I’ll emphasize one result that explains why the XPG3 is a poor thrower. The data in the last three rows in my table shows that 20% or more of the total light output is not coming from the die itself but from the light that has escaped the sides. The actual die luminance was unaffected evidenced by the X6 luminance being unchanged, but the total light output dropped significantly when the side-escaping light was blocked. You can’t expect to have high die luminance when 20+% of the output escapes to the sides.
I think we’re pretty much talking about the same thing…
Since it emits light to the sides that’s why they used a blob of phosphor rather than a flat surface, so that it goes over the edges and then produces the white light.
The first two images I posted have the die IN that dome, it is not simply the phosphor from on top of the die.
Here is a better image to explain:
You’re right though, dome shaving probably wouldn’t touch this phosphor dome.
Kd/lm is fixed parameter for optical system&led combination.
If you have enough lumens but not happy with candelas - just change optical system.
Fresneles are good to focus big cobs like cree cxa tight, why not try them with side light distribution leds?
Fresnels are only good for very large diameters where an aspheric lens becomes too heavy or expensive.
Other than that, they are much worse than a single aspheric.
And no, the XP-G3 still sucks for any high intensity application regardless of the type of optic you use.
I agree some sort of TIR could utilize all the light from a XPG3 and properly mix the colors to solve the tint shift issue, but it won’t solve the low luminance issue.
Yes there are ways to collect all of the light from an LED including that from the sides.
A regular flashlight reflector also uses 100% of the flux, if you count the spill.
The point is that the light emitted at the sides is not useful for increasing the surface intensity, which is what contributes to throw.
Throw is pretty much what this entire topic is about.
Here’s the intensity graphs of the G3 ad G2 overlaid:
G3 definitely emits more light to the sides, even though the difference is very reduced due to the dome being on when the tests are done.
G3 is red, in case that wasn’t obvious enough already
Great ‘detective work’ here I would have never considered that there could be such a significant amount of light from the side of the die. If that could be captured and sent downstream, the lumen loss could be mitigated but the tint issues would remain.
fwiw, some people are reporting tint shift across the beam, on their new Zebralight SC5C MK II H CRI, with Cree XP-L2
your post helped me understand that it is not an isolated incident:
[QUOTE=jruser;5106891]I got mine today from the new batch. Beam pattern is terrible. yellow hole in the middle of the hotspot, surrounded by decent tint hotspot, surrounded by yellow ring, surrounded by decent tint spill.
What causes this? reflector? LED? centering is fine.[/QUOTE]
I’ve been puzzling over the cd/lm differences between the G3 and G2 in the Carclo performance data. Thanks for solving the mystery. Now I have to perform a root canal on my light bar and replace 45 XP-G3s with XP-G2s.
Oh yes! I finished my build in May. Documented here: DIY Offroad LED Light Bar It will be more like 45 root canals.
I’m a noob and didn’t understand the factors that affect throw, so I was solely focused (no pun intended) on lumens per watt and the G3 looked best on paper. Now my beam pattern is broader than I wanted, so I’ve been searching for ways to improve it and found what a tremendous resource this board is.
I’ve ordered some XP-G2 S4 E3s to compare to the G3s I have.
I’m glad the measurements from me and others helped you.
At the same current, the XPG2 will have more throw than the XPG3, but only by a factor of 1.3x or so (in beam intensity). If you want a more significant increase you could dedome the XPG2 emitters. This would increase the beam intensity by a factor of ~1.9 over the stock XPG2. It would be more work to dedome them all, but it would be a significant bump in throw. The beam size should narrow by a factor of ~1.5 due to dedoming.
That carclo data gives you the cd/lm, which is not exactly the relevant quantity here because the XPG3 and XPG2 output different lumens at the same current, although they aren’t that different. My estimate could be low; you might see a 1.5-1.6x increase in beam intensity at the same current.
A 1.5x increase in beam intensity won’t seem like a lot. It is a tough game we flashlight enthusiasts play: due to the ~logarithmic sensitivity of our eyes we have to go many times brighter to get the “wow factor”.
I would have never considered that there could be such a significant amount of light from the side of the die. If that could be captured and sent downstream, the lumen loss could be mitigated but the tint issues would remain.