Emulators of Classic Computers

From: William Donzelli <aw288_at_osfn.org>
Date: Thu Jan 22 21:01:45 2004

> THe electronic design was abandoned for now. Basically I was to
> (will...?) build a 4 x 4 switched cap memory which will be a nice
> reality check on flops, tubes, etc.

Please keep us informed.
 
> Also, I chased down and read the exhaustive life-testing book the US Mil
> sponsored (forget the title) in the early 60's re: tube reliability.
> Though it applied specifically to subminis, it quantified what the
> computing people knew: derate, derate, derate. They got easily 16,000
> hours from selected devices.

Subminis might also be a good choice, if you can find them. They "never"
go bad. I don't think I have ever found a bad one. Their reliabilty was
due to production in a clean room environment using very high purity
materials. One interesting side effect was that submini tubes lost their
sockets - the things were soldered right into the circuits, as the sockets
became the most failure prone components.

Most computer tubes were also built in the same clean room environment,
and also tended to be very reliable. In the good old tube computer days,
there would be a flurry of tube swapping for the first few months, maybe
a year, of the computer's life, but then as the weakling runt tubes were
purged, the machines became quite reliable.

> The suspected more was easily possible, but
> that was the extend of their testing. It was interesting that some
> devices worked better/longer at slightly elevated Vff.

The biggest advancement was the use of the IERC tube shields - the type
that have ribbing inside to conduct the heat away. It took ten years, but
the industry finally figured out the old style tube shields were a really
dumb idea.

> Also, if you look at general tube electronic environments throughout the
> 50's, excepting the smartest stuff (Tek scopes etc) the power supplies
> are total crap. Unregulated and unsequenced. That did a lot of harm. It
> was fine for radios but not for "high speed" switching where margins
> started to matter.

The military radar things, specifically for IFF systems, had good power
supplies. They were digital system, after all (specifially SIF encoders
and decoders).

William Donzelli
aw288_at_osfn.org
Received on Thu Jan 22 2004 - 21:01:45 GMT