Thanks, that means a larger die size than XM-L (2mm)
Measuring a mule is ALWAYS going to be much different than measuring a thoroughbred.
The reflector collimating output makes all the difference in the world in how the sphere or lightbox reads the result.
I use a light pipe as built and calibrated by manxbuggy1 and rdrfronty. They used 20 something ANSI lights to determine the multiplier, which for me is .345 x the meter reading. If I have to figure anything and shift or make adjustments, then the game is over and they’ve already turned the lights off cause that ain’t happening! 
Yes, these measurements are in a built light, I should have gotten beam shots last night but was exhausted after a very long and hot day out and about.
PS: Let me also add that holding a Couri D01 reflector over this emitter gives spectacular results with an even tighter hot spot and very smooth spill. 
VOB,did you drive led with some current source,or did you adjust pwm of DD driver for each current?
Well - i have some numbers...
my not scientific "setup" (man! - i really need a 30/20 bench power supply....)
i used several cells in 3S and 4S to simulate different input voltages (due to sag)
(the 3,3A could only be reached with full HE2)
i think my "lightbox" given readings are about 15-20% high (1000 Fenix-Lumens are displayed as ~1200)
| Volt | mA | "Lumen" |
| 10,5 | 19 | 43 |
| 11,03 | 110 | 263 |
| 11,95 | 520 | 1050 |
| 12 | 590 | 1170 |
| 12,12 | 620 | 1250 |
| 13,7 | 1700 | 2560 |
| 13,9 | 1900 | 2700 |
| 14,12 | 2230 | 2900 |
| 14,5 | 2500 | 3000 |
| 14,9 | 2700 | 3100 |
| 15,2 | 2980 | 3200 |
| 15,8 | 3300 | 2800 |
as Graph:
Volts from left to right are meassured at the cells
numbers are taken after 10 seconds (fast drop due to heat above 2 A !)
i let the LED and copper heatsink cool down to ~25° after each meassurement
i tried severall cables and even SMD resistors (holding by hand between the contacts) so get Amp-readings around 2A - but i ran out of time.
i think led4power guesstimated the sweet spot at 2.5A just perfect ;)
I have a mastech bench power supply.
Right now I have a U4 XML2 in the gas for de-doming. Tomorrow I will run this in my setup and see what I get apples to apples. I will do a test both on a copper sink and inside identical lights.
FWIW, the light I used is a Maxtoch SN50. The Luminus SST-50 had to go, of course, and I had already put an SST-90 in it. This Cree XHP-35 is tighter, whiter, and more powerful and probably a good deal more efficient.
Edit: Beamshots…
First, a comparison shot… my EDC Eagle Eye X6 Triple with V2 1A XP-L HI emitters and a polished CUTE-3 optic…
And the same scene, same settings, with the Maxtoch SN50 and XHP35, using LED4POWER’s LD-2
The above 2 shots were at 28mm, here’s the XHP-35 zoomed in to 112mm for detail…
And aimed at a little barn 610 yds away, first at 28mm then at 112mm
Works for me! 
Oh yeah, forgot to say… the X6 Triple is making ~3880 lumens while the SN50 XHP-35 is doing 2060.
I can’t believe the beam this XHP-35 is making, this isn’t exactly the best host for this but it’s still pretty dang nice, I can only imagine it being that much better in a bigger better reflector.
has anyone a word to my results?
M4D M4X, looks like you did a fairly exhaustive testing set on this and other than the lumens reading on the high side it all looks fabulous! I wish I could chart like that!
Did you get the lumens numbers with the emitter in a light? Or how did you derive those? Do you use a lightbox or sphere? Curious, is all.
I used the components Neven suggested, except for the resistor at the moon set, I didn’t have the correct one and frankly I don’t like a super low moon on a light like the SN50 anyway. So the 32 lumen moon is fine with me. I used a 270Ohm in place of the 154Ohm.
I stacked 3 (I think, 2 tall and 1 beside em laying on it’s side) of the 750KOhm resistors at the battery set-up side and engaged the 2 cell button. Then, with a 180KOhm sense resistor (I had already pulled the stock one, so I put on a new one) and the new big resistor at 2M instead of 1M, I got really close in-light reading’s to what was expected. I didn’t want to push it REALLY hard, but wanted it up there fairly close to top end. So far so good. I’ve run the light 5-10 minutes at a time and it seems just fine. When all was done with the assembly and I was happy with it, I smothered the inside of the board with Arctic Alumina Thermal Adhesive and topped it with a new 16mm Noctigon, masked side out. The driver itself is soldered into a thick copper disc (same thickness as a Noctigon), as this light requires like a 24mm driver (didn’t measure, just cut to fit) so the driver has heat sinking through the thick copper adapter it’s soldered to, which in turn is clamped into the lights heavy pill with a thick brass retaining ring.
I have no idea if all that will help anything, but it occurred to me to give it the benefit of doubt and add that copper for a bit of a sink. Whatever, it’s working and I’m very pleased with it.
The light has a 3 cell tube, so 4 18500’s fit nicely. I don’t know what it’ll do with hotter cells, but you’re showing the best output from HE-2 or HE-4 cells. Are we giving it too much Voltage with 4 big dog cells? That’s what I was afraid of, so I went with the Sanyo lap pulls. Figured they’d sag pretty much at load and then be within reasonable realms for this modification.
Ya I thought it was good too. I know what a pain it is to do that stuff without a bench supply, I had to do it like that for a long time too.
I am curious to what your set up is too. Also what are you doing to extrapolate the lumen number from you foot candle reading?
on my pictures cou see all
first one shows the bare led on noctigon srewed to a copper pc heatsink
the cardboardbox is my lightbox (turned inside outto have the white surface inside)
the lightmeter is looking to the back where the led is pointing…
so its kinda “ceilingbouncelumens“
the graph comes from google tables and is screenshotted 
Sweet! Nice thinking “outside the box”
Reading lumens from a mule configuration is always much different than reading from an assembled light. The collimation changes things, as do the losses from the actual reflector and lens. So it becomes somewhat confusing trying to relate the different methodology.
I’d like to see the XHP-35 in a BIG light like the SR-90 Intimidator, but don’t want to change mine again. lol Maybe a TK61? The Nitecore TM36 also has a massive solidly made aluminum reflector that would do well here I think, getting the 12V set-up would be the issue, boost driver? Extension tubes? At any rate, I’m looking forward to seeing where this one goes…
Edit: Aw man! Y’all are making me think things I shouldn’t be thinking! I actually DO have a second tube for my Intimidator, could 2 sets of 6 18650’s (3P2S x 2) be made to power this 12V XHP-35? Why yes! Yes it could! :bigsmile: And the massive reflector should give excellent throw with this emitter, right? So if I can find someone to make me an adapter collar to allow the two tubes to be mated….
Edit II: Duh! There are storage caps for the second tube, the solid end could be cut off and thus, a threaded collar born. I really really like the SR90 as it is, with it’s 96CRI XHP-70 and Illumination Machines 5-90 reflector! Why me? Lord help me…
Not sure how that’s applicable Fritz. A plastic light is a bad idea for high output, and we need 4 cells in series to get over the 12+A draw of the XHP-35. What am I missing?
Than how about this one with 14500’s in series max say 1.5 amps. http://www.illumn.com/flashlights/other-brands/starry-light-sa-22-880-lumens-4-x-aa.html For those of use with pockets.
That one might just work, but I’m not sure what it would gain. If it’s doing 800+ lumens to start with and you swap in a XHP-35 at or under 1.5A then you would only gain, in theory, a few hundred lumens. Almost not enough to be able to see it. And it wouldn’t be easy, most likely.
What about the Convoy L4? You can get an extension tube for it, run 4 18350’s and the reflector is nice enough that it should put out a surprising beam. The light has a tail clicky for lock-out, a side e-switch for modes, so while it wouldn’t be exactly easy it will take Neven’s LD-2 (I know this because I put an LD-1 in the L4)
So, while the XHP-35 is showing an impressive potential, it’s rapidly becoming apparent that it will take a rare set of conditions in a host to properly utilize it. The Fenix TK series has stackable tubes with carriers, so the carrier could be modified to 4S and a second tube also modified would extend run time. With that huge reflector (even though it’s plastic) this emitter should really rock!
i hope i can finish my build the next days….
i don’t want to tell to much early because if i fail…
I bet the Starry Light SA-22, 4 X AA with HD XHP-35 top bin LED might push out 2000+ lumens on a 1.5 amp driver.
This is a P series, not L, so it is aluminum except for the battery holder. It takes 4 x AA cells, so 4 x 14500 would fit.
It comes with an XR-E, probably why it is offered for this price. So it is no harder putting an XHP-35 and 14500s in it than to put an XM-L2 or XP-L and AAs. The mechanical driver should work as is, or one could change the resistor values. It is about half as compact as 4 x 18350, but 14500 is a more common cell size.