Very interesting stuff and I appreciate you sharring your work! When I saw that you make new socketed accelerators for the SE/30 coupled with recently learning about Bolles grayscale card, I got inspired by the possibility. I might give a 4 card set up a go and see what happens in the future, but the prospect of an actual socketed 68040 accelerator is really cool. Did you ever run into compatability issues with your custom cards?
I am in no way qualified to make something like what you did, but it would be a fun goal for me to work towards as I'm interested in learning more about electrical engineering and PCB designs.
The reverse engineered stuff is pretty safe, IME.
Things I've designed from scratch like the booster 2.0, 30video cards, NuCF, etc have more room for incompatability but in every case it's been an existing issue with the other card rather than "my" fault so to speak. Sometimes this can be mitigated, sometimes not.
For example, with my early 30Video grayscale cards sometimes exposed an issue that certain carrera cards had with their /DS line that was apparently vulnerable to an unterminated line introducing noise (best guess). Colorboard video cards don't handle their /DSACK lines correctly and are incompatible with boosters for that reason. NuCF had an issue with IIfx having a poorly /AS strobe. That sort of thing.
It‘s probably time to condense everything into a new replacement logic board
Truthfully I don't think trying to condense it makes sense - the SE/30 (II/IIx/IIcx) architecture is the worst Mac II architecture simply due to its age, it doesn't make a lot of sense to use as a basis for something new. You'd have to increase the entire logic board to 6 layers (or continue to use a daughtercard) and would still be saddled with the worst possible 040 performance.
I think it'd make more sense to build a native 030 system that happens to be mac-like enough to boot a patched ROM. Similar to what was done on the products like Daydream for NeXT that allows booting Mac OS on NeXT hardware with a ROM-box. The GLU chip in particular isn't terribly complex, a functional replica would be straightforward to make as it doesn't need to be cycle or behaviorally exact at the bus level - just present the same interface (or lack thereof, actually) to the software. The SWIM can be bypassed. The ASC presents a problem, but technically I think there's nothing precluding you from blindly acknowledging writes to that space as I seem to recall
@Bolle saying a SE/30 board will boot without the ASC. The sony chips likewise aren't important if the ASC is set aside. At that point all you have left are commodity chips or functionality that can be replicated with optional additions (sounds, SWIM) for completeness if desired.
040 would be a tougher sell, but if the rom patches done by the Turbo040 can be reverse engineered there's no reason it couldn't be baked into a ROM.
