19 Jul
2014
19 Jul
'14
1:15 p.m.
No memory pluggen in, the system complains, that it
didn't find any.
One memory module plugged in, Memory fault at fffefffc ...
(All other boards removed, tried different slots, different RAM Boards)
Interesting... It doesn't seem to be mapping the RAM on the CPU board
And as this backplane seem to be completely passive, my guess is, that
it is the CPU boards problem.
Thge backplaen that the I/O and memory boards go into is just connectors.
Thge motherboard, flat in the base of the machine, does have the keyboard
cotnroller and HPIB circuitry on it, but these should affect things.
Anyway I beleive you've got it workign with another CPU board, which
would tend to suggest the motherboard was not the problem (it may be...)
So my guess would be the transceivers to the bus (there is a whole line
of '245s)
My first thogut hwas the name of that old newsletter 'DTACK/ Grounded'.
As you know, the memroy boards you insett have their address selected by
DIP switches on the board, but there are no such swtiches for the RAM on
the CPU board. WHat happens is quite nice. At power-up, the machine runs
some code from ROM (this you are doing, the machien initialises the video
system, etc) and then starts looking for RAM starting at the top of the
address space.
When the first bus timeout aoccurs, maenaing no DTACK/ for that location,
U53a on the CPU board is cleared. This stores the address in U42, and
also enables the first block of RAM via U31a. On the second bus time out
the proceess is repeated using U53b, etc. And now bus timeouts become
BErr/ 's via U31d and U30c.
So, if there is never a bus timeout, meaning DTACK/ is alsways being
asserted by something (not necessarily tied to ground!), then the CPU
board RAM will enver be enaables. And accesses to RAM boards will fail,
they are not fast enough to complete amemory cycle.
So I would start lookign at DTACK/ and the various thigns that can drive
it.
-tony