High Power LED Collimating optics for projector project.?


My projector currently has a high pressure fancy schmancy bulb and when it goes (after another few thousand hours hopefully) I want to replace it with a high power (~100 watt) LED.  That's the easy part.  I can interface with the projector and convince it there is a real bulb firing up - but I need the optics.  Basically, it's a DLP projector so all the light has to be focussed on just under a 2x2cm square hole to hit the colour wheel.  The LED in question is already over that size on the diode die alone. 

There are parabolic reflectors available (I think my best option) but they still have a positive beam spread - I need a shrinking beam.  It doesn't have to be perfectly collimated, but I bet that helps.  The original lamp definitely has a converging beam but its hard to measure as you can't operate the lamp with the cover open (safely).


I'm thinking diode >> reflector (like flashlight) >> convex or fresnel lens.
I can't get better measurements at the moment for I'm not near the machine and its finnicky to take off the ceiling.  Penny for your engineering thoughts?

What I need but have never seen commercially: 

LED to use:
dimensions about 5x5.5cm

Reflector I have in mind:
(rather wide beam still)
Alternate reflector:

So...what I need are lens options.  Based on my estimations I'll have several cm (up to 10cm) to come up with a good focus distance.

sort by: active | newest | oldest
The thing with the collimating optic is it needs to be designed with the beam pattern of the source. The HBO lamps used in conventional projectors are great for the optical engineers, because they are essentially a point source.

That LED is certainly cheap, but you'll make your life easier the closer to a point source you can get. What other choices are Nichia, Luxeon et al offering this week ?

frollard (author)  steveastrouk5 years ago
Regardless of source/maker, to get in the 100 watt 3-4000 lumen range, diodes simply have to be massive grid arrays like this - any smaller and you can't get the watt density + heat dissipation you need. Agreed the point source would be super ideal.

I saw someone on hackaday do a retrofit with a 30watt led with no optics, just relying on the 5% of the beam that left the 180 degree field to go 'straightish' and it was okay but rather dim...I figure if I can concentrate just a bit of a wickedly powerful led, I'll still be saving half the electricity I was before AND getting cheaper replacements.

My last headache line of reasoning is to use 'fiber' optics of curved 1/4 inch acrylic to feed many streams from many 5+ watt leds aimed at the port - the final straight stretch of tube acting a bit as a collimator (very rudimentary, I agree).
The best way is to focus it down as small as you can manage, then re-expand/collimate it.
dyzie3 years ago
Try using the primary mirror of a reflector telescope (i.e newtonian). You could cannibalize a cheap newtonian or just buy a second hand mirror off ebay.
Anything around 110mm should be more than enough for what you need.
Check Wikipedia's article on Newtonian telescopes for a better idea of what I'm suggesting.
I do not recommend trying to buy a new mirror as most manufacturers charge an arm and three toes, but you can find a 6in mirror for less than $20 if you look around.

BTW I made this account just to give you back what i can. I was ready to give up on an old led project of mine when I stumbled upon a convo on here where you recommended FastSPI. That library helped me more than I thought a *.h could.

Thank you
dyzie dyzie3 years ago
Whoops I just reread your post and noticed that you said you had 10cm to focus with ... maybe a telescope eyepiece? Glob. I was all "I am awesome for having a great idea and going out of my way to help someone". Well it was fun while i believed it.

Good luck and all that jazz
frollard (author)  dyzie3 years ago
if it had to be outside the projector to fit, that wouldn't be the worst thing in the world, just more mirrors. Compactness is the game and it's hard to be xenon for that.
iceng dyzie3 years ago
Nice of you...
frollard (author)  dyzie3 years ago
Thanks for the advice and the kind words! It's why we're all here! Contribute to the collective knowledge. I don't know everything (and it shows) - but I like to learn and share! Post a link to whatever you're working on to my orange board ('instructables wall') - I'd love to see it!
dchill5 years ago
You might look at finding a TIR optic to match the LED you want to use. There are other sources around, but one place to find them is here: http://www.fraensrl.com/prodinfo.html They have a list of lenses designed to the various light outputs from some high powered LED's. Might be a little more pricey, but you should gain quite a bit in efficiency.

The concept would be to use a TIR lens to collimate the light(you won't get perfect collimation, but maybe something around 5-6degrees) , then a short focal length lens to focus the light down through the color wheel into the integrating tunnel. You might find something suitable at thorlabs or edmund optics for lens/holder.

One other thing to look out for is the color balancing. The projector was set up around a high pressure lamp that probably has a color temperature in the 6-9K range. The LED you are looking at is in the low 3K range, so you will probably need to make some severe adjustments in the custom color settings. This will be another brightness hit to you, since this virtually shrinks the sections of the color wheel to get the right color mix.

A final note - your current bulb might last a bit longer than you think if you don't mind lower brightness. Those lifetime ratings are a bit confusing, but the wording in the standard is that the lifetime gets defined as the time it takes for 50% of the test population to drop below 50% of its initial brightness. I've seen systems limping along at 3-4x original lifetime rating. Good Luck! D
76543215 years ago
If it still works dont worry about it :D
frollard (author)  76543215 years ago
Again, this is planning for when it stops working.