rickb at bensene.com
Fri Jun 29 15:25:41 CDT 2007
> Billy wrote:
> Delightful machine wasn't it? To accomplish so much with so
> little logic.
> Those were the days when an engineer would be consider a
> genius and could get a big bonus for eliminating one flip-flop!
The guy that desined the LGP-30 architecture was indeed truly a genius.
His name was Stanley Frankel.
He was a physicist, but was extremely well-versed in almost all of the
sciences. Frankel was a key member of the team of folks that worked on
the Manhattan Project to develop the A-Bomb. His understanding of
nuclear physics, as well as extremely complex math, lent itself to
Frankel being a prime mover in developing the algorithm for the
"computers" (people with desktop mechanical calculators who would get a
slip of paper with a problem to solve, which they would do with the
calculator, then record the result, and pass the paper on) that "ran"
the nuclear diffusion calculations for the folks designing the bomb.
The "computers" were slow, and sometimes made mistakes. Frankel, along
with some other team members, came up with ways of making the
calculations cross-checked by others, so that the chance for error was
minimized. It was essnetially like programming a computer is today, yet
it's all done with paper, pencil, people, and mechanical calculators.
Later, to speed up the process, an IBM Card Programmed Calculator
(essentially an electronic math unit that was programmed with
plugboards, and used punched cards for input and output) helped speed up
the process, and again, Frankel was a major force in developing the
algorithms to make the calculations possible on this machine.
After the Manhattan Project ended, Frankel did consulting work. It is
during this time that he developed some computer designs. One of his
designs was for the architecutre of what became the LGP-30. He
presentated a paper on his design (which was indeed extremely elegant
and efficient), and attracted the interest of folks with Royal McBee
(later acsuired by General Precision, then Librascope), and a deal was
cut for Royal McBee to license the design for creation of an actual
computer, which became the LGP-30. Frankel worked closely with the
engineers to come up with the working design of the LGP-30, a machine
that utilized a magnetic drum for main storage, large arrays of
diode-resistor gating, and a relatively small number of vacuum tubes for
flip flops, buffering, and drive circuitry.
Later, Frankel ended up doing early electronic calculator design. He
developed the design of what became the Smith Corona/Marchant Cogito 240
and Cogito 240SR calculators, which were all-transistor, bit-serial
calculating machines that utilized a magnetostrictive delay line for
register memory, and an oscilloscope-type CRT tube to display results in
Smith Corona Marchant had an OEM marketing agreement with Diehl
Calculating Machine Co., in West Germany. Diehl was very successful in
the mechanical calculating machine marketplace, designing desktop
electro-mechanical calculators that were both extremely fast, and
virtually bulletproof. Frankel had designed a small-scale
microprogrammed logic engine in his home office (he called it Nic-Nac),
as a personal project. The idea was to make this "programmable logic
engine" the core architecture for a calculator. The folks at Diehl were
looking for a way to break into the blossoming electronic calculator
market, and learned of Frankel's idea through Frankel's work with SCM on
the Cogito 240/240SR. Diehl licensed the architecture that Frankel had
designed. Frankel, and his son, Allan, would travel to Diehl's plant
to help them with turning the microprogrammed concept into a working
calculator. A somewhat improved version of Nic-Nac was built in
Germany, which was then used as the development test-bed for what became
the Diehl Combitron calculator. The Combitron was Diehl's first
all-electronic calculator, and it was indeed a microprogrammed machine.
It had a roll of punched metal tape that contained the microcode for the
machine. The Combitron used two magnetostrictive delay lines as memory,
one which would be loaded with the microcode at power-up (from the
punched metal tape), and another that contained the working registers of
the calculator. This generalized architecture allowed the calculator to
have a very low component count for a complete desktop printing
electronic calculator with memory registers and square root. A later
revision, the Combitron S, added facilities for additional stroage which
allowed the machine to be user-programmable by "learning" calculator
operations into memory from the keyboard. The Combitron S was picked up
by SCM because of their OEM agreement with Diehl, and marketed by SCM as
the SCM 566PR.
Frankel received royalties for every Diehl Combitron, Combitron S, and
SCM 566PM that were made.
Just thought that perhaps some background on the genius of Stan Frankel,
and his contribution to electronic computing, may be of interest.
The Old Calculator Web Museum
More information about the cctech