[sci.electronics] QUADRAM MEMORY

abc@adm.BRL.MIL (Brinton Cooper) (06/01/90)

I purchased a QUADMEG-AT (128K) memory board for an old-ish IBM-PC/AT.
The board was designed for the 4.77 MHz machine.  However, it turns out
that this machine contains a 12 MHz crystal.

  (To justify my stupidity, let me point out that this machine belongs
to a nearby school for whom I'm doing a 'volunteer' project.  I planned
to donate the board.)

The specific problem is that the machine does not "see" the memory on a
cold boot; it gives a diagnostic of the form:

          080042 AA00 Memory Error 201

which appears during memory check after machine has indicated that 512K
is OK.

The (64K) memory chips in the QUADMET board are 150 ns Fujitsu chips
(MB8264A-15);  these are 16pin chips.

For those of you who know this product:  

	1. Can I get reliable operation by changing the memory chips?

	2. Is it only the chip speed that is likely to limit things?

	3. I have only a JAMECO catalog here at the moment, and the only
faster chip of the correct configuration is a 100 ns model.  Is
something even available that will work?

	4. If nothing will make this board work reliably in this
machine, what am I offered for the board?

Other advice appreciated, including pointers to boards that will work
and that don't cost an arm and a leg.  I'll entertain buying a used
board from you.

In view of the large number of articles posted to these groups and the
probable lack of general interest in this topic, you might wish to
e-mail your responses to me.

_Brint Cooper
H: 301-879-8927
W: 301278-6883
<abc@brl.mil>
-- 
_Brinton Cooper		BRL - Where "Research" is our Middle Name.

smasters@gmuvax2.gmu.edu (Shawn Masters) (06/02/90)

In article <23502@adm.BRL.MIL> abc@adm.BRL.MIL (Brinton Cooper) writes:
>I purchased a QUADMEG-AT (128K) memory board for an old-ish IBM-PC/AT.
>The board was designed for the 4.77 MHz machine.  However, it turns out
                                ^^^^
>that this machine contains a 12 MHz crystal.
                              ^^ 
The cycle time on the trubo XT's was 150ns.  The original PC used a time
of 200ns(64-256k motherboards, 250ns for 16-64k motherboards).
You might be able to squeak by with 120ns chips on a 12 MHz board, but I
have had the least trouble on this type of system with 100ns chips(plus
the higher speed chips are more reliable due to higher manufacturing
constraints).

>
>The (64K) memory chips in the QUADMET board are 150 ns Fujitsu chips
                                                 ^^^
These will not work unless the board has way of holding the memory wait
line on the bus(unlikely).  You can do some quick work with the
calculator and see why.

>	1. Can I get reliable operation by changing the memory chips?
>
>	2. Is it only the chip speed that is likely to limit things?
>
Since this is a board designed for the XT bus, and you are using an AT
there may be problems.  I'm looking for a description of the AT bus
myself right now, but I don't know enough right now to know if the buses
are incompatible for memory expansion units(not just the bus is at work
here I guess.  What about AT bios and other AT cards?).

You can buy 10-12 MHz boards now for around $100(0k), so if you need the
memory and your going to buy it anyway, maybe it'll be cheaper just to
buy a new motherboard then one of these $200-400 memory expansion board.

Shawn Masters
smasters@gmuvax2.gmu.edu