11 Apr
2007
11 Apr
'07
10:42 a.m.
On 11/04/2007 17:17, Ethan Dicks wrote:
On 4/11/07, Alexandre Souza <alexandre-listas at
e-secure.com.br> wrote:
> Ethan, this is news to me. I work as an electronics' designer for
> years,
> never seen ANY kind of memory made to be piggybacked.
Hmm... this would be in a TI paper-bound book from,
say, the
early-to-mid 1980s. I don't recall the exact part numbers, but it was
probably some variant of the 4164. I remember the specifically
because of seeing folks piggyback 16K and 64K DRAMs in other designs,
and having to do the "flying select" wire and thinking how cool it was
that chip makers solved the problem with a "standard" part.
It wasn't that rare an idea. I'm sure I've seen Intel parts like that.
Motorola did it with two types. The MCM6664 is a 64K DRAM with a
refresh signal on pin 1, and the MCM6632 is the same chip with A7 acting
as a chip select when /CAS is low. The data sheet says
- Upward pin compatible from the 16K RAM MCM4416
[and 5V-only MCM4516/MCM4517]
- one half of the 64K MCM6664
- the operating half of the MCM6632 is indicated by the device marking:
MCM66320 tie A7 /CAS (A15) low
MCM66321 tie A7 /CAS (A15) high
In other words, the system's A0-A7 are used when RAS is active but CAS
is not, and system's A8-A14 are used when CAS is active, at which time
system A15 acts like a chip select.
They did the same thing with MCM6633 (half of MCM6665, which is a
similar chip but with no connection to pin 1).
--
Pete Peter Turnbull
Network Manager
University of York