Quantity   Component name
1 × Home 3d Printer We're working to make this project as flexible as possible so that it works on top of most home 3d printers. Our guidance system connects to the marlin firmware which is used by a lot of different printers (LulzBot, Průša Research, Creality3D, BIQU, Geeetech, and Ultimaker). We haven't tested them all so some adjustments may need to be done depending on what printer you have.
1 × Soldering iron The soldering iron will be used to "de-solder" the components on the PCB. Currently we are using a basic off the shelf soldering iron since we know how common they are for the user. We may move to a DENON DN-SC7000Z and JBC PULSMATIC 55N in the future to increase the automation aspects for more complicated uses.
1 × PCB Image dataset We collected a dataset of approximately 50 images of PCB's with hand-soldered components. You will have better accuracy if you teach the machine learning algorithm based on pictures you take of your specific set up. We show you how we did it so it's easy for you to create your own data set and build your own model.
1 × Edge impulse account We used Edge Impulse to train an object detection on our dataset of PCB images to recognize solderof PCB images.
1 × Pi cam/other webcam A webcam will be used to view the layout of the PCB. The machine learning algorithm will recognize areas with solder and guide the soldering iron to de-solder them.
1 × Raspberry Pi 4 We use this both for teaching the machine what a solder blob looks like, and for guiding the 3d printer to desolder them.