Damus
eujc21 profile picture
eujc21
@9b2590c4a7be5aee113d8fff6e192943b23feb2d2c7928da5ed8252cad0f2982
Relays (12)
  • wss://nostr.wine – read & write
  • wss://nostr-pub.wellorder.net – read & write
  • wss://relay.damus.io – read & write
  • wss://relay.roadrunner.lat – read & write
  • wss://nos.lol – read & write
  • wss://eden.nostr.land – read & write
  • wss://relay.current.fyi – read & write
  • wss://nostr.orangepill.dev – read & write
  • wss://brb.io – read & write
  • wss://relay.nostr.band – read & write
  • wss://relay.snort.social – read & write
  • wss://relay.hamnet.io – read & write

Recent Notes

eujc21  profile picture
Brainstorming a CLI-Driven Analog Design Flow

Lately I’ve been thinking about how far we can push AI + simulation workflows for analog circuit design — without ever opening a GUI.

What if we could go from idea → schematic → simulation → optimization → BOM entirely from the command line?

Here’s what I’m experimenting with in the lab:
• Use Vertex AI’s Gemini to generate SPICE netlists for basic amplifier circuits (starting with non-inverting op-amps)
• Run simulations via Ngspice in batch mode
• Parse gain, clipping, or frequency response with Python
• Feed the metrics back to Gemini to refine the design
• Loop until the output behavior meets the design goal
• Export the final netlist + BOM for import into Eagle or AMCADD

No mouse. No clicking. Just AI + CLI + electronics fundamentals.

The long-term goal is to automate parts of the analog design process and create a reproducible design-feedback loop that could scale from lab tools to PCB fabrication.

If you’ve explored circuit design automation, SPICE scripting, or LLM-generated hardware — I’d love to hear your thoughts. What’s missing in this loop? What breaks when complexity scales? Let’s riff.