What did you mod today?

Thanks for testing that!

So the red will run at 350ma below 2v, so thats only about 0.7w so that’s in my favor.

However I’ve heard the red XP-E is more affected by heat, it won’t be as efficient as the XP-G (so more heat?), and there’s leas cooling in a sealed pill than on a workbench.

So with all of that, do you think I’m still good with 350ma,or should I dial it back to like 200ma?

My pleasure. Yeah, I guess 0.2A would be safer with the unknown resistance of the non-DTP. Is it really not touching anything in the pill?

Well, it’s a 12mm mcpcb that’s been cut down to a narrow strip, still 12mm long but barely wider than the emitter itself. It has some small contact with the side of a pcb and the lip of a brass retaining ring. It’s packed in a small space, but it’s not actually pressed against anything.

Oh, it sounds like there is a lot less surface area than the MCPCB I tested with. I think it should get considerably hotter in this case, so maybe try .2A and see how it goes.

Try some heavy wire to the pads if you can. Heat will go through there too.

Unfortunately no room for that, just some 26awg. Maybe I can pot the top of the board a bit, but 200ma is plenty, just wanted to see how far I could push it

Here we go again…
BLF X5

You beat me to it :rage:

I guess I’ll have to go bigger….

Had a really frustrating morning trying to make a 22mm FET+7135 driver with zener mod, using the components from a BLF-A6 driver, a 22mm wight board from Oshpark, a zener diode and a 220 Ohm resistor. I will not go into details but in the process of troubleshooting I must have soldered and desoldered (by hand) at least 30 almost invisible parts to the board. And it never worked. And it was part of a nice mod that I can not do now :frowning:

To cheer myself up I skipped the project and went on to another mod that to my surprise became a total succes. The host is an ultra lightweight 4.5 dollar AA zoomie: Ultrafire AT-007, one of my favorite cheapies because of the compact build but still good lens and optical path.

*The led is a dedomed XP-G2 S3 3D on a Noctigon. I sanded the Noctigon to 1mm thickness to get the focus right, it is clamped with the stock plastic thingy (that I made black with a marker) over the hollow pill (I love hollow pills :partying_face: ), bit of Arctic Silver on the edge of the board for better contact to the rim of the pill.

*The driver is a 15mm diameter Bistro driver, I transferred the components of a Banggood Bistro-driver to a 15mm Oshpark FET+1 board (I believe a design by Warhawk), and it works flawlessly. After the failure this morning that came as a total surprise.

*The tail spring was bypassed, the pressed-in switch-tab received an extra solder blob on the side for extra tight fit and improved electrical contact with the body.

And it works, and does not even get incredibly hot too fast, even on max setting.

Numbers on highest setting and purple Efest 14500:
–3.75 A current and quite constant for the first minute. This is measured with a clamp meter around a thick copper loop, the light without tail.
–450lm flood at start, 422lm at 30 seconds, the high setting appears pretty stable, at least for a useful minute or so . So just 7% drop, did I mention that like hollow pills? :wink:
–33kcd throw after 30 seconds, measured at 7 meter. Nice for a just 18mm diameter lens.
-and a good tint, this led dedomed gives the best dedomed tint around IMO.

So although meant as a freak light with too much power in a flimsy host, it just works and is a great EDC, and anything with bistro is pleasant to use :slight_smile: . And if throw/gram is a quality for a flashlight, this light scores 0.89 kcd/g (without battery, with battery it is 0.58 kcd/g). Beat that!

Edit: just checked and unexpected by me my Brinyte B158 with dedomed S4 2B has better kcd/g, with battery it is 1.07kcd/g LOL

Pop-pop-pop-pop-pop-pop-pop

That’s what my 7x SRK did a couple of minutes ago.

Did some crash testing with this light. Mounted a FET+1 on the empty stock PCB, springs have been brigded earlier, the traces on the tail PCB have been strengthened. There were 6 fake XM-L on the MCPCB and 1 spare Cree XM-L2, since one of the fakes died earlier.

Measured 18 Amps for more than 4000 lumens with old Keeppower 2900 protected, and then I tried my new 30Q. Was able to measure 26 Amps but this might have been a little to much … :smiley:

After less than 10 seconds all LEDs died. Actually I am not surprised, I expected something like this. New Cree LEDs are on the way.

Wow, sounds dramatic. Was there adequate thermal path and do you think that was the cause of the failures?

No good thermal path yet - but I don’t believe the cause was bad heat dissipation, since the LEDs died after a few seconds already. I haven’t opened the head yet but as far as I can see the bond wires are blown.

Interesting. In either case the MCPCB would get pretty darn hot. Pumping 70W thermal into a 10g Al MCPCB would cause the temperature to rise ~8 degrees C/sec.

I’ve been reading through this thread from the beginning (skipping every post without pictures :partying_face: ) I missed this one in April:

Dale, is that a Damascus light in the background?

again not really a mod but a combination of parts
a elcheapo ebay mag base and a ultrafire light from gb


usefull I hope

Nice flashlight, same good as our one new released flashlight TK4A Amazon.com, :blush: :blush: :blush:

New LEDs arrived, reflowed 7 genuine XM-L2. Very bright, none of the LEDs popped up to now, but it’s pulling only 24 Amperes out of brand new Samsung 30Q button top. FET+1, thick/double wires, all springs bypassed - shouldn’t it be more Amps? I wonder if the cells are genuine.

???that sounds like a inappropriate plug of your new flashlight??? That light has nothing in common with that cheap AA zoomie in the picture, it is even not a zoomie.

Ahh, are you using the stock cheap MCPCB? Yikes! Living dangerously with good XM-L2's. There's some significant points of loss's in these cheap SRK clones. Many we can fix, but I suspect some not so easy, I've been finding. I beef up the traces, or reduce/eliminate them on the tail PCB, then there's the bridge from the batt+ contact to the LED side of the driver -- not sure that's good. In single cell lights, you can run heavy gauge wire from the driver spring bypass direct thru the driver board, but not so easy in a SRK. Then there's that stock MCPCB - oh boy... I'm also not confident with the the connection of the tail PCB to the housing - actual conductive contact could be suspect, cheap screws - no brass or copper there, maybe limited contact surface.

Could be one of these things, or a combo of small loss's over them all. Doubt it's the cells.

Are there better MCPCBs for sale anywhere?

Yes, I’m using the stock MCPCB, but it’s not that bad, rather thick. Even no copper plate below (too careless, I know), don’t have any yet, but the housing gets very hot quickly, so heat dissipation should not be that bad either. I did most of the things you proposed, even managed to lead batt+ through the driver board with a very fat cable (1 more ampere). Haven’t done this with batt- yet, but here are more stock lead throughs. What’s still missing is a better connection of tail PCB to the housing, should have removed the anodization there. Will do it when I have to repair the spring bypasses one day, I don’t want to open it too often since one of the screw threads in the housing is already damaged.

I’m using the cheap XM-L2 U2 3D from fasttech, the warm tint is very nice, althought I’m not sure if all LEDs really have the same bin. Replaced the stock lens with a coated one. I’m using my own firmware with optional stepless ramping.

Anyway, I fell in love with this light, with 30Q it delivers about 6500 lumens at start, with older Keeppower protected still 5500 lumens. Not bad for a 20$ light and 30$ modding parts.

Btw., thanks for your idea of doing a wire loop at the tail PCB in order to measure current with a clamp meter easily.