Hijacking Address and Control lines for memory at the source? Diodes?

[Trash80toHP_Mini]

Curious about this one. A pair of diodes wired to address and control lines of two banks of a memory controller seems to open up the possibility of hijacking those lines from Bank B and sending those from Bank A down that pathway along with the data bus which goes both ways at the same time anyway.

The diodes would prevent feedback to the memory controller when one line OR the other is active on their suddenly shared trace. The Mac can't use the both banks at the same time anyway, or is that a huge ASS-ME error?

Thinking of mounting a tiny PCB with diodes on the raised A/B leg pairs and short jumper to each trace pad vacated by raising the Bank B Control or Address leg?

It'd be more work, but a passel of really short antennas would be a major improvement over the jumper cable length antennas currently on the table.

 

[Trash80toHP_Mini]

Forgot to mention R/W Enable for Bank A would still need to be jumpered to the relocated Bank A SIMM at the end of Bank B's traces. One long antenna would be a lot better than 23 of them.

Also, the PCB would be mounted on top of  MDU so its raised legs can be soldered to it.

If possible, this is the least evil of the ways I can think of for taking a IIsi to 128MB.

 

[Trash80toHP_Mini]

View attachment MDU-000.PDF

 

[techknight]

You still have to keep the inactive bank refreshed. Otherwise it wont work. 

Also the address decoder/mmu/system must be aware of the available memory and be able to do the bank switching via decoding. Not sure how or if this would ever work IMHO.