I agree he meant “cannon” (simple typo) but “canon” is not just a camera company:
Sk68 died a smoking blue death (single mode with liion cell) so I pulled out the pill and reamed the hollow shell to the OD of 1/2” copper pipe(.625”). 
And made up a copper insert. 
I might go old school and direct solder an XRE since a sinkpad would raise the die out of focus.
Thanks guys 
I actually made a mistake in my calculations, there’s a thing called etendue, and due to this the formula 1/r^2 doesn’t directly apply to my light because of how it is almost collimated…
The farther I take the measurement from the light, the more lux @ 1m I get.
The correct way to calculate it is by using this tool: Candlepower & Divergence Calculator For Uniform Beam Lights or you can do a bit of trig to manually figure out the distance behind the lens
This means that I’ve always been underestimating my measurements, and I might be getting actual candlepower in the tens of millions….
I need to make new measurements at a longer distance, and measure the diameter of the spot to figure out the real number, so hopefully in the next week I can actually tell you guys the real numbers 
Yes, but there are so many “light cannon” named lights I wanted mine to show up on google search, so I named it “lightcanon flashlight” so that it is pretty much the first result 
SHHHHH let’s pretend it is what I said above
The last emitters I bought to dedome died under direct drive, I think they were U3’s?
What bin of XP-L’s are known to be more “durable”? I know the W2’s on Aliexpress are an option, but I’m not sure I want to spend that much yet
The XP-Ls are new cree production. with the really low Vf, so pretty much any bin in direct drive will die from too much current.
You need to use a current limiting driver for these, not FET or DD 
I’m really not sure what you are talking about. I have heard zero reports that XP-L’s suddenly have a lower Vf. Unless you are talking about XP-L2’s?
This is interesting. The details of this are not often discussed. I think for largish diameter lights this definitely makes a significant difference in how we would calculate the throw (cd).
I think you explained it fine; this is just some further explanation of my understanding of the subject. The inverse square law, which we use to calculate the cd, is valid for a point source of light and the distance in the formula should be from the point of light to where you measure. With flashlight beams, the light appears to come from a point behind the flashlight, not the face of the flashlight, and so this effective distance should be used in the formula, (cd)=(lux)(d^2), instead of just the distance to the flashlight.
Something bothers me about that calculator you linked, though. It implies that (with all other things the same) the size of the emitter affects the divergence distance behind the aperture. This somehow doesn’t seem right to me. I did some quick tests with my UF 1504 lens light. To get a more accurate measure of this effective divergence distance, I measured the beam size at different distances (2.5m, 4.2m, 10.2m from the lens), then simply graphed the points and drew a line through them. The point at which the line crosses zero is the effective point from which the light is emanating. I did this with two different sized emitters, the dedomed old XPG2 and the XPL HI, and got equal divergence distances behind the lens, 2.5m. This means, effectively, that the beam size is not the size of the lens at the lens. Anyway, I don’t understand all the implications, just wanted to put it out there.
Mitko and DB custom reported the W2 would do ~6.6A on direct drive, so a bit lower Vf. But I haven’t heard they are dying from direct drive.
Did you have a U3 XML2 or XPL?
They were XM-L2’s that died, but when I reported it I realized that it wasn’t just me, U3’s were dying for everyone. So I was wondering if there is a similar “cursed” bin of XP-L’s
THANK YOU !
Burned 3 xm-l2 U3’s in one C8 .
Now it’s running U4 without issues
I see. I have not heard of any cursed XPLs. The V6 will do 6A on direct drive, typically. The W2 a bit more.
Well this XP-L had a section of the die die (lol) under 6A, imagine if you drive it at 4.2v from a lipo…
My first 7135 stacking. Thorfire BD04 now with XM-L2 T6 3B
It’s not the size of the emitter that matters, it is the size of the aperture of the flashlight.
I actually didn’t know how the website algorithm worked, so I took out my trusty calculator and did some simple trig, and I did get the same value as the website.
Basically imagine a triangle, starts at one point, and at 10m has a spot that is 10cm.
That is far more divergence than a rhombus that starts at 5cm and at 10m has a 10cm spot. (which is a section of a much looonger triangle)
As you can see, the second case has a lot less divergence than the first.
The square law of light “calculate back to 1m” method works for case 1 but not case 2.
I will draw a pic later today to illustrate this for other people
Well according to that calculator the beam size and the aperture are what matters. The beam size is determined in part by the emitter size (they are roughly proportional).
I agree with what you say, but I think the calculator is not completely right, for the reason I explained above. Imagine you have your lens, but you can switch between two different sized emitters. The two emitters will make different sized beams (measured at whatever distance), and so that calculator would predict different “divergence distance behind aperture” for each emitter. I am arguing that this is not accurate; they should have the same “divergence distance behind aperture”.
My measurements showed that this distance is the same for different sized emitters, and I think this makes sense. Imagine having one emitter, then putting another emitter right next to it to effectively double its size (in one dimension). Nothing has physically changed about that first emitter, but the calculator predicts a significantly different “divergence distance behind aperture” value. I think that is not physically accurate.
The calculator is essentially using two data points to draw a line to calculate this distance, the beam size at the aperture (which it assumes is the same size as the aperture) and the beam size that you actually measure at some distance away. The way I calculated the distance is by actually measuring the beam size at 3 different distances and drawing the line. I get a different result than the calculator; the effective beam size at the aperture is not actually the width of the aperture. More measurements are needed to fully understand this.
Nice clean solder work. :+1:
Thanks RBD,
2AM project… hope it works.
With that said I did something stupid. Hooked to Bench PSU with reversed polarity. It’s from a Thorfire BD 04 and the driver is build different (wrong, backward).
Original emitter is dead so, will see shortly if driver survived.
Very pretty solder work, and not an easy thing to do. At least, its not easy for me to do
If you want to save a bit of time and effort, next time skip the center pins. Its continuous with the ground tab in the back so as long as one of those is connected it’ll all work fine.
And in most cases the base level center pads aren’t connected to the ground ring anyway so the only possible benefit is a smidge more heat sinking from the added solder mass.