-
1Install a mount for the soldering iron and camera
In our components list you'll find a mount that works for our set up. Your printer, camera, and soldering iron might be different sizes so you may need to make the holes bigger or smaller to accommodate them. Make the needed modifications, 3d print it, and install it on your printer.
-
2Attach the soldering iron to the 3D printer
Mount the soldering iron to the right of the extruder. It should slide into place and be secure since it will need to withstand quick movements. Make sure to give the cord enough slack so it doesn't get pulled out even at the furthest position. You don't want a hot soldering iron going in unexpected places.
-
3Attach the camera to the 3D printer
Mount the camera to the left of the 3d printer. It should be looking down through the hole. Secure it to the mount however you see fit (tape, glue, clamp, 3d print a seat etc.) Make sure the cord has enough slack even in the furthest position.
-
4Hook up the Raspberry Pi 4
Connect your pie to the camera and the 3d printer. You can use it both to train your machine learning model on what a solder blob looks like, and then use it for guiding the 3d printer to desolder them. You can use our trained model, but we'd highly suggest training your own model after the camera is on your 3d printer. That way the lighting, surroundings, etc. are all customized for your specific application. This should really improve accuracy.
-
5Add the machine learning capability
The machine learning component of this project will consist of a camera which streams a live video feed of a PCB with solder to a raspberry pi or micro-controller. The pi or micro-controller will run software which utilizes a machine learning model to recognize areas with solder on the PCB. After locating the areas with solder, the device will send the location to the 3D printer. The 3D printer will then position the tip of the soldering iron at that location and melt the solder away, freeing the components.
Access the machine learning model here in order to get started. You can choose which pi/micro-controller and which camera to use. Make sure you choose an ML model which is appropriate for your device type.
-
6Use the machine learning algorithm to run the guidance control
The machine learning algorithm identify solder blobs. The information it has is sent to the 3d printer to move the soldering iron to heat up each blob. As they are heated up, the component should drop down (from the other side) due to gravity.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.