Corda Albert J DLVA wrote: 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] 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. 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. 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... I'll see what can be dredged up. It's been a *very* long time. 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. 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. 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(a)mainecoon.com http://www.mainecoon.com PGP fingerprint: 4E99 10B6 7253 B048 6685 6CBC 55E1 20A3 108D AB97