Given that
there was already a bad joint on the replacement socket, I
wonder if there's other damage.
Possibly. The socket does correspond to the bad bit in the Sad Mac
code, so one of my thoughts is that they saw the code, tried to replace
the chip and the problem did not go away and the board was shelved.
YEs that's possible. The originalRAM was in fact good, but the error code
suggested it had failed. So it was replaced to no avail.
Have you tried 'buzzing out' all the pins
on the socket (and other RAMs
if you want to be sure) to wherever they should go (address muxes, data
bus, timing/control PALs). My first guess would be an open connection
somewhere.
I have buzzed the socket to adjacent DRAMs and the 14 common pins
all buzz fine. I have not walked Din and Dout everywhere (yet).
I would. I don't know how many layers this PCB is (I've never acutally
worked on one), but given the size, I would guess there are internal
signal layers. It's eitlerly possible one internal trace has becomd
disconnected somewwhere .
Are there any series termination resistors for the data lines? If so,
check those too. It's not impossible for one to be open-circuit.
-tony