-----Original Message-----
From: Jos Dreesen / Marian Capel [mailto:jos.mar@bluewin.ch]
Sent: Monday, January 20, 2003 12:10 PM
To: cctalk(a)classiccmp.org
Subject: Re: Core memory speeds
On Monday 20 January 2003 12:02 am, ben franchuk wrote:
Since I am building a classic TTL style
computer,
what was the access and cycle time of core memory
before it vanished and just when did it do that?
Also when did 74LS come out? I am aiming for late
70's early 80's time frame here since this looks
to be the transition stage from the the old to
the new. Ben
My 1975 core memory products catalogue ( Philips components )
lists modules with cycle times ranging from 0.65 to 1.5 us,
both for 8kx18 stacks.
My newest core memory stack sits in my Philips P856 :
dated 1980, cycle time unknown.
My own TTL computer, dated 1986, just uses 8kx8 SRAMs......
Jos Dreesen
With core memory the general rule was the smaller the
donut the faster the cycle.
The real limiting factors have always been the physical
size of the array of cores. The plane, a mat of wires
with little teeny-tiny ferrite donuts at each intersection,
looks electrically like a lot of inductors all strung
in series. This is a good configuration for a low pass
filter that we are going to put a very fast pulse through.
The bottom line of all this is that even fast core
memory tends have around 0.490us write cycle times.
Read cycles are always twice the write cycle times.
This means you need to read the core memory specs
carefully. Be sure that you see the both read and
write cycle times. If you can only find one it will
most likely be the write cycle time.