DIY reflector paint

Klaus Kragelund wrote in news: snipped-for-privacy@googlegroups.com:

Nothing hand applied would be good.

Do a youtube search for aluminum sputtering or mirroring. Or Bismuth or... other. But there are metallization videos out there I am sure.

And model car makers used to chrome plastic trees full of 'parts' for our models. So it can't be too hard to metallize it. Electroplating follows the surface it is put on though so your 3D printed part would have to have some good surfaces on it where you want the mirroring to be or you would have to post process them before you 'chrome' it.

A paraboloidal mirror is your friend. You can also get a LED reflector:

ASSY REFLECTOR LUMAWISE 10DEG 43MM SNAP 1-2154430-1 TE Conn $3

Cheers

Phil Hobbs

What about total internal reflection? Is that viable in this case? Can you get the geometry to work?

The LED I am using is this one:

I have tried parabolic reflector, but due to the construction of the PCB I could not position the LED in the focus point

Maybe I should redo the setup, since I think you are right, the off the shelf reflector is a lot easier

Cheers

Klaus

It sounds as if what you want to do, is something vaguely similar to what Bausch & Lomb did with their "Nicholas" illuminator for their stereo microscopes. If I recall correctly the B&L Nicholas illuminator uses both lenses, and a reflective tube to collimate the beam.

Any sort of paint I can think of, is going to turn your reflector surface into a diffuser... the unevenness of the paint surface is going to be quite a few wavelengths. You'll get a lot of light coming off of it at angles which result in a "sort of parallel" field, overall... that may or may not be adequate for your application. If you just need effective illumination, maybe so... if you're expecting sharp-edged shadows, probably not.

If you really do need a well-collimated beam, you're going to need something approaching an optically-accurate reflective surface. To me that suggests "smooth it carefully, buff on some graphite, and then electroplate it". A commercial silver-plating company might be able to deposit-and-polish.

I suspect that a lens-based approach is likely to be easier to get working properly. Sure you can't find anything in Edmund's catalog which could be pressed into service?

Lenses are heartbreakingly inefficient for this job, trust me. A large-aperture paraboloid with the LED at the focus (pointing at the mirror) is the ticket.

LEDs are spatially incoherent, and so radiate into a hemisphere (pi steradians projected solid angle). The projected solid angle of a lens with a given numerical aperture is pi*NA**2. (A hemisphere corresponds to NA=1.) So the light collection efficiency of a lens is just NA**2.

A super-large aperture lens, such as 25 mm FL, 25 mm diameter, has an NA of 0.5, meaning that it collects only a quarter of the light. It also has horrible aberrations, so the collimated beam quality is poor. Most lens-based condenser systems are much less efficient than that--a 100 mm FL lens of 25 mm diameter collects only 1/64 of the light.

Cheers

Phil Hobbs

We did try with a lense and a fiberoptic cable. The loss was horific. So like you say, I will need to try another solution and the reflector looks good

That stuff is garbage. IME there's nothing better than carefully cut and pasted strips of kitchen foil. Unless you want to splash out and get the thing professionally silvered, of course.

Might a Fresnel be better here?

Maybe press aluminum foil into a mold?

Vacuum forming can be cool. I wonder if you could vac form aluminized plastic.

There's probably a suitable cheap reflector already available somewhere. Maybe in a car tail-light or something.

pull a lens from one of theose $3 flashlights, they can project an image of the die on the wall.

then what?

you pretty much have to use metal to make a mirror.

If you have a vacuum chamber you can deposit aluminium, if not there's a chemical silver process.

It's really hard to make a Fresnel lens faster than NA=0.5. Higher refractive index helps with the NA, but the Fresnel reflections get quadratically worse.

Cheers

Phil Hobbs

Agree 3D printing using FDM is coarse, stereo lithography is much smoother and has post processing options. Check out Shapeways.com or many other 3D prototype vendors.

We have 100k dollar 3D printers in our facility, so we do not have problems with coarse surfaces. Cheaper machines for simple prototypes outputs samples that can be sanded lightly to provide smooth surfaces

Cheers

Klaus

Have you checked

Your best bet for a lens is one off those cheap LED torches which will almost generate a parallel beam image of the chip. Approx 15mm diameter.

Take the lens off something like this one (other brands are available).

The one I have is almost good enough collimation to point at stars with. You could certainly point at a whole constellations.

Huh? Hit up a thrift store, get any old point-and-shoot film camera, and plop the LED into the middle of the focal plane where the film used to go. It might not capture all the light, but from anything similar to a point source, it'll give you collimation.

Or if you want a wider beam, toys like this are basically an LED, a Fresnel collimating lens, and a projection lens to focus the image of a transparency onto the ceiling.

Wouldn't it be easier to start with an LED that emits a tighter beam?

That is with the light output already roughly collimated by the shape of the water clear plastic on top of the die. You can get tiny reflectors that fit around certain designs of power LED - modifying one of them to fit your particular device might be easier than starting from scratch.

If it is a one off then classic wet silver deposition chemistry isn't too hard and will give an optical quality finish if the original surface is suitably polished. Old amateur telescope books have the method.

Yes but even then you are up against the finite size of the die.

Doing something clever at the receptor end to capture as much of the incident light as possible might be more rewarding.

The 15mm diameter cast aspheric lenses on the cheapo LED torches don't do too badly at collimating a parallel beam from a power LED. I grant you that maybe half the light output isn't in the beam itself but it is pretty good and you can project an image of the LED die onto clouds.

A decent monochromatic laser with lens collimation is way better.

The OP's requirements are ultimately up against laws of thermodynamics.

You can't make a perfectly collimated beam from a LED, but good ones are brighter than most thermal sources, so it isn't a bad place to start.

"pretty good" isn't too quantitative. ;)

Sure, because its phase space volume is a single mode. Much much easier.

Yes, but the arguments are a bit subtler there since a LED is not exactly a thermal source.

Cheers

Phil Hobbs