Great progress today, some instructions already working on Logisim (check it out the files on github).

But...

I figured out that for this design to work, there will be necessary two 3-state output drivers (sixteen gates) in the out of each register (and there are are 8 registers...).

24 chips only for registers is a huge overkill for a home cpu. Imagine the mess of dozens of wires on 4 (or more) breadboards... JUST for registers!

Going back to paper and redefine it. Maybe it may be needed to fix the ALU A input and output to one register. Many homebrew and old CPUs use this approach, it's the "Accumulator" register. Always wanted to know why, now I know...

These design trade-offs are driving me crazy.