Hello, imperfection! | Random decap #1 - ATmega328p

It works!
If you can read this, it means I've correctly put stuff out there. It's not perfect, but "done is better than perfect".
The driving force behind finally publishing this blog is to help with job hunting (If you're an employer near Vienna, you might have come here via a link in my CV, in which case I appreciate your diligence), but the real target audience long-term is the maker community, from which I've learned more than any school, and naturally I wish to give back to the community.

perfectly imperfect panorama of a ATmega328p die broken during cleaning The above picture is one of my first attempts at stitching a panorama of, in this case, a crappily decapped ATmega328p - one of the world's favorite microcontrollers, especially among makers. I don't have access to industry-standard acids, so I'm just using a flame torch to burn away the plastic aka "thermal decapsulation", which tends to leave a lot of residue on the chip and requires further cleaning. Again, as I currently don't have the usual chemistry for that, mechanical cleaning plays a big role and it takes a bit of failures to get familliar with the absolute brittleness of pure silicon and get a feel for it. However, despite all of the imperfections along the way, the microcontroller in this image is still identifiable and subjectively, quite beautiful and hooked me for more (slightly less as I get more experienced) shitty decaps.

Thanks to the real pros that have created and shared their (better quality) decaps on the Internet (for example this decap and delayer by ryancor), we can see that It's fairly certainly the same chip/mask and both of them have a matching Atmel logo (so I probably didn't get a fake).

Additionally:

  • You don't actually need an expensive metallurgical microscope just to be able to tell if your chips seem legit (spoiler: I don't have one of those yet and will soon post about my process).
  • Low quality decaps can still be useful and fun, even if you damage the chip in the process.