Small 3D Printer Rebuilt as an OpenPnP Pick and Place Machine

Oct 12, 2026 Last reply: 1 hour ago 3 Replies
Jump to replies (3)

The hardest part of turning a hobby printer into an assembly machine is getting placement accuracy out of hardware that was never meant for it, and tweezy1337 handled that from two directions. Cutting a hole in the bed allowed the upward-facing camera to be mounted underneath, which cut down the number of Z moves the machine needed. Later, a spring-loaded pickup tool removed the need for very precise Z moves altogether, and top and bottom vision were both working, with live camera feeds used to calibrate how components sit.



The base machine is a Creality Ender 2, and the frame is black aluminum extrusions. The moving head carries a downward-facing camera, a vacuum nozzle and LED illumination, with a small control board sitting on top of it. Printed mounts hold the cameras and the pickup tool, which started life as a cheap SMD suction pen. The cameras are ELP-USB100W03M USB models feeding a PC, where OpenPnP sends all the gcode to the machine. Suction comes from a budget vacuum pump, and a 12V solenoid on a fan port routes it between the pickup tool and ambient air, switched with m106 and m107 commands. Lighting runs through m42 on pin 29, and the stock firmware stays untouched, with configuration done through gcode driver tricks in OpenPnP.



The board workflow around it is stencil plus solder paste, then automated placement, then a toaster reflow oven, and the machine served many PCB builds in this form.



One trade-off: at the time there was no part rotation yet, and a hollow shaft NEMA8 plus a Juki nozzle were on the way, though how that upgrade turned out is not stated. The machine was eventually stripped for parts after the need for big batches of boards went away.



Project by tweezy1337. Watch it on their YouTube channel:

formatting link


The part that stands out to me is running the vacuum solenoid off a fan port and toggling it with m106 and m107 while leaving the printer firmware stock. That seems a really clean approach, since all the switching logic lives in the gcode that OpenPnP sends anyway, and it avoids touching the Ender 2 firmware at all. The trade-off, as I see it, is that the whole vacuum behaviour depends on getting those gcode driver tricks right, but given the machine ran through many PCB builds, it clearly held up.

I am curious about the under-bed camera arrangement. How was the hole in the bed finished off around the upward-facing camera? With a camera sitting underneath looking up through the bed, I would imagine keeping dust and stray solder paste out of the lens area matters a lot for the vision alignment to stay reliable over many builds. Did the maker add any kind of guard or window around that opening, or does the camera just sit exposed below the cutout?

Since the maker mentioned the tool head was designed for an Ender 2 but might also work on an Ender 3 and a CR10, it would be great if someone with one of those printers tried the printed parts and reported back. That would settle whether the head really transfers across machines, which is currently untested. Pairing it with a spring-loaded pickup tool like the one the maker moved to seems worthwhile too, since that style of tool allows part rotation, which this setup was still missing at the time.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required