LitePlacer.com forums

So why can't you use the upward facing camera plus edge detection to get the part outline for leadless parts?

Reza

I haven't had enough time to work on this to get it public-ready. Based on the down looking camera performance, the algorithms need work, the current accuracy is not always good enough for small pitch parts.

How often does the needle need calibrating? This is what I'm thinking - add a light ring above the pick up needle w/ difuser. Cover difuser with black paper for needle calibration, then have it pick up a part, move it over the upward facing camera and turn on the illumination and turn off the upward camera light. You should be able to see a clear outline of the part.

Something like that.

this seems interesting - template based matching - http://www.azom.com/article.aspx?ArticleID=4401

> How often does the needle need calibrating?

When you turn on the machine or change the needle.

> You should be able to see a clear outline of the part.

The current algorithms in the software will get a decent image already. However, noise in the camera, glare from part edges, sometimes asymmetrical, and other issues like this all diminish the resulting accuracy. This is why I'm saying it needs work. I will look into this soon.

My 2 cents:
Use the bottom camera with red light to get rid of the glance for the alignment.

Hiyas!

I have a whole bunch of those factory-standard packing trays laying around. It'd be awesome if they would be supported. As such, I am super-willing to measure them with my trusty (mm scale) callipers for you, if you so desire. Shoot me an e-mail.

It is indeed a "fact"-ish thing that those trays are cheap to free at several places. They are made by the billion, probably somewhere in the more Asian regions, and as such the market has sort of saturated itself a little.

It might even be an interesting idea to have them stocked somewhere to sell to people wanting them, that also allows some control as to which trays you already support. I'd be almost willing to search a plastic molder in the Taiwan/China area that already makes them and ask them if they can make 500 of each size, but in different colours for camera recognition. Though I think 500 is way below MOQ and that idea will die swiftly. But, blue sky thinking and all.

But since an all encompassing machine vision solution would be the best, if or when I do order your kit, I'm happy to play around/experiment with some of the camera issues as well, loads of stuff in my lab, from hundreds of colours and types of LEDs to just about every allen-head screw from M2 through M12. Happy to run some experimental FWs/SWs if you promise they won't fry

I'm still pondering the purchase thing.

As we are discussing pickup and (optical) alignment of components from tapes in other threads, the IC trays came back to my mind...

I think it would be great if we could enhance the "setup tape position" dialog to not only support cut tape strips with sprocket holes but also simple tray geometries for ICs.

Let's say I have such standard tray from mouser, I need to teach the machine:
- the number of columns and rows
- approximate center of top left and bottom right component (could be done manually or semi-automatic) to calculate pickup locations inlcuding rotation error

Assuming we had optical aligment using upward camera implemented (as discussed here http://liteplacer.com/phpBB/viewtopic.p ... 6&start=20), that would be good enough too place a series of ICs.

Just order a kit for myself, and see support for this cheap trays as a great advantage.
Most of the IC's I get is delivered on this kind of trays anyways.

What is the resolution of your cameras?
- could not find this in the specs.

Is it possible to install cameras with higher resolution?
- what would be the boundries for that?

Would I get any advantages from a higher resolution in combination with the actual software?

Rainer
from Hannover, Germany
www.steamboating.de