In jjRobots we have been involved in the world of electronics manufacturing from day one. We have gone from manufacturing PCBs in a DIY home oven (and abandoning it after a few weeks) to having to outsource the manufacturing to external companies (that was mandatory just after a few months).
Quality control and visual inspection of the electronic boards has always been a necessary, but extremely cumbersome and time-consuming job. And we always discussed how to improve that process, not having to expend a large amount of money on this (and time learning how to use a classic AOI)
Mrobotics, our partner in California, using their Mirtec AOI system, has always helped us by inspecting the PCBs/panels , but increasing the Mrobotics´s workload became a bigger problem more than a solution for both companies.
Three years ago, we started working with neural networks for external projects and saw the potential of using this technology in optical inspection systems. Thus, AgnosPCB The neural network powered AOI was born. A company that has given birth to a software capable of checking the integrity of a PCBA /panel or circuit board through a simple photograph, within seconds. And highlighting the location of the faults on the image.
The company was created in September 2021, after processing more than 60K images from our manufacturing experience. And today, companies such as Bosch, Intel or Apple use our services (adapted to their particular needs) to inspect their PCBA. AgnosPCB obviously, with the idea that anyone can use its services without the need to invest large amounts of money in an AOI, created the AgnosPCB inspection platform. A simple but capable desktop AOI. With it, you just need to press a button, and the camera will take the best possible image of a PCBA and start the inspection process.
How does AgnosPCB AOI work?
But what makes AgnosPCB special?
Well, it behaves like a highly trained human technician who is able to find the errors present on the PCBA in a matter of seconds. No matter the number of components, the size of the components, the dimension of the PCB or the small differences between PCBA and PCBA: The neural network KNOWS what an electronic component is and what is not, so it will not pay attention to differences such as light reflections, if a resistor is rotated 180º or if the solder volume is slightly different.
This AOI service will only point out where the faults are, as a skilled technician would do. Type of faults detected:
- Short circuits
- Missing elements
- Very much displaced
- Problems with polarity of diodes, capacitors…etc.
So, we have here an affordable but dependable AOI system. As the Apple´s technicians called it, “easy to use and capable lab friendly AOI”.
If you want to use your own camera system to take the photos of your PCBs, just do it. We have an API available too. So you can embed Agnospcb AOI in your own AOI system. As long as you can take a photo to the PCBA you are set to go. But, obviously, the better the photo, the better the inspection results.
Any other advantages?
Yes. This inexpensive AOI system learns. The more images it sees, the wiser it becomes. Just like a human. That is (the most?) impressive feature of this technology. You can let it know if there is a certain area / element you want it to pay special attention just pressing a key. The neural network will take note and the Agnospcb technicians will process and prepare that additional info so, in the future inspections, the behaviour of the IA will be, lets say… “wiser”