LLM Based Computer-Aided Design and Simulation For Electromagnetic Devices

• Mekhai Waples, Computer Engineering, University of Delaware

• Nathan Lazarus, Electrical and Computer Engineering, University of Delaware

Additive manufacturing has allowed even casual users to make parts through low-cost printers available at makerspaces and libraries. 3D printing of electronics has, however, remained the domain of the highly specialized user. Large language models (LLMs) enable users to generate functional code in a variety of programming languages, even with minimal training. Here, we demonstrate that LLMs such as OpenAI’s ChatGPT and Google’s Gemini can be applied to 3D printing of electrical devices such as resistors, capacitors, and inductors, including geometry definition and device simulation. Both conversational-style textual prompting and diagram-style visual prompting are investigated in ChatGPT 5.3 and 5.4 and Google Gemini 3’s Fast and Thinking modes to generate STL files of electrical components. LLMs are also demonstrated to generate physics simulation code that predicts the resistance and capacitance of 3D-printed conductive structures without manual programming. The resulting devices are then printed in fused filament fabrication (FFF) using dual extrusion with thermoplastic conductive composite filament conductors and non-conductive thermoplastic filament mechanical dielectrics. This approach is a powerful technology for enabling access to additive manufacturing of electronics for those without specialized expertise.