Focal Point of an HD Emitter?

[Otokoyama]

My empirical observations suggest that the focal point for an HD emitter is essentially in the center of the hemispherical "dome" lens, about 0.06" off the body (black part).

The Lumileds HD "Spatial Radiation Pattern" graphs indicate no light being generated beyond 90 degress off axis ("angular displacement"), but it is not clear the exact point from where the angle is being measured. I suspect this is measured from the bottom of the lens where it attaches to the base, not the actual focal point.

Measuring from the apparent focal point, there is some light being emitted beyond 90 degrees which should be captured and used for maximum efficiency.

All of my emitter-based lights so far have some final focus "slack" which I'd like to remove. Does anyone know exactly--to the thousandths of an inch--where the HD focal point is?

 

[Doug S]

It is my belief that the dome on a HD is designed to do *as little* as possible in terms of redirecting the light. Since Lumileds was designing for a lambertian light distribution and the die can reasonably be expected to already have a lambertian distribution, the only required function for the dome would be to protect the die. The LED does not emit from a single point but instead from an area that [mostly] lies in a plane. It appears reasonable to me for most purposes to treat the focal "point" as the elevation of the top of the phosphor coating on the die. I suspect [intuition here, I'm too lazy to work out the ray tracing diagrams] that at least for some reflector designs if the tightest possible beam is the objective, the apparent focal point will be a fraction of the die width *below* the surface of the phosphor.
BTW, Otokoyama, I am a great admirer of your work.

 

[Ginseng]

It's right.................................there.

No seriously, since all four dies radiate (5W) you cannot presume a point source. I think I hear you asking about the point source of the light emission and not the focal point of the HD lens. The best you can do is to do the optical analysis that allows you to back out a fuzzy virtual blob source at the actual focal point of the HD lens. Because the die area is much larger than the focal point...well, fuzz happens.

Now, if you had bare emitter dies you might be able to stack some corrective optics on top and generate a virtual point source, but I have no idea what combo of lenses would be required. I guess what I'm saying is that focus slack is there for better or worse because the HD optic is itself a suboptimal, though largely functional, optic system for controlling the emission from a non-point die. I take the dip between +/- 20 deg off axis as an indicator of this inability to control the non-pointness of the die emission.

Wilkey

 

[Otokoyama]

Thanks. I should mention this pertains to the 1W white HD. Even though the HD is of course not a point source, in my lights with adjustable emitter positioning, there's a very distinct point that yields the tightest hotspot for a given distance. Farther in and a hole opens up; farther out and the hotspot loosens up. So, empirically, it behaves like a fuzzy point source.

Right, Ginseng. I shouldn't have called the point I'm looking for a focal point since the HD lens focal point is probably the die; "virtual point source" is a good term.

Since I just made a reflector when a known focal point, I can guess that this virtual source point is roughly 0.06" above the body into the lens, which approximately corresponds to the center of the lens hemisphere.

I had somehow expected the virtual point source to be the base of the HD lens where it attaches to the body, but this is clearly not the case.

I thought that someone must know or be able to determine what exactly this virtual point source is. Maybe I should just go with 0.06" above the body.

 

[McGizmo]

Otokoyama,

I have assumed a light origin somewhere in the neighborhood of the point you mention as well as the die plane itself as Doug has suggested. I have matched these points with where the coil of an Incan resides within a reflector. Since you *know* the focal point of uyour reflector, you may be in a great position to advise us as to where the LED resides in the optimized position. I have been monitoring this thread also hoping someone would post this information but it may be that you will be in a position to enlighten the rest of us; should you be so inclined.

- Don

 

[Otokoyama]

The reflector I built isn't truly parabolic since I wanted enhanced spill, so I can't tell exactly where the virtual point source is ... my 0.06" figure may be off by up to 0.02" either way.

Hmmm, the program I wrote uses ray tracing, so I should be able to design an elliptical reflector with known focal points. It should then be very obvious when the virtual point source is at the focal point since the other focal point would be a relatively tiny spot.

Sounds like the thing to do, but it may be awhile before I get to it.

 

[Ginseng]

Oto,

Upon reflection, it seems to me that if you have the capability to ray trace, you might tackle this bull by the horns instead of the tail. By this, I mean make use of the fact that we know the HD radiation pattern is clearly not pointlike. As a curious comparison, Welch Allyn provide a spot size/diameter graph for their reflectorized elliptical LAs that looks suspiciously, amazingly similar to the HD double-hump curve. I think this is a result of the non-point nature of the filament.

Ok, so what am I saying. Only that if we already know that the emission pattern is two-humped (in 2D, donut-shaped in 3D when rotated about the optical axis) rather than the ideal single-hump point source why not assume a ring source (in 3 dimensions) or a closely spaced twin-point source (in 2 dimensions, but valid when rotated around the optical axis). You could conceivably run simulatons to see if it would be possible to design a reflector profile to bring those two humps together. My guess is you will probably be able to eek out a bit more, but without optical (lens) tricks, you will not be able to converge them with the reflector profile alone.

Wilkey

 

[Otokoyama]

Upon reflection (good pun), my guess is that trying to optimize for whatever causes one hump would end up de-optimizing for whatever causes the other.

To collimate everything, I like the McGizmo technique of harnessing the nonreflected light with appropriate refracting optics. A lens with a refracting stalk would make a nice aftermarket addition for a Mag.

 

[lambda]

Otokoyama,

Have you seen THIS article on Luxeon lens/reflector design ?

I've not had the time to study it yet, but may have some info for your project........

 

[Otokoyama]

lambda: Thanks. I read that article when you first posted it. I found the section on Compound Parabolic Concentrators interesting. I designed something that's a combination of an elliptical reflector that captures 100% then pumps it sideways into a parabolic for collimation. I'd need CNC capabilities to build it though.

By the way, I love the Micro Illuminator. Something finally displaced my Arc AAA!