18 Feb
2010
18 Feb
'10
6:04 p.m.
On 2/18/10, Jim Brain <brain at jbrain.com> wrote:
On 2/18/2010 3:58 PM, Ethan Dicks wrote:
I'll go see what I can dig out of the bin - I can easily throw the
Fluke 9010A on one and ensure the RAM and ROM test out OK. It's
harder to test out the bus interface, but unless you plan to stick
this on a Unibus, I don't think you care if the 8641s are all 100%
known-good. ;-)
The basic board is 8MHz 68000, 128K of parity DRAM (array of 4164
DIPs) on a 74S409 DRAM controller, 2x 28-pin JEDEC ROM sockets (2764s
at least, if not 27128s), a socketed COM5025 which could be a good
place to tap into the data bus and some select lines, an MC6821 with a
16-pin DIP exporting at least one of the 8-bit ports ("programmers'
interface"), some sort of sub-100Hz timer, IIRC, for heartbeat
interrupts, two 40-pin off-board BERG connectors (one for sync serial,
one for LP05-type printer, driven by the rest of the MC6821), and 12
edge-visible power and status LEDs (mostly serial status indicators).
I'd send you a link to a picture, but I don't seem to have posted one
(and the salad days of Software Results pre-date the Web). It's a
hex-height card that expects to pull +5V, +15V and -15V of of the
Unibus, so you'd have to either plug it into a DEC module block or
hack in power somewhere.
Naturally, I have all the schematics and PAL equations, but they are
in paper format, not electronic. Overall, my recollection is that the
memory space is divided by A23 and A22 into four 4MB quadrants, RAM,
ROM, I/O, and Unibus DMA engine. Retooling PALs could, of course,
alter that. Unfortunately, the PALs are not always socketed (depends
on the age of the board) and it's a six-layer PCB. I know of no
reason PALs couldn't be replaced with 16V8 and 22V10 GALs, but it
hasn't ever been tried.
Still interested?
-ethan
P.S. - the other models are variations on the theme. The older model
has 32K of 2114 SRAMs and two 6309 PROM sockets, newer models (of
which I have only a handful) have 41256 or 44256 DRAMs and Z8530
DUARTs, but are otherwise quite similar architecturally, if you ignore
the specifics of the host bus interface.
P.S. - this project revives, yet again, my
occasional interest in
hacking a COMBOARD into something less "embedded" - I usually get
stuck at the same stage - whether to hack in a 5380 SCSI chip or a
TTL-based "IDE" interface. COMBOARDs have serial, but no "disk", so
they'd need _something_ (they already have between 32K and 2MB of RAM,
depending on the model - I have piles of working boards with 128K of
DRAM and a COM5025 USART, smaller piles of other models).
Send me one and I'll hack in an SD interface complete with FAT support
:-) Heck, I could even put it on the serial port, if you didn't want to
hack it in.