A few more minor updates were made to the circuit. The original design used a diode ROM to select the RAM or register that is output to the data bus of the data space. It started out fairly complex, but there were ways to rationalize it by optimizing the instruction encoding. This ROM eventually boiled down to just two 4-input AND gates and it was worth the extra chip to just replace the ROM with a 74F21.
I don't have a 74F21 on hand, so the prototype build got blocked. This provided some time to start on the PCB layout. The original plan was to use two boards with a riser and the schematic showed two pairs of 2x20 headers to join the boards. The board size requirements are starting to drop with the elimination of the diode ROM and a switch to using resistor networks for the video DACs. Switching to a single board would eliminate the risers and save even more room.
The current enclosure will fit a board up to 5" x 8". It now looks possible to squeeze everything on to a single board of this size. An example layout is shown below:
Not a lot of thought was put in to the layout of the chips, other that to see if they could be placed around the mounting holes a leave a central gap for the decoupling capacitors. The autorouter gave up with this and a lot more thought will be needed to see if it is even possible to configure the board to route at this density. I'll continue along this path though and see if I can make it work.