For the MP-1 mod, I didn't use a custom PCB. It is generic stripboard. Of course, to put any dual-inline-pin IC onto stripboard, you have to cut the traces, or get a stripboard which is prepared for this use already.
It is very simple:
- There are only two resistors and a 100 uF capacitor, plus a long jumper wire to help with the capacitor.
- Also: a 300 mil wide 28 pin socket for the chip, and two rows of headers in the 600 mil width to simulate IC pins to go into the MP-1 socket.
- The jumper wire conveys the ground from pin 14 to an extra strip-board trace that runs parallel to pins 1 and 28. Pin 1 is the "VCAP" pin on the new chip: NC (not connected) on the old.
- Thanks to the above jumper wire, we can nicely mount the capacitor between VCAP and GND, since these are on adjacent strip-board traces. You cannot easily see the wire in my pictures because I routed it under the IC socket!
- Header pin 26 is not installed, so pin 26 of our IC socket does not connect to pin 26 of the MP-1's socket. This is due to the concern that the MP-1 drives pin 26 high with too stiff a voltage source (it is directly tied to +5V) which will overpower the Cypress chip's attempt to pull pin 26 low during the AutoStore operation, which will cause that operation to abort.
- Instead, we connect pin 26 to +5V (obtained from pin 28) via a 10K resistor, like in the datasheet. This means pin 26 is held high by us, and the chip is able to easily pull it low when it needs to.
- Pin 27 ("WE": write enable, active low) is also connected to +5V through a 10K resistor. This is due to a recommendation in the data sheet, to prevent a possible situation whereby the SRAM contents become corrupt on power up due to flakiness on the WE line.
So you can see that the two resistors and disconnection of Pin 26 are due to diligent adherence to the data sheet; perhaps the chip can work without them.