have been over 3.5, maybe 3.9A even.
It’s the original driver, high/low/strobe, didn’t know the numbers could get that high with spring bypasses.

Although paste is easier he could do it with rosin core solder and 5$ heat gun. Flux it a bit and just heat pill a lot and when solder starts to melt put mcpcb with led(press it) and finished.

Is that method invented by you Djozz? I think you first started with mcpcb to brass soldering.

FWIW, if anyone is wanting to do the thing where you project an image through a loupe, it actually works better with a regular XM-L2 than it does with a de-domed XP-G2. Throw isn’t as important for that as lumens; you simply have to get as many lumens as possible into the loupe.

I got a UF-T20 a while back to throw-mod it, but I ended up leaving it stock to use as a projection light… and got a Jax Z1 for throw instead.

I did make that up and posted about it in this thread, but others mentioned that they had been thinking about that too.

edit: I think the most significant point of this thread is that it was the first time that the implications of match's then very recent tests of Cree leds on Sinkpads were explored in a flashlight. From his graphs it was crystal clear that XP-G2's (and XM-L2's) on Sinkpads could be overdriven to at least twice the specifications, but people were a bit reluctant to read/trust a bench test (match did see it immediately btw). And back in 2013 the common believe (it still is actually) was that driving leds hard (3A through a XP-G2 was unheard of until match's tests) required massive chunks of copper everywhere, and in this build I realised (as well as match who already mentioned in the test thread that it was no fun anymore with these Sinkpads, too easy, and that he had build his last copper pill, and then he retired :sad: ) that just the very near surrounding of the emitter (the DTP-path under the led) required careful attention to the heatpath, further away any limited heatconducting measure will be fine (the T20 pill screws into a rather thin aluminium tube, but with that socalled 'bottleneck' present, in 10 minutes (surprising to me) the throw went just down from 108 to 102 kcd which is totally insignificant.

+1 for what djozz said above. As long as one uses DTP board the LED should be fine, and there is no need to have copper everywhere in your flashlights unless it is for other reasons.

I have modded a Convoy C8 for one of my friends before by driving a dedomed XM-L2 at 4.2A reflowed on a 20mm Noctigon. On high mode it was measured 90kcd at startup and 89kcd at 1 minute, very insignificant output drop and it was still the stock aluminium pill inside, didn’t add any extra copper plates in the aluminium pill as some sort of “heat buffer” either.

But I know some people still believe that using lotsa coppers does make a difference, or maybe it is just for the peace of their mind. Well, each to his own I guess. :slight_smile:

That was you?

I don’t think I had even done my first mod at the time.

Thanks for paving the way forward! :slight_smile: