Help me pick S2+ emitter / optic

I’m lost. Until today I knew what the perfect setup would be - 10507 with three XP-Ls. But I was told it doesn’t work.

OK, I need something else then.
I want to have a dragster, with silly peak output. But when I’m in need, it should provide several hours of useful output from its lone 18350. Also, I’m an efficiency junky, so the 10% output loss from dedoming totally kills the XP-L deal. I’m looking for rather floody output with some throw and no hot spot. I’d like to get 0.5 OTF lm/mA at low levels (100-200lm) and ideally 2000+ OTF lm after cold start, though I can settle at less than this; efficiency matters to me more than peak power.
This morning I thought it was possible, now I’m not really sure.
Quick research has shown several options:

  • 3*XP-G3 S5, direct replacement to what I intended
  • 4*XP-G3 S5, slightly better, killer PCB shipping prices. It’s cheaper to buy a complete Astrolux S41 than to ship from Cutter.
  • 1*XHP70 rewired to 3V, but dedoming destroys efficiency
  • 1*XHP70 with a boost driver, but is there a driver that will meet my size / efficiency goals?
  • 1*XHP50/50.2/70.2 like above. Actually assuming that driver losses are the same (not true I guess), 50.2 would be the most efficient
  • 1*XP-L W2. That may reach 0.5 OTF lm/mA, but will provide lower peak output. What optic?
  • 1*XP-L2 V6, slightly less efficient, but a bit more powerful than XP-L
  • 1*XP-L W2, how long until it’s available?

Any suggestions?

XPL2 does fit under the 10507. A triple XPL2 would be very efficient on lower modes using the single 7135 chip of a FET+1 for example. But keep in mind the efficiency of TIR optics is different than traditional reflectors. I would estimate they are 90% of the efficiency of a traditional reflector. But the TIR puts more of the light into the beam instead of the spill. So if you look at total out-the-front lumens a reflector wins, but if you look at just beam-lumens a TIR would probably win.

Now that’s news! Looks like the winner then.
I’m well aware of TIR efficiency. Carlco provides this info (though not for 10507 with XP-L2, I don’t expect it to be much different from others).
XP-L2 V6 would be .54 lm/mA minus bezel losses (which I want to improve, but won’t negate) minus electric losses when driven @ 350 mA. And would do 2500+ OTF lm @ 7A :slight_smile:
BTW, @380mA it would drop to .53 lm/mA.

Thanks a lot!

Take a look at djozz’s test of optic efficiency, if you haven’t already.

I don’t know whether it is consistent with the manufacturer’s specs.

I built a pair of S2+ lights with triple XPL2 from mtn electronics with 10507 optics and mtn DD drivers on copper spacers. They’re awesome little hand warmers… you won’t be disappointed.

to enhance TIR efficiency use optical UV hardened glue to get an AR lens on top of it

Great thread, though it’s too late to digest it now. Thanks!

:slight_smile:

I may try to go w/out lens on top of 10507…I need to research the topic though.

As to driver, I don’t like FET+1 option as the distance between 1 7135 and DD is just too large and efficiency in higher modes would be poor. H17F would be better because it has 1+7+FET. Though I don’t like 380 mA low. greg @ swiatelka.pl sells a driver that has 1+1+10 350 mA 7135s. The high modes have 700 mA w/out PWM + fill the rest from PWMed 3.5A. Maybe he could replace all but the first with 380 mA, pushing the driver to 4.5A. DD would be missed, but efficiency would be top notch, which makes it my favourite. There’s also HEDBEDES 104A with multiple PWM-free modes on high, though I like greg’s software more. Considering also LD-3, though it’s probably not efficient enough for me. And there may be H17F+ coming. I’m not in a hurry, I could wait if I knew it was worth it.

I just built a light similar to TBJ’s.

This is the build list with the parts coming from Mountain Electronics and the batteries came from Illumin. This light is a gift for my brother so I included the short tube, two 18650’s, two 18350’s, and a holster and clip from Simon. Battery selection matters as you have to have good high drain batteries for this kind of build.

(3) CREE XP-L High-Intensity V3 3C LEDs on NOCTIGON 3XP MCPCB + PRE-BRIDGED 3XP-XPL-HI-V2-3C-JP 1

10507 Carclo Lens - 3-UP Narrow Spot Optic CARCLO-10507

Convoy S2+ Flashlight Host - Red CON-S2±RED-HOST

FET + 7135 Driver - 17mm - MTN-17DDm

  • Wires + Spring Option: Complete | Wires + B…
  • FET Firmware: BLF A6
  • Turbo Timer: 60 Seconds MTN17DDm-NUV

20mm AR Coated Glass Lens ARLENS-20MM

Convoy S2+ 18350 Flashlight Tube - Red

Bob

ETA:

You also need the triple spacer/adapter but I already had one of Hoop’s custom pills for this build.

Still the carlo TIR is not AR coated
If you fill the gap with UV hardened display glue you have almost no loss on that side
And the advantage of AR coated glass on the other

Why not use TA driver 7135*1+7 + FET

And glass is far more resistant to scratching than plastic. That is why I use them on my triples and why I always get glass lenses for my eyeglasses.

Bob

OK, I redid calculations with information from that thread. It seems that a single XP-L W2 with reflector will be in the same ballpark as a XP-L2 V6 tripple@350mA. A single will be between 4% better and 1% worse. It will lose badly at high currents though.
A quad would be 1% better at 350 mA and 6% better at 7A. You need much more current to see the difference.
W2 bin would be 4% better all around.

Not sure if I follow, do you imply that with AR glass and such glue it will be as more efficient than without?

the standard TIR surface is not AR coated, so there is always like 5% light reflected back in the flashlight
a good AR lens has below 1%

Thanks for the info. I guess the best would be to coat the lens, but it won’t be cheap to replace then. And something tells me that it would be hugely expensive. :wink:

BTW, now that MTN sells quad PCBs, a quad would be an option. Does XP-L2 fit with 10621?
10621 would be even more floody than 10507, right?

There is a problem with glueing the AR glass lens to a Carclo 10507 though…
Glue will end up in the front cavities of the 10507.

Thanks for the link, very interesting (and a little disappointing).
I thought TIRs were more efficient than normal reflectors, but they’re not…

They are, just use proper leds and dont close tirs edges with the bezel.