Building a manual pick&Place machine

After building 3 of the UniversalTx boards with tweezers, I was getting really fed-up with that method. While there are some open-source fully-automated P&P machines in development, nothing really ready to go now, and they are quite expensive. Manual machines are also quite expensive to buy. SadSack posted this link in the UniversalTx thread: vpapanik.blogspot.gr/2012/11/low-budget-...-pick-place.html?m=1 and it seemed like something that could be thrown together in an afternoon (with the right parts on hand) at reasonable cost.
I already have a bunch of 2020 v-slot 2020 extrusion around, so I thougt I'd make mine out of that. I whipped up a quick sketchup to see how it'd look:


Building it out of extruded aluminum is certainly more expensive than SadSack's link, but since I have most parts lying around, it won't be too bad. I just needed a few extra wheels and the gantry plates as well as a better pump, a foot switch and needle bearings. I am still trying to decide how big I need it. I think I'll make mine a little smaller than vpapanik's. roughly 500mmx250mm which should be plenty large for any board I am likely to hand-assemble. Once the spare parts come in, I'll try to throw it together and hopefully have it working before it is time to assemble the next iteration of the UniversalTx board.

I built the p&p frame:




it seems to work well, except for the spring-loaded vacuum tube. The needle bearings don't work really well for sliding in and out. I tried a steel tube instead of aluminum but it didn't help much. So I still need to work on that still. I also need to 3d-print some end caps for the vacuum tube.

I ordered some linear bearings and a matching tube, which will hopefully take care of the issue.

yep, that is exactly it. In this case, I use an aquarium pump (which can easily be reversed to a suction pump), so not too much flow. I'm planning to use a foot-switch to control it, because from prior experience I found that any movement of my hand tends to cause jiggle that mis-aligns the part. The vacuum assembly is spring-loaded and can rotate and move in the z-axis, which makes it easy to precisely align parts.
I found that the needle bearings can work ok if you use 2 AND you get them precisely aligned. I did that by abutting them, and now my tube slides easily enough, but I still plan to replace the assembly with linear bearings which should be more precise.

As far as motorizing it, I think this isn't the best design for that. The complication in pick&place is the machine vision which identifies whether the part has been picked up and its orientation. They aren't easy tasks to solve. But for prototyping, I am hopeful this machine will make assembly much less stressful. We'll know in a couple weeks once the next generation of boards come in.

Now that I have all this equipment, I may run another round of multi-module boards using SMT components.