>That's the first thing that puzzles me. It
was my understanding that
>these were 128Kbit chips, so 18 of them would give you 256K bytes (+
>parity) of memory. Can you confirm, please, that 18 of them really are
>512K
>> The top chips are stamped MCM66128L20; the bottom chips have no part
>> number.
Maybe the bottom chips are just EMI filters/pull ups??? :oO
There are NO possibilities that memory can be piggybacked this way. When
you piggyback two memory chips, at least ONE of the pins must be routed
separately. At least this is what logic tells me.
Of course.
The thing is, a normal 64K DRAM chip -- a 4164, for example -- has 15
connections used. That's 8 address lines (multiplexed address, of
course), data in, dara out, RAS/, CAS/, WE/, +5v, ground. That chip comes
in a 16 pin DIL pacakge with one pin not used.
It is my belief that in thses piggybacked modules, each package contains
a 64K DRAM die, but the pinout of the 2 DIL packages in the piggyback is
slightly differnt. One pin is not-used on one device and RAS/ (I think)
on the other. And they're the opposite way round on the other package. So
when they're piggybacked, you end up with effectively 2 64K chips, with
common connections for everything but RAS/, which is separate for the 2
chips. That's hwo you can select one chip and not the other.
I never seen a 128K *D*ram chip. SRAM exists. This
is something that is
I've seen an module with 2 surface-mount ceramic pacakges on top. It was
used in one version of the PERQ memory board. Again, much the same idea
as regards selectign the 2 devices.
I've never seen half-good 256K DRAMs :-). There were some 32K DRAMs which
were hald-good 64K chips, though.
-tony