I’m working on a research project to analyze how people lay out components, and nets in KiCad and the ways engineers design PCB layouts. I’d like to collect KiCad projects (schematic + PCB pairs) for my dataset. If you have any public (open‐source) KiCad projects you can share, especially ones with at least 10 to 50 components and multiple nets, please consider sending me a link or a ZIP.
Requirements:
Must include at least one .kicad_sch and one corresponding .kicad_pcb file.
Should be saved in KiCad 7.x (or exportable to that version).
Ideally, a small‐to‐medium design (10 to 50 components, 10+ nets).
Must be licensed so I can analyze it (GPL, Creative Commons, Apache, etc.).
I’m not looking for commercial or highly proprietary files, just small-scale hobby or proof of concept boards are fine. I appreciate any help you can provide.
(If you’re comfortable posting a public GitHub/GitLab link, that’s easiest. Otherwise, you can send me a direct link to a ZIP.)
Are you training AI? For whom are you doing the research. Also, no email addresses allowed, mainly for YOUR spam protection. Scraping the web for email addresses was the #1 source last I knew.
EDIT: A quick search shows AI under your name a couple of times. AI training IS NOT research. I would consider calling it that, theft by deception. If I have this wrong, please correct me.
Due to the KiCad 7.x file syntax (the .kicad_sch/.kicad_pcb lexer/parser I built expects version 7 conventions). Although KiCad 8 and 9 files are very similar, there are a few subtle differences (like changes in how certain attributes are named or nested and so on), I am trying to keep the initial parsing reliable, so I’m starting with projects saved in KiCad 7. up to version 9.
I have the spam filters set, but I appreciate you removing the email, thank you for that.
The research is done for my next conference paper, which will be published under my University (I can disclose the institution/university name in a private message). But I think that information is also publicly available under my name online.
I am training an ML model (if precise info is needed, it is a T5‐style sequence‐to‐sequence network) to predict small schematic or PCB edits based on previously seen layouts. In other words, it’s machine learning, which falls squarely under AI since AI is a broad field.
Also, in academic research, training ML models to identify patterns (in code, images, or in this case in design files, and so on) qualifies as research, and this project aims to understand and model KiCad design workflows. With the hope of building a small “assistant” that can suggest next steps or catch common routing mistakes, once again, purely as an academic proof‐of‐concept, that is why I required the projects licensed as GPL, Creative Commons, Apache, etc. so no proprietary or private designs are ever used, just publicly shared .kicad_sch and .kicad_pcb files.
And just out of curiosity, “A quick search shows AI under your name a couple of times” In what way do you mean that, like from the Google Search, or…?
I hope I addressed well all your points and concerns.
A few years ago, i created RepoRecon which collects Github repos on particular subjects and places them in a searchable table. It updates nightly. The last i checked, it had over 30,000 KiCad-related repos. A large number of those would be KiCad projects that you could use.
I’ll also mention that you can clone the RepoRecon repository and then get the entire “database” of KiCad repositories from the docs/kicad.json file. Possibly you can do a little ML on that and sift out the repos most likely to have KiCad projects.
Yeah. My search was quick and I wasn’t going to do too much guessing. AI developers seem to feel pretty entitled to information lately. One went so far as to protest that asking artists permission would kill the business completely.
Obviously I’m a little opinionated on that subject. I wish you well on your research. If appropriate, please share back when it’s done.
It can happen, but if you have a good work ethic and follow the regulations, you can’t harm anyone. I will definitely share the research upon completion. Thank you once again
