Sharp MZ

Welcome aboard! look for SP-5025. That's the basic which was delivered with the MZ-80 K. They are a lot of variants but that one is the original sold with MZ-80 K.

Let us see later, my actual situation is not propice for doing so.

And yes, on MZ-80 K schematics, Z80 /WAIT is directly connected to 5V through a pullup. At least, it is clear. For MZ-80 A, just a label /WAIT. It cannot be floating so it lacks at least a circuit mixing /EXWAIT to connect Z80 /WAIT.

Usually it is up to the peripheral card to set the /EXWAIT pin, not the GAL. MZ-700 LSI outputs /WAIT but not for $F000-$FFFF so I don't think it is the case for MZ-80A too. The /EXWAIT and /WAIT should have a circuit so it can feed the Z80. But the actual schematics show nothing about the circuit a...

And how does MZ-80 A handle the VRAM conflict?

So you means when GAL decodes $F000-$F3FF, it generates one cycle /EXWAIT? That's so weird.

20µs is indeed very short. I'm not sure why the GAL is connected to /EXWAIT as an input or as an output especially for $F000-$F3FF. It remains a mystery for me.

Yes but sampling should be done upon $E008.D7 (BLNK pulse) and not upon $E008.D0 (TEMPO pulse) as you seem to suggest to Ben.

And I cannot find the link between /WAIT (connected to Z80) and /EXWAIT (connected to GAL) in the MZ-80 A schematics. /WAIT is just connected to Z80 with no pull in sight and I cannot spot /WAIT somewhere else. I suspect the schematics missing that part.

MZ-80 K: MZ-700: