Smarter Automation for Complex Analog Circuit Design Using AI

What is it about?

Designing analog and mixed-signal (AMS) circuits—used in things like smartphones, medical devices, and cars—is a difficult and time-consuming task. Unlike digital circuits, AMS circuits need to follow a lot of specific layout rules to work correctly. These rules, called design constraints, help ensure that the final chip works as intended. For example, some parts of a circuit need to be symmetrical, some need to be grouped closely, and others must match in size or shape. Today, engineers often have to manually identify and input these constraints into layout tools, which is slow and can lead to mistakes—especially as circuits grow in size and complexity. Some modern tools use AI, like graph neural networks (GNNs), to help with this, but they only work well for certain types of constraints and small, simple designs. This research introduces a new method that uses advanced AI techniques to automatically learn and apply a wide variety of layout constraints for large, complex AMS circuits. It combines two new algorithms—Net-First GNN and Selective Topological Search—to accurately confine layout design space and apply them across different levels of circuit design. The system was tested on very large and complex circuits and was able to handle them with high accuracy and efficiency. This new approach makes AMS circuit layout design faster, more reliable, and more scalable, helping engineers save time and reduce errors.

Featured Image

Photo by Laura Ockel on Unsplash

Photo by Laura Ockel on Unsplash

Why is it important?

As electronics become more advanced, designing reliable analog components quickly is critical. This method could significantly reduce manual work, speed up development, and make it easier to create high-performance electronic devices.

Perspectives