How to Make [almost] Anything

Nathan Melenbrink

Electronics Design

In Week 4 we learned how to mill and stuff the FabISP in-circuit programmer, which was generally a succesful exercise. However, it was using a schematic that was provided and didn't require any skills in electronics design. This week was my first time designing and fabricating a board. Initially, I was thinking to use Grasshopper to create my own platform for evolutionary optimization of PCB design. However, I decided there wasn't time; also, I didn't yet understand enough about what makes an optimal board. I later found the auto-route function in Eagle, and doubted I could produce anything more sophisticated in a week. Also, it seems the difference in performance between an optimal and non-optimal arrangment is actually minimal. Anyway, now that I understand a bit more about the process, I might revisit this idea in coming weeks. But for now, I designed the schematic and traces in Eagle.

  • Tools: Roland Mill, soldering tools, etc.
  • Materials: Electronic components
  • Software: Eagle, fab modules
  • Year: 2014
Project 01a
I found Eagle to be relatively intuitive to use. I experimented with the auto-route function, though it was never perfect. It was interesting to see other possibilities for traces, but overall I don't think it actually saved me time.
Project 01b
Milled traces. After milling these traces, the fab module cut to the inside of the outline PNG (I couldn't figure out why), which just meant I had to mill the traces again, with a new outline that overcompenstated for the bit radius.
Project 01b
Components labeled and spread out on paper.
Project 01b
Finished stuffed board.
Project 01b
Creating hex, make and out files to program the board.
Project 01b
After programming the board, we noticed some odd behavior. The LED would be off until I pressed the button...though after pressing it again, it wouldn't turn off. Actually, nothing would make the LED turn off, other than removing the FTDI cable. We tried adjusting the code and reprogramming, but nothing seemed to work. So for now I'll have to be satisfied with a board that more or less works, and leave the LED mystery until later.