An introduction to the Bit Preserve project by James Lewis. The aim is to generate proper schematics using KiCAD from (badly) scanned PDFs of schematics of various computers.
There is a decent book that discusses the technical details of the Apple IIgs, but it does not get into chip or board level design. For that detail, I had to look at the original schematics. While I am ecstatic that someone archived these original documents as PDFs, I quickly became frustrated. Sometimes the scan quality is not very good, and it is nearly impossible to search for symbols across multiple pages. I thought to myself, “There has got to be a better way!”
Instead of relying on these bitmapped documents, I decided to re-create the entire Apple IIgs schematic in KiCad. Why KiCad? Long story short, I settled on KiCad because it is an open source tool and has an accessible file format. Those two qualities make it ideal for preserving a schematic long term. For fun, I started streaming my efforts on Twitch. What surprised me is that people started contacting me about other computer schematics. A friend and colleague emailed me a ten-megabyte zip file with Commodore 64 schematics he collected over time.
The idea is to recreate in machine readable form: it’s not PDF as such that’s necessarily a problem, but that the PDFs are imperfect scans and not searchable. But a captured schematic is more valuable even than a legible and searchable PDF, because it contains connectivity as part of the data structure. That does mean that the circuit can be simulated, implemented, modified.
Regarding original schematics on PDF: Some of them are not so readable and some even contain errors. Also, habits change, as do notational conventions, and some older schematics are apparently hard to read today (e.g., ICs split to multiple symbols, where older schematics tend to focus more on the conceptual aspect than on board layout and actual building blocks.) I’ve seen this with the Computer Space schematics, where modern, verified schematics were apparently a substantial improvement.
(One of Lynn Conway’s stories about IBM in the early 60s was that different people, even in the same department, used different conventions for logic gates. You couldn’t know if something was NAND or NOR without context.)
I’ve put scanned schematics of old radios into modern schematic CAD in order to better document and understand changes. The differences in modern notation are a huge help there, too - e.g., on those schematics mf or MF probably means microfarads, and mmf or MMF is picofarads (milli-milli-Farads). Some schematics use uf or uF (probably not an actual mu), though, so as you note regarding gates at IBM, context is essential.
