basePnP - pick and place machine

It´s long time since last update. Sadly this project is hibernating and waiting for a reborn. I did this project for a reason - to make a small batch of PCBs for my planned product. But it turns out that making a hardware product is not that easy. Making a prototype is one thing, but serial production is completely different story...

Not enough time for updating this blog, sorry :)

But soon or later there should be a big update with all the documentation and of course some videos from real jobs.

Seems that detecting parts on the nozzle will be quite easy.

First section of the graph is with closed vacuum valve - sensor is on atmosphere pressure. Then valve is opened (no device attached to the nozzle )and ADC value goes quickly down to 3300. After device is suctioned to the nozzle, there is another drop down to 2900.

This test was performed with 0.9mm nozzle. So for smaller nozzle diameter (for placing 0402 packages) difference in pressure will be much smaller, but still be detectable.

Seems that two moths is really big challenge for such a complex project like this :D

As my thesis supervisor suggested, I did implemented pressure sensor for check if component was picked up or not. For this purposes I ordered MPXM2202AS from Farnell. This sensor has differential output with only 40mV full scale. So the differential amplifier was needed.

Here is the prototype

Output is set to 3V at atmospheric pressure and goes down to 2V when pump is directly connected. This should give me enough range to detect presence of components.

Feeders

Most of the components which i normally use came in cutted tapes. After few minutes in OpenSCAD and two iterations I came with following solution.

For the rest of the components there will be "single component" type of feeder implemented.

Make a good feeder is crucial, so I expect a lot of redesigns.

Just a quick update with current progress.

Z axis prototype.

Here is how my vacuum setup should look like.

As a vacuum pump I use reworked aquarium pump. Pumping speed and ultimate pressure is so poor, but it's enough for testing purposes. Id like to use a diaphragm vacuum pump in final setup.

Buffer is in ToDo list, since I didn't find any suitable vessel in our hackerspace.

Valve is SMC V114A-5GU, not a best choice for vacuum application, but it works :)

Trickiest part was to make a custom manifold. I've started with this block of aluminum:

and on the second attempt (+ one broken tool) I ended with:

I hope that tomorrow I will be able to pick up some parts with improvised Z axis!

I've spent all the time since lat post only by writing the SW. The latest version can be now found at github if someone is interested: https://github.com/Hyna/basePnP

Implemented features:

• communication module - GCODE sender/receiver
• Simple board alignment - use only bottom left corner of PCB as zero position
• Fully automatic board alignment - use vision to find the fiducials and then affine transform coordinates
• device manager - static tape implemented

So the SW is almost ready for the first job :) Now is the time to hack on Z axis and vacuum.

Now the trickiest part - the SW. All coding will be done in Python + OpenCV + PyQt. Programming is really not my hobby, but I'm able to to write some code which sometimes eventually works :D Friend of mine from our hackerspace offered me a help with SW. He has different approach how the things should work, so plan is to have two independent control SW. The reason why i want to write my own version is that this project is also my Diploma thesis.

Back to the topic. I did some tests with fiducials and it was really straight forward. Same as detecting the holes in SMD reels. But one problem already appeared. How to create the proper light conditions! Which color should I use? Which intensity, from which angle etc... Seems that even the guys from FirePick project do not know.

SMD tape:

Fiducials: