Corda Albert J DLVA wrote:
> The unit seems to be in pretty decent shape (apparently it did not
> go through "Military" (i.e. sledgehammer) decommissioning,
> as the party I bought it from indicated that he obtained it from a
> NASA surplus auction). Unfortunately, there were no docs
> whatsoever with this critter. I seem to vaguely remember that
> some of the Rolm systems were just repackaged and beefed-up
> Data General Novas, but I could be wrong.
Yep, you're wrong :-) With the exception of one model (of which
a grand total of three were built) ROLM never did "punches"; the
machines were always of their design and frequently had architectural
extensions.
The 16XX series are post-nova 800 with stack support, but in
an utterly different and incompatable fashion from that
found in the Nova 3. I don't recall if it had 64K extended
mode a la Keronix and DCC, but I wouldn't be surprised.
A consequence of this is that not all Nova code will run on
the 16XX; just as DG reclaimed some meaningless instruction
codings (which could be used as NOPs) to implement the Eclipse
instruction set, ROLM claimed some for their extensions. The
exposure to problems, however, is probably quite small --
less for an early machine like the 1602 as opposed to the
later 1666B.
[snip]
> Some of the things I'd like to know (or find pointers to):
>
> Hardware interface pinouts/docs:
> i.e. where do I connect an ASCII
> terminal... (There is no obvious serial connector,
> i.e. DB25, etc. All the connectors are some
> sort of mil-std twist-lock jobs)
Yep, and not all of them are fixed-function. J1-12 on
one machine might well be something utterly different
on another. I'm not positive where the basic I/O stuff
comes out, but I may be able to find one of my fellow
ROLM alumini who does.
> Does this critter have a disk interface of
> some sort, etc.
It might -- or might not. It depends on the application
that the machine was used in; a nontrivial number of
the 16XXs were employed in diskless applications
running a purely memory-resident operating system. A
very large number of these were used in the GLCM/SLCM
erector and launch control systems. If the machine
*does* have a disk interface it's likely to be a
diabloesque interface to one of ROLMs semi-proprietary
drives.
> Power requirements and pinouts:
> The previous owner had an AC line cord
> attached to a connector on the back,
> but I don't necessarily trust that he knew what
> he was doing. Is this really capable of running
> at 117V_at_60Hz, or did it need something oddball
> like 400Hz?
It depends on how it was optioned. Most will accept a
fairly wide frequency range on input. Unlike some other
manufacturers, ROLM built a single machine for all branches
of the service. This in turn caused some other problems;
hardware capable of withstanding the Navy's sledgehammer
test was usually on the heavy side as far as the air force
was concerned and the Navy really didn't care if the
machines operated to 60,000ft...
> Instruction set documentation:
> So I can play with the front panel :-)
I'll see what can be dredged up. It's been a *very* long time.
> Software:
> i.e. Did this thing have a simple executive program of some
> sort?
Not in the sense of something in ROM. ROLM software for the thing
included ARTS ("advanced real time system") and some memory-resident
thing whose name escapes me. Much of the latter stuff was written in
a C-like language called MSL, which tried to dodge around the problems
inherent in byte pointers on the Nova.
> I downloaded Bob Supnik's Nova emulator in the hope that
> it might provide me with some hints as to Nova architecture,
> but there wasn't much documentation there. Will a Rolm 1602 run
> DG Nova code? If so, is there an archive of DG Nova
> software somewhere?
It should run most Nova code; it will be things like MMAP controls
that might break. As for a Nova archive -- not that I know of,
although it's on my to-do list.
> The back of this critter uses a bunch of what looks like mil-std
> twist-lock connectors. Does anyone know an (affordable)
> source for these? I'm going to try pulling it apart tonight
> (It appears to be held together by about 17,000 screws :-).
And you'll find about 17,000 more inside. Each board has a
sealed metal cover; if you peel that off you'll find the chips
mounted so they straddle metal rails. In order to meet fungus
and salt-spray (and to a lesser extent, altitude) Mil Specs
the cases are gasketed and some were pressurized with
dry N2 -- so obviously blow through cooling won't work. The
metal frames and the metal lids comprise the so-called "thermal
frame" which conducts heat to the sides of the ATR chassis, where
in most machines they attached blow-through heat exchangers.
Boy, I haven't thought about this stuff since 1984, so I may
be off on some of it...
--
Chris Kennedy
chris_at_mainecoon.com
http://www.mainecoon.com
PGP fingerprint: 4E99 10B6 7253 B048 6685 6CBC 55E1 20A3 108D AB97
Received on Mon Jun 05 2000 - 16:57:35 BST