Dear group,
today we started put my machine to production for some simple PCBs and its working (of course after wasting hundreds of parts). As I did not have that much time to built the machine it now took almost a year to get there, as it is only a side project.


I started in the OpenPnP project without my own machine and first did some software stuff like the logging and signalers. But as I am a mechatronics engineer of course I wanted to build a machine which fits our prototyping needs at my company. We do contracted development of devices, mainly in the IoT, lighting and measurement area. So we do a lot of prototypes in quantities of 2 to 25.The main goal was to have a fun project and maybe easen the manual production a bit which means reducing the time spent placing tiny components on the pasted PCB not reducing the total production time.
The machine itself (I did no CAD, as I was a too optimistic about my imagination) is a frame of aluminum profiles with Hiwin Rails on them. I decided to go for a
CoreXY setup which I have not seen here before, so for the sake of having something special I went for this belt driven solution (My idea was to reduce the moved weight, but actually the cable carrier and stuff makes most of the moved weight in the end). I got Peter`s head and modified it a bit by adding the necesary belt holders on the back (I need to adjust the belt heights at some point). The precision and repeatability of the machine is already good, we have not put too much effort in making it precise yet and right now we are placing 0603 without problems, I could imagine placing 0402 with it in the future.


CoreXY setup, crossing belts (they are at different heights) on the back of the machine. The rails are 600mm, the machine itself is 880*880mm outer dimension. I wanted everything to be within the frame, so of course the travel length is not ideal, but still quite large. Optical endstops for both x and y, for Z I need the special smoothie firmware for doing the special homing, but I would like to have a up to date version of the smoothie software, as the one Peter provided is not up to date.

I put the Electronics in a box, nothing special, Smoothie controller, ELP cameras (bottom with
microscope ring light) Leadshine DM556 for the X/Y steppers, robotdig pump and valves, I used XLR connectors for everything, they are cheap and can carry the current. The box because we have a lot of customers here and this way it looks quite decent (at least from the outside).

Total costs of the machine is about 1300€
Lessons learned:
- Avoid 12V (as I have for pump and valves, will replace them soon, parts are already here), go for 24V for everything
- Do a CAD drawing ;-)
- Cable carriage puts a lot of force on the system, next time I maybe want to try something https://www.igus.de/wpck/1783/overview_TriflexR for bringing the cables to the head from above. This design (think of filament in 3d printers) could be interesting for PnP machines.
- Do not try to run a machine without endstops
- Software can do a lot if it comes to squareness
Plans for the future:
- Add top lights (Priority 1)
- Fix the belt heights
- Replace the wooden board by a pcb holder with springs for quick removal
- Put the PC in the frame and add a monitor holder with keyboard and mouse fixed to the frame on a moveable arm.
- Add more feeders and maybe even automatic ones
- Add a tool changer
I bought a Prusa i3 MK3 3D printer which will arrive in a month or so, then I will start doing some printed parts (The hose adapter from Peter melted a bit when I wired the steppers wrong and they got hot).
It was a lot of fun to build it and maybe I will built another machine at some point. Its fun doing such a project and you can learn a lot along the way!
Special thanks to OpenPnP especially Jason, Peter Betz, Daniel Dumitru and the Smoothie project who made this possible!
Best, Friedrich