Dear Ed,
I really like your site. I've been into electronics since I was a little kid. Over the years,
(I'm 63 now) I've amassed quite a collection of books and "legacy Items" (girlfriend calls it "my junk")
pertaining to obsolete technology, Bell System, Cold War, etc. Thanks to photos on your site,
I've identified a few of my items as likely components of Nike-Ajax systems.
I'm writing in reference to your Univac Solid-State page:
http://ed-thelen.org/comp-hist/univac.html
I worked on one of these machines just out of High School, around 1970 to 1974. This was in a
small town in southeast Colorado. A friend of mine's Dad made a deal for two complete systems.
These had originally been owned by Safeway Stores Inc., and removed when they upgraded.
We also got all the books, drawings, and support equipment such as key punches, interpreters,
sorters, collators, etc. The computers were made in 1959. We were able to get one system
working and used it to do card entry, tape storage and printout of inventory records. A few
years later, when a local utility began selling computing services, the data was moved over
there, and the old Univac was sold for scrap. They used a lot of gold contacts in those old computers.
The Univac Solid-State Computer came in two versions, 80 and 90 column. The 80 column systems
used the same type punch cards as IBM, but we had the 90 column version. The 90 column cards
were the same physical size as the 80s, but divided into an upper and lower zone, with 45
columns of round holes in each zone. Unfortunately, I never took any pics of the Univac.
Nowadays, "solid state" means, "no vacuum tubes," but back then the meaning was, "at least
one semiconductor." The USSC was very much a hybrid of tube and solid state technology.
Main memory was magnetic drum of 5000 "words" capacity. The drum was in a hermetically
sealed container filled with helium for cooling. It spun at around 15000 rpm as I recall.
The computer had a 3-phase 300 Hz alternator in the base, belt-driven by a 60 Hz
single-phase motor, that generated power to run the drum's motor.
A word was ten 4-bit digits plus a sign bit. Clock frequency was approx. 707 KHz, right
smack in the AM broadcast band. The "clock" was actually a CW transmitter using six
4CX250B tubes in push-pull parallel and run from a negative 1600 volt power supply.
Power from the clock operated the "ferractor" logic elements, which were similar to
tiny saturable reactors or magnetic amplifiers. Logic gating (ANDing, ORing etc. was
done by semiconductor diodes, I don't recall if these were silicon or germanium.
Jan Lindeboom has a very good site on a similar machine here:
http://www.ussc90.nl/
The biggest differences between Jan's system and ours were, his Univac had the three optional
index registers which ours did not, and our system had magnetic tape storage which is not
mentioned on his site. This almost doubled the physical size of the system. A Tape
Synchronizer provided the interface between the CPU and the "Uniservo II" tape drives,
and was as large as the CPU itself. Each tape drive was approximately 6 feet high x 2.5
feet wide x 3 feet deep, and we had four on the system. These read and wrote 8 tracks
on half-inch tape, as I recall. (6 data tracks, clock track and parity track)
We also had a paper tape read-punch unit on our system. It could read and punch up to
8 tracks on standard paper tape up to one-inch width. The machine's tape guides were
adjustable for smaller tape widths, and also had a patch panel so any tape track could be used for any function.
Thanks,
Dan Entingh
Denver, CO
see Picture of Grace Hopper with USSC
and its Uniservo tape drives
here
and
here
(links to Computer History Museum).
|