Lens and Optics Question

[mribble]

I designed some electronics for a high speed 1 microsecond LED flash and I'm pretty happy with those.

Now I'm working on optics to reduce the scatter of the light.  I posted a video about some of my optics experiments here:

Since the video I've come to realize the orientation of the fresnel lens matter, that a reflector doesn't help, and that I might need a difuser.

The problem is I have about a 1x1 inch array of LEDs and I want to take all the light from those LEDs and focus it on a 6x6 inch square.  The source of the light is not even (since it's an array of LEDs) , but I want that 6x6 inch destination to be a smooth light.  I image this problem is solved in projectors.  How do I solve this problem simply and efficiently?  I'd prefer to keep the optics very simple/cheap and will sacrifice some performance to keep the solution cheap.

I couldn't find a optics forum on the web where I could post some detailed questions (the ones I found were applications based like for guns).  Does anyone know of such a forum?

[thermistor-guy]

...
The source of the light is not even (since it's an array of LEDs) , but I want that 6x6 inch destination to be a smooth light.
...

I can't access your youtube link (due to employer's firewall), so I haven't seen your experiments.

One idea I can suggest, for smoothing the intensity variations of the composite beam, is to mix the individual LED outputs using a small integrating sphere, and use a small output port of the sphere to illuminate the target, possibly via a lens.

The sphere could be small, cheap, and DIY, internally coated with white paint (TiO2) from an Arts supply store. The LED outputs could illuminate the sphere via cheap optical guides e.g. plastic optical fibres.

DIY integrating spheres are interesting per se, and may help you with your experiments.

[mribble]

That is certainly a different direction than I was thinking.  The issue is I'm using 25 fairly big LEDs so I think even cheap plastic fiber optics will get expensive.

I can imagine a solution using a parabolic mirror to focus all the light to a point and then lenses to distribute that light evenly.  But I think there must be a way to do it with just a few (hopefully 2) simple lenses.  I've been reading about condenser lens configuration, but I still haven't wrapped my head around it.

[RandallMcRee]

Yeah, I think projectors use (used to use?) diffusers.

All kinds of items for you here:
https://www.surplusshed.com/

They only have one diffuser, it looks like. You could also get a cheap optical bench and ditch the alligator clips.

Good luck!

[thermistor-guy]

...
That is certainly a different direction than I was thinking.  The issue is I'm using 25 fairly big LEDs so I think even cheap plastic fiber optics will get expensive.
...

One option would be to have two spheres. Mount the LEDs directly on one. Couple the first sphere to a second sphere, e.g. via a POF or even a short hollow reflective tube. Use a small output port on the second sphere, with the port placed at a right-angle to the inter-sphere coupling.

The potential advantage spheres have over a diffuser is reduced optical losses. However, I don't know if this is practical for your application.

[mribble]

Clever idea again, but I don't think it works for me.  The issue here is it requires excotic manufacturing capabililties.  I want to mount the LEDs onto a flat pcb because that is what pick and place machines do well.

[mribble]

RandallMcRee, I missed your response.

I've seen several pages explain simplified projector optics.  Maybe they are all so simplified that they don't mention the diffuser, but my understanding is they don't use a diffuser because it would lower light output significantly.  The bottom of this page explains how I understand projector optics to work: https://www.edmundoptics.com/resources/application-notes/optics/optics-application-examples/

If I need the optics to be more accurate I could get a cheap optics bench, but this is very much a temp rig.  My plan is to 3d print a case to hold the lenses once I understand the optics I want to use and how I want to space them.

[StillTrying]

"Maybe they are all so simplified that they don't mention the diffuser, but my understanding is they don't use a diffuser because it would lower light output significantly."

As it's easy to get an image of the filament on screen, they make the filament a small rectangular shape so it fits!

https://www.medlamps.co.uk/osram-hlx-64655-ehj-a1-223-24v-250w-g6-35.html
http://www.freesciencefairproject.com/physics/Optical_Instruments/index.html

[mribble]

Oh, so they are cheater and starting with a good light source 

[PlainName]

Interesting projet 

Perhaps it might be worth investigating individual lenses for each LED. Then you can have the proper focus but arrange for the image of each to snuggle up to its neighbour (or overlap slightly).

An alternative might be to point the LEDs at a reflecting, but matt, surface and focus that onto your target.

[mribble]

I've thought about lenses per LED, but with 25 leds that just is going to be too expensive.  I could see a cheap custom plastic lens/light guide working, but that is also probably beyond the scope what I can reasonably do.

A bumpy reflector seems more possible, but I think I'm going to see if I can get the simpler solution I mentioned in the video working first.  If I can't then a reflector will probably be my next attempt.

[BBBbbb]

As you're looking at the most cost effective solution the way I'd proceed:

1. Try to get better alignment - i.e. better DIY 'optical bench'. Try to isolate the environmental factors by putting it into a dark tube/box with only the lens at the output. At the first glance I'd blame the alignment first, but not exclusively. A lot of things in theory don't matter, but in practice they do (lens orientation for example).
2. Try with different focusing lens and/or LEDs. Does the lens have AR coating... If not, how much in price would the coated one add. 
3. if nothing works a diffractive optical element - diffuser/homogenizer might help, and some might fit your access and budget.

Anything more complex than this would increase the complexity/price significantly.

[TerraHertz]

I had a similar problem with an illuminator for a microscope.  http://everist.org/NobLog/20150510_Olympus_bhm.htm

If your light path is all lenses, then there's no way to even out irregularities due to the light source.
The only solution is to include a diffuser stage. And that has to be as small diameter as possible, with all the light from the source(s) directed on it as evenly as possible. THEN you can use lenses to control the spread from the diffuser.

A good source of diffuser film, is the backlighting structures of old LCD monitors.

Why not use the more powerful flat plate LEDs?  Like in http://everist.org/NobLog/20150628_LED_lighting_mk3.htm

[Brumby]

If your light path is all lenses, then there's no way to even out irregularities due to the light source.

I was thinking this all through my reading of this thread.  Even focussing with a parabolic reflector doesn't change the fact that each ray passing through the focal point keeps its direction vectors and they will affect how each ray traverses subsequent lens elements.

These, however, seem to offer a rather good source geometry (from the link offered by TerraHertz):


I'd call that a solution.

[CatalinaWOW]

You might simulate an integrating sphere by making a cube of LED boards.  Soldering the boards together is a chore, but not advanced manufacturing. 

The shape of an integrating volume is less important than the ratio of the internal area to the exit aperture.  A big key is the reflectance of the inner surface.  Be sure to paint all non-LED surfaces white.

[TerraHertz]

These, however, seem to offer a rather good source geometry

Not as good as you'd wish, because optically they are an array of bright dots. Even the ones with spectrum-shifting phosphors, still look like an array of dots. Just fuzzy blue dots in a background of the desired color. After any number of lenses the result is STILL an array of dots. The best you can do is blur them.