Hacker News new | ask | show | jobs
by Vrpe 2688 days ago
Well layed out, my understanding of electrical engineering is a bit lackluster, but I appreciated the article nevertheless.

How hard would it be to self develop a multi color laser projector? How would it work?

3 comments

Sure, rather easily in fact. You just need three laser diodes (Red, Green, and Blue), each pointed into the appropriate side of an RGB dichroic optical combiner cube. The cube redirects each of the three beams, combining them into a single beam emitted out of the remaining side (1) (the prisms can also function in reverse to split collimated a white light into constituent RGB components).

In lieu of a cube, you could use three (R,G,B) dichroic notch reflectors (mirrors) (2) and combine the lasers 'manually', but the cubes are nice because you don't have to build a mount for the I individual mirrors.

You can either scavenge a cube or dichroic mirrors from an old standard projector, or pick them up online for a few bucks. After that, just place the combined beam in the same path at the red laser shown in this example set-up, modulate the three lasers according to an image's RGB intensity values (you can separate an image frame into three RGB frames using openCV), and there you go.

Tangentially, the basic idea of scanning a laser to form an image is what underpins the Virtual Retinal Display (VRD) - a head up display technology developed at the Univ. of Washington and later attempted to be commercialized by Magic Leap (ML) for Augmented Reality eyewear. Even though ML hasn't panned out, I still see the VRD as a very viable path forward for high resolution/field of view AR displays.

1. https://en.m.wikipedia.org/wiki/Dichroic_prism

2. https://en.m.wikipedia.org/wiki/Dichroic_filter

Hmm. That sounds like maintaining the alignment of the three beams would be a total nightmare.
it's funny, we're building this set up (https://github.com/PRNicovich/NicoLase) in our lab for microscopy. From the docs:

"When aligning your laser launch, do not rush. This is not something to attempt late on a Friday afternoon in a free hour before heading to the pub. This is a task involving a great amount of precision and small, careful movements. There will be very little in the way of positive results until the task is nearly complete. Set aside up to a several hours to complete the alignment, especially on your first try"

You only have to do the alignment once, on the input to the combiner cube. Then you get a single output beam sent to the galvo.
You could use three lasers (rgb) and merge their beams using a set of dichroic mirrors, then use the resulting beam as if it was your red laser. The control logic then switches the three lasers instead of only one.
This, and then projecting the RGB beam from the rotating mirror fixed on the ceiling to a circular screen surrounding the user for full 360 degrees experience.
...or what frabert said much more succinctly :)
Both are good :)
Combining multiple beams into a single white beam is an answer, but is horribly inefficient. A better (ie brighter) approach is to scan with multiple beams, overlaying them as necessary to create various colors. It would be more complex (3x moving parts, more alignment) but the net result would be faster scanning and a greater percentage of laser time being used.
if the 3 beams were miniaturized as in surface mount diodes:

these for example

http://www.excelitas.com/Pages/Product/Surface-Mount-Laser.a...

and were carried or fibre coupled in a single "gimble?" there would not be duplication of moving carriage parts. a "trinitron like" emitter gun would be possible. moving a mirror as done by the author would also reduce the mass that would move at high speed while scanning across a large display field, it would then be a matter of timing a laser pulse of desired color to the desired position of the mirror set. a look at CRT television/display circuitry would be very inspirational, only in this case the color is emitted rather than stimulated by flouresence of a phosphor target.

Something to keep in mind is that mirror scanning for television is a very old technology, dating back to the beginnings of mechanically scanned TV:

http://dci-forum.com/back-roots-tv-history-beginning/24/mech...

Yes. There's a long history of ideas in this space. DLP projectors, with a MEMS mirror for each pixel, are one of the few successful examples. Any system where you illuminate only one pixel at a time flickers horribly, which is why this sort of thing is used for laser light shows and not much else.
- now if you can send the beam[s] through a tft display minus the backplate, and darken pixels you dont want to illuminate then scan the beam across the whole thing...This has a madmax or a maxheadroom sort of appeal to it.

i was trying to find the mems mirror around but couldnt , still looking for dead projectors to throw in the morgue of parts.

I do have wide streak of steampunk, i love retrotech, espescially when using junkyard parts and a few mailorder bits together.