Close

Glare, and a plan.

A project log for Arcus-3D-P1 - Pick and Place for 3D printers

Open source, mostly 3D printable, lightweight pick and place head for a standard groove mount

daren-schwenkeDaren Schwenke 09/25/2018 at 05:013 Comments

In the last iteration of the mirror arm, I precisely masked the camera with plastic near the aperture to match the FOV of the part it gets from the mirrors.  I did it the wrong way..  I get a ton of glare coming in from the part ring light.  I think I need just a square mask way out front, with dead space behind it.  Right now it's a channel, which has the effect of giving the light something to bounce off of all the way to the camera.  Changing that.

Since then though, I discovered that the main light does a pretty good job of lighting the parts with just the light bouncing off the larger mirror, except for the two ends of it which are directly in the FOV of the camera when looking up at the parts.  

If I masked those bits off, or perhaps added the ability to turn those bits off, then the light bouncing off the bigger mirror will be enough to light the part and be completely outside the FOV of the camera.

I'm going to explore that, and probably make the second mirror larger to enhance the effect.  Since I can ditch the part light ring then, I could go 2x the mirror size and still be at the same weight.

One issue here is I have to use the same exposure, brightness, contrast well, ALL the same settings for both the down looking and up looking version of the camera.  Only one thread can have control of the camera, and it doesn't look like I'll be able to switch the settings at this point (right now.)  Also, I can't just use PWM to control the amount of light as the rolling shutter combined with my required 400Hz PWM frequency causes terrible banding in the images at anything other than 'full on'. I need to use 400Hz PWM, as that is how I'm able to control the servos with enough resolution to meet my rotation accuracy goal and I can only have one PWM frequency (without breaking down and coding just that for the second PRU anyway).  There are circuits which amount to adding an LC filter which could solve this too and allow dimming, but at that low of frequency, the required component values are huge.

The most straightforward solution is to just strike a delicate balance of segmenting the main light ring, and using the falloff from the distance to the board, to get the same brightness for both up and down vision.  Sounds like fun.  :)

For the best vision results, I'm also going to need to add a mask to the nozzle and/or print the head out in black as well.

On it.

Discussions

Morning.Star wrote 09/25/2018 at 05:39 point

If you are having trouble with glare its because you need shadowless light and that will invariably bounce into the camera from somewhere. To prevent it you can either put the light behind the camera or surrounding the part. Either location prevents shadows but doesnt allow it into the aperture.

A ring around the camera works best, similar to my jewellers loupe. I've only lit the sides, same with the camera used to film the microbugging I've been experimenting with, but 360 just improves it further.

Tentacles crossed you get round it, glare is a sod to cure. ;-)

  Are you sure? yes | no

Daren Schwenke wrote 09/25/2018 at 07:14 point

The  plan was basically that.. a ring around the mirror for a non-diffused source lighting the parts from the edge at a 45 degree up angle, and then a diffused light ring aimed down from around the camera for the lighting the boards.  Problem was for the up, I'm only using half the camera FOV, and that other half is causing the glare.  I could probably get rid of most of it with a redesign of the camera mask, but I now like the alternative better.  The edges are plainly visible on all sides, and that's what I need.

I played with the ring and moving the lights a greater diameter and in closer to the part does fix it nicely, but it then occupies the same space as the end effector when you fold it up.  I'm going to think on it a bit to see if there is a simple/clever way to fold it that won't cost me weight, but I'll probably just bounce the down light off the mirror and make it bigger.  That worked well.  Video uploading...

  Are you sure? yes | no

Morning.Star wrote 09/25/2018 at 09:59 point

Sounds like a proper nightmare. Have you considered polarising filters on the lens, thats another way of getting rid of stray photons...

  Are you sure? yes | no