150uf, as I did not have anything else of suitable value handy, and put the card
back into the case. This time the POR worked perfectly, and it has done ever
since. (It should be equally valid to increase R13 to say 12K/15K.)
I have not analysed the action of the POR circuits of the N2, but it seems
that the value of the time constant of C3/R13 can be critical. The time constant
C2/R11 in the manual reset circuit is identical in component values. In
consideration of the ‘cure’ for this fault it seems that C3/R13 should have a
longer time constant than C2/R11 for correct circuit action. By the laws of
averages, this should be true for 50% of the Nascom 2’s with 10K/68uf fitted in
both positions, allowing for component tolerances. In any case, if this
assumption is true, the fault will only be noticed in a proportion of those N2’s
that are trying to reset to some other address than 0000H.
I am not an expert in digital logic. Perhaps someone can send in an
article explaining the operation of the circuit, with some timing diagrams, as my
cure may not be the best one for this problem. It does work though.
I would like to thank Alan Stevens and Phil Harvey for the loan of cards,
which helped me to isolate the problem. I hope that this article may help others
who have experienced similar problems with the power-on-reset on their system.
LARGE RAM SYSTEMS USING THE MAP CARD
by Richard Beal
This note refers to the
review of the MAP 256K RAM card
on pages 40 to 44
of Vol 2 Issue 1 of 80-BUS News. A problem has arisen when using the Gemini
CPU/64K RAM card with more than one
card. The reason is that the
GM813 does not provide an A19 signal – see
“The Great A19 Debate”
on page 43
of that issue for a detailed explanation. This means that the 64K of RAM on
the GM813 is addressed both at the bottom of the one megabyte of addressable
memory, and also half way through. This obviously causes problems, since if
two 256K MAP cards are added, the last 64K of the second card is overlaid by
the bottom 64K on the GM813.
The solution is to disable the 64K on the 813, and alter the header plugs
on the MAP cards so that they provide memory starting at 0K, 256K, 512K and
768K. This is very simple and MAP will happily tell you how to do this.
The best way to disable the memory on the 813 is a simple hardware
modification to permanently disable the RAM. MAP will provide a recommended
alteration. Alternatively, the 813 RAM could be addressed on a different 64K
page, using the original paging scheme which has four pages of 64K. This has
the disadvantage that as well as the GM813 needing modifying, a patch to RP/M
is needed, which makes it nonstandard. I don’t approve of that, but MAP may
prefer to adopt this solution.
You may feel that it is incredibly wasteful to effectively throw away 64K
of RAM, but if you have a system with 512K or more, you probably don’t mind!
Note that with Gemini’s own
512Kbyte RAM-DISK board you can go upto
8 Megabytes with no mods. to either the GM813 or GM833!