[vcf-midatlantic] Bendix G-15 was Re: Univac, our first real work!

[Brian L. Stuart]

On Sun, 6/26/16, David Riley <fraveydank at gmail.com> wrote:
> Agreed on all points. I'd be highly interested in working on that, if
> only I had more time... a drum emulator should be extremely simple
> to make in principle, the interface will probably be the hardest
> part.

I've done a little thinking about that in a different context, and
I think the best way to conceptualize the interface is to think
in terms of two separate power domains with the interfacing
being done with optoisolators.  The timing aspects of the interface
won't be as difficult as one might guess.  The main clocking
for the machine comes from the drum, so the whole machine
is slaved to the drum for timing.  As a result, we wouldn't have
to worry about getting the data rates exact as long as simultaneity
is preserved.

_If_ such a restoration were to ever be attempted, I'd suggest
approaching it as four tasks:
1) Develop a full emulation of the machine in such a way that
the drum emulation and the rest of the machine are connected
through the same interface used on the real machine.  For example,
using one Raspberry Pi to emulate the machine's logic and another
to emulate the drum.  That way the drum emulator can be developed
against an emulation of the machine and the major functionality
of it can be developed before attempting to tie to the real machine.
2) Restore the power supplies.  Of course, we're mostly all familiar
with that by now from other projects.  The good news is that tube
electronics of that era generally had relatively simple power supplies
when compared with the complex switchers we've been using the
last 40 years.
3) Set up a test bench with fixtures for testing all the tubes and
all the plug-in modules.  Then one-by-one go through every module
in the machine, replacing tubes and repairing modules as needed.
4) Restore the original drum.

Once Tasks 1, 2, and 3 are completed, we could integrate the drum
emulator and would have a running G-15.  I'd recommend doing Task 1
first as a proof of feasibility and competence.  Then if the machine's
owner were amenable, Tasks 2 and 3 could be undertaken concurrently
with relatively little risk.  Finally, only if it were decided to be worth the
risk would any attempt be made to touch the drum.  Of course, we'd
want to carry out any such effort using the same level of preservation
as the CHM PDP-1 project, documenting every step, preserving and
labeling replaced components, etc.

But I know what you mean about available time.  I've been meaning
to do an emulator anyway for some time, but other projects keep
getting priority.  What I should do is force myself to spend a Christmas
break doing an emulator.  Then if a restoration project were ever
approved, I could try to spend most of a summer when I'm not teaching
on it.  Of course, the same thing applies to restoring George.

BLS