Finding Your Place

Pick and place machines are essential parts of automated manufacturing processes. They are designed to efficiently handle and place components on circuit boards or other substrates. These machines play an important role in industries such as electronics, automotive, and medical devices, where precision and speed are essential. The primary function of a pick and place machine is to accurately pick up individual components from a supply source and place them on a designated location on a substrate, following a predefined pattern.

Flyby cameras are an essential component of improving the efficiency and accuracy of pick and place machines. These cameras are strategically placed to capture real-time images of the components and the work area during the machine's operation. They follow the dynamic movements of the machine head as it navigates across the work surface. This allows for more accurate placement of components and a faster overall operation.

These cameras are needed to continuously monitor and align components. They provide a dynamic view of the work area, allowing the machine to adapt to changes in component position or orientation in real time. The captured images enable the machine's vision system to make rapid adjustments, ensuring that components are accurately picked up and placed with high precision. This real-time feedback mechanism significantly contributes to the overall speed and accuracy of the pick and place process, reducing errors and increasing the throughput of the manufacturing line.

Pick and place machines are typically quite expensive, so many hobbyists choose to build their own. YouTuber and engineer iforce2d is still on the fence about building a DIY machine, but decided to dip a toe in the water anyway and design a flyby camera system that could eventually be incorporated into a full machine. The build showed that while these precision systems are complex, it is quite possible for a hobbyist to build their own.

The system is designed to interact with the open source CNC machine controller called LinuxCNC. This runs on a weeny PRU board that is hooked up to a Raspberry Pi 3 single-board computer. An OpenMV Cam H7, modified to use a global shutter image sensor, is used to capture images of components as they zip by, with a ring of LEDs for illumination.

A test rig was built that consists of a semicircle of aluminum, with a set of 4 holes drilled into it, that is attached to a Nema 23 stepper motor. A signal is sent to set the semicircle in motion, and when one of the holes is nearly perfectly aligned with the image sensor, the weeny PRU sends a signal to the OpenMV Cam H7 that triggers it to snap an image. The OpenMV Cam then performs blob detection to locate the position of the holes, which are then displayed on a small LCD display, and the coordinates are transmitted to a laptop via a UART connection.

After tuning the system using this test rig, paper drawings that represent the footprints of SMD components were taped over the holes in the aluminum semicircle. It was shown that the exact position of the features on each “chip” could be identified in just a few milliseconds. In theory, that information could be used by a pick and place machine to adjust the rotation, offset, and more of real SMD chips before they are placed on a board.

This project looks like it would take quite a bit of work, and it is just one component of a pick and place machine, but iforce2d makes it look like it would be pretty reasonable for an experienced hobbyist to build their own machine — if they have got a bit of patience, anyway.