How to specify/order aspheric lenses

[SKU427]

I want to order some aspherical lenses to use with focusing the light output from a single LED. I am not familiar with optics and I would like to know how to calculate the needed specifications so I can order from an optics supplier.

Is there some sort of formula that I can use?

 

[MacGyverS2000]

Call Edmunds... they can help you with specifying anything you need.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[lumenharold]

Dan is correct that Edmunds will probably be able to help you spec your lens, but if you still want to understand the process you might just surf a little. There is a lot of information available regarding moldeling LEDs and associated optics. Depending on what you are trying to do it more difficult than it might seem.

This is a link to a paper on secondary optics design for LEDs and includes lenses (it comes up short of aspheres though):
C:\Documents and Settings\All Users\Documents\Reference Docs\LEDs & Lightpipes\Secondary Optics for SuperFlux LEDs.pdf

This is a link to anthother site that does cover aspheres although for use with Laser Diodes. It's not a big leap:

http://www.phototechnica.co.jp/pdf/lightpath/asphericlens.pdf

If you care to post the LED you are planning to use and what you are trying to do with it perhaps you will get more specific recommendations.

Harold
SW2009 SP4.0 OPW2009 SP2 Win XP Pro 2002 SP3
Dell 690, Xeon 5160 @3.00GHz, 3.25GB RAM
nVidia Quadro FX4600

http://www.lumenflow.com

 

[lumenharold]

That first link won't do anyone but me much good as it is on my hard drive. Dang....
I'll try again:

http://www.philipslumileds.com/pdfs/AB20-5.PDF

Harold
SW2009 SP4.0 OPW2009 SP2 Win XP Pro 2002 SP3
Dell 690, Xeon 5160 @3.00GHz, 3.25GB RAM
nVidia Quadro FX4600

http://www.lumenflow.com

 

[SKU427]

Thank you all for answering. I have contacted Edmunds and skimmed through the literature provided above. I find that perhaps an aspherical lens is not what I am looking for.
What I want to do is focus the light from a Cree XP-E LED

http://www.cree.com/products/pdf/XLampXP-E.pdf

Through a 12.7mm lens (11mm opening) mounted 2mm from the base of the LED. I want to produce approximately an 8 degree beam that does not project the square shape of the die. Something similar to the beam in the following picture.

http://www.expandku.com/smooth.jpg

I am looking for a thin assembly as opposed to a thicker reflector system. Perhaps a Fresnel lens.

Any help would be deeply appreciated.

 

[IRstuff]

So, you actually need an anamorphic lens. Note, by the way, the typical LED lens does NOT project anything to do with the die. The beam profile is a consequence of the structure of the optical cavity, and is a "virtual" source, i.e., there is no physical structure corresponding to the angular spread of the beam.

If you don't care about losing some lighting efficiency, there's

http://216.154.222.249/files/images/downloads/LightShapingDiffusers.pdf

which is a Fresnel lens specifically tailored to shape beams.

Otherwise, see:

http://led-spot.com/index.shtml?gclid=CIKG7dvViJ0CFSWlagodFgjNaQ

and others


TTFN

FAQ731-376

 

[SKU427]

Thanks for replying IRstuff. But loss of efficiency would be a definite no-no. And the beamshot picture above is taken from a LISA XP lens designed for the XP-E. But I choose not to use it because a single lens would take about 5mm off the overall length.

 

[DanStro]

By "...loss of efficiency would be a definite no-no." I hope you don't expect to get ALL of the output of the LED die to your target/screen. In my experience since the light is being emitted from the die into a very large solid angle it is difficult to get a decent amount of it exactly where I want it to be.

And the more efficient optics I believe, correct me if I am wrong, have to bonded directly to the die.

 

[lumenharold]

If you are try to maximize the out-coupling of the optic to capture the most light from the LED then the input surface will need a very steep curve. The 2 theta 1/2 of that LED is about 120 degrees and the theta 90% is about 160 degrees. In other words, considering that you want the lens nearly touching the LED dome the lens curvature will need to be small enough that the incident rays don't just bounce off at the outside edge of the aperture. That is not going to be a standard lens. Are you making a bunch or one? As far as getting close to the emitter, the Cree has a dome and unless you want to buy a bare chip they all do.

There are a lot of low profile moderately efficient solutions and some larger very efficient solutions. I believe a search for RXI would turn up a decent optic. If you are interested in considering a different LED source, these people seem to have a line on a very efficient design (though I think the 94% they claim in the white paper comes with a disclaimer). Bigger than the Cree though.

http://www.innovationsinoptics.com/products/lumibright.html

Harold
SW2009 SP4.0 OPW2009 SP2 Win XP Pro 2002 SP3
Dell 690, Xeon 5160 @3.00GHz, 3.25GB RAM
nVidia Quadro FX4600

http://www.lumenflow.com

 

[SKU427]

I'm making a bunch.

The link you sent me looks interesting, but it's way too big.

 

[lumenharold]

OK, you want to capture as much light as possible, you have an 11mm clear aperture, a 12.4mm mounting diameter, you want to use the Cree and mount the optic within 2mm of its PCB. And you plan to make a bunch. A CPC is too long, an RXI is too big around, and the diffusors will lose too much light. Have I got all that right so far?

How thick is too thick? What about glass vs. plastic? When you say "a bunch" would that justify custom tooling? Is cost a factor (a coated lens will cost more than a plastic optic that operates on a TIR principle)? Is lighting a current area of expertise within your business (is there anyone there with the software to model this for you)?

Harold
SW2009 SP4.0 OPW2009 SP2 Win XP Pro 2002 SP3
Dell 690, Xeon 5160 @3.00GHz, 3.25GB RAM
nVidia Quadro FX4600

http://www.lumenflow.com

 

[SKU427]

12.7mm mounting diameter
a lens mounted exactly 2mm from the PCB
The rest is right.

How thick is too thick? hum. A LISA XP lens adds about 5mm to the front of the LED assembly to produce the beam posted in the picture above. But I don't like it because it's plastic. So I would have to say 3mm would be the limit. Glass only, maybe even sapphire if the cost is not astronomical.

Yes, custom tooling would be justified. As for cost, I would like for each unit to come in under 25 USD and I am looking at ~200 units in the first run.

I am making high-end flashlights, but I deal mainly with the mechanical side so I do not have access to software that models light.

 

[IRstuff]

It's plastic, but so what? So is the dome on the LED itself.

I don't have anything solid to go on, but I think reducing the lens thickness by 40% isn't going to be solved by going to sapphire. Sapphire's index is only about 1.77, compared to 1.49 for PMMA.

Cubic zirconia, or Cleartran, perhaps; they both have indices >2.15. Cleartran (zinc sulphide) has an index around 2.3 in the visible.

However, in both cases, the index varies over the visible band, so there may be some limitations in color correction.

TTFN

FAQ731-376

 

[lumenharold]

I can understand not wanting plastic on the outside of the product if you think there will be a quality perception issue. As far as performance, plastic optics have do quite well (take a look at a SureFire LED based product that uses TIR - they are wicked bright). You can always have a sapphire window over the plastic optic but then that will make the whole thing thicker again. Is this a retro fit? Like a Maglite hotrod? If not, why the tight space constraint? If this is retro fit are you considering how to dump the heat?

IRStuff is right as always. You will need a higher index glass if you want the optic to be that thin. I have the Cree in my library so I stuck a couple lenses in front of it and changed the curves and material and didn't come close to 8 degrees (assuming 2 theta 1/2) until the radii were really small and the index was approaching 2.0. I didn't do a "design" so take it for what it's worth.

I would suggest you contract with an optical engineer who can do the modeling for you. If you are going to spend money on tooling it will save you money in the long run. An internet search or a search in the Photonics Spectra Buyers Guide should turn up a bunch of capable contractors. Local is nice but Gotomeeting works fine too. Lens venders might have referals though many don't like to play favorites.

The other problem you might have is getting to your price point given your volume. If you use a molded glass optic you will need to have it molded here in the states. 200 is considered a sample run overseas. 2M is production. That is not so much the case state-side. If you go with a ground glass lens then the curves might be a cost driver. With the high index and steep radii I don't know if you'll get around using an asphere for the color correction IRStuff refers to. You don't want a bright rainbow maker. Again, I'd suggest you hire an expert.



Harold
SW2009 SP4.0 OPW2009 SP2 Win XP Pro 2002 SP3
Dell 690, Xeon 5160 @3.00GHz, 3.25GB RAM
nVidia Quadro FX4600

http://www.lumenflow.com

 

[SKU427]

Plastic is not scratch resistant.

It's more or less a retrofit. The original electronics are being moved to a new shell. 2mm is the distance between the rim of the pill (tube containing the LED, driver, MCU) and the LED's base. And the shorter the distance to the lens, the less light I lose.

Seems like there isn't going to be a single thin lens that will beat what I get from the LISA XP. I'll probably just go with that one then. Thanks for all the help.

 

[IRstuff]

If you need scratch-resistance, you might explore getting a coating on the PMMA, similar to those on eyeglasses and sunglasses.

As for thickness, I think you're mistaken. PMMA's visible light transmission is quite high.

TTFN

FAQ731-376

 

[IRstuff]

Hit SEND too early. The reduction in thickness would only result in perhaps a 8% increase in transmission.

TTFN

FAQ731-376