I tried (and failed) to make a homemade brushed DC electric motor from scratch. I did this to prove a point:
Are hardware projects as easy as building Legos?
This is obviously an overstatement. Legos are just about assembly. Hardware may need unique tools to solve the problem (e.g soldering).
The point of this project was not only to get a spinning electric motor working (it ended up not spinning :( ) but also to see how reliable llms are at guiding hardware projects. It’s easy to vibecode a calculator, but how can an llm spatially reason when assembling parts or predict proper current levels through a given material?
I also feel like learning complex topics like this can be efficiently learned by doing. And there’s no better way to learn about Faraday’s law than to see inductance prevent your motor to move in real life :( (that and many such cases).
The design is based on someone else’s blog which ChatGPT scraped off the internet. The Bill of Materials (BoM) was validated + procured by ChatGPT. And finally, I extensively fed images to ChatGPT throughout the assembly process, ensuring the geometry and functionality matched the blog’s instructions.
It felt like I was cooking a recipe.
The difficult part was actually validating an e2e current response after the whole thing was built. I added some reflections at the end that summarizes my shortcomings.
LLMs can explain the theory better than me, but I will try to describe the high-level intuition without going too deep into the weeds.
There are two electromagnets. The stator is the base coil at the bottom and the armature is the smaller coiled bar containing a bundle of soft iron core nails. We could replace nails with any ferromagnetic material, so long as it concentrates and intensifies the magnetic field.
When current flows, both of these coils create a magnetic field $B$:
$$ B \propto \mu N I $$