This page focuses on the 18 bit Navy computers which were eventually replaced by 16 bit computers in the 80's.
The first unit isn't illustrated on the genealogy chart, i.e. the 1218 brass board was called the Computer Unit Tester (CUT) , whose logic design was created
using circuit card types from the CP-642A in order to effect more efficient testing of UNIVAC commercial and NTDS peripheral equipments.
An innovation of these units were the Input/output (I/0) channels. They could be configured to
operate in a normal 18 bit length or an odd/even pair could be
ganged into a 36 bit I/O channel.
The Defense computers were programmed using the Trim III assembler while the commercial software was developed with a COBOL compiler. [lab]
According to the Univac Products St. Paul, Volume II printed
in 1961, the 4 was because of a 4 microsecond memory speed and
the 18 because of the digital word length. The 18 bit machines discussed hereunder were a 'commercial-to-military' ISA transition.
The series was the Computer Unit Tester, the commercial 418,
and then the 1218. The 418 commercial line became the 418 I, 418 II, and 418 III as new technologies were incorporated.
This genealogy clip from the 60s shows the commercial units in red, the defense in blue and a couple of airborne units in yellow.
3. Computer Descriptions
There were about 950 computers delivered with the basic 18 bit architecture - not counting any commercial 418 series computers.
The basic Instruction Set Architecture for this series came from the Computer Unit Tester (CUT),
a laboratory unit initially assembled using printed circuit cards from the CP-642A. Manufacturing test engineers need to be able to test the
interfaces of peripheral devices with a variety of word lengths; the 1100 series 36 bit machines down to the 6 bit 1232 I/O Console.
The CUT characteristics were 18 bit instruction, 8 microsecond 16k word core memory, eight duplexer I/O channels -
Type A per DS4772 (-15v) with a rate of 62.5 Kwords per second. The unit cabinet took 25.5 cu. ft, weighed 900 lbs, and consumed 1000 watts.
[written by Lowell with inputs from Don Mager.]
3.1 The 1218 (CP-789)
The CP-789 became the core of the Navy's on board Management Information System to keep track of all the logistics
such as ammunition, foodstuffs, etc. A variation with more memory was designated the AN/UYK-5 (Moonbeam) that used special software
to provide on ship logistic management. [lab]
1218 Characteristics - project start October 1962:
18 bit word size, memory capacity from 4k words to 32k words with a 4 microsecond cycle time. Data size was either 18 or 36 bits [i.e. dual word]
There were 98 instructions with a basic add time of 8 microseconds. A maximum of 8 I/O channels with -15V interface were available.
AN/UYK-5 ( )
A version of the 1218 was sold to the USMC for their 'Landing Approach System-3', the first delivery was Oct. 23, 1965. [lab]
A system - Based on the Univac 418 design, the 1218 came with COBOL, very necessary for the software development. This system used for inventory control
was developed for the AFS-1 through -7, the MARS Class Combat Stores Ships. John Markfelder, Univac headquarters marketing, Skip Wren in D.C. were the primary customer contacts.
The system consisted of the 1218, a 1232 paper tape and typewriter unit, the 1240 magnetic tape unit, and the Card Reader, Punch Interpreter (CRPI) unit.
The CRPI was the 'old' John Bull taper pin punch/reader which had previously been used with the Mod 0 file computer.
Based on user feedback, the unit should have been called CRaP because the I portion didn't work very well. [Lyle Franklin]
Hi Lowell: I was the program manager at DCS/Air at HQ-MC responsible for introducing the U-1500 into all Marine Air groups.
After I retired from the USMC, I assisted John Markfelder in the UNIVAC deliveries to the Marines as well as responding to their customer problems world wide. [Earl Lillestrand]
From Ron Q. Smith Subject: UNIVAC 1218 in Australia - A UNIVAC 1218 was used at the Honeysuckle Creek tracking station.
The CP-848 became the base computer for TARTAR, TALOS, and TERRIER missile launching from ship installations.
The government nomenclature for those systems was MK-152. These units required a motor generator for power. [lab]
1219B Characteristics - project start July 1967:
18 bit word size, memory capacity from 32k words to 65k words with a 2 microsecond cycle time. Data size was either 18 or 36 bits [i.e. dual word]
There were 102 instructions with a basic add time of 2 microseconds. A maximum of 16 I/O channels with -15V or -3v interfaces available in four channel groups.
First delivery was March 1968. [lab]
As the first CP-890 was in initial debug and test, a small team put together a proposal to use the card and
cabinet packaging style with a 1 microsecond memory to develop a next generation 18 bit computer, tentatively to be
identified as the UNIVAC type 1236 (1219C). This was a
'concept' computer, never designed nor built. The target market was the Sea Sparrow, to be bid exclusively with
Sperry to Navord. The characteristics were to be either 15.1 or 11.6 cu. ft, 900 or 550 lbs, 2330 or 925
watts depending up the 131k memory or 32k memory capacity.
Lyle Franklin of Marketing, Ernie Lantto of
Engineering and Curt Christensen of Program Management were the principals of this proposal. Raytheon won that
competition. As part of the debriefing, Lyle Franklin told NAVORD: 'Perhaps I'll come back with a 16-bit computer
proposal.' A prophecy which came true a year later when the 1616, UYK-15 was proposed for the ITT
Cutter' application. [Item written by Lowell with inputs from Lyle Franklin and Curt
ILAAS was an airborne 18-bit machine using mechanical
technologies from the CP-823 30-bit computer. The
characteristics were a 2 microsecond memory in 4k to 32k
words. This unit had only 27 instructions with 3 index
registers mapped onto core addresses. It also had just 9
interrupts and 1 Real time clock. It had an Assembler and
Utility Package. The Instruction Set Architecture differed
significantly from other UNIVAC machines of the time in that
bit 0 of the registers was at the left and bit 17 at the
right. There were 5-bits for the instructions versus 6 bits in
other machines. there were just 2 bits for index registers
versus 3 bits.
Project start was March 1966, first delivery in May 1967. Gary Bosworth, Barbara Halvorson, and Lee Sheldon were engineers on this project. [lab]
3.5 CP-914, 1819
The CP-914 was another 18-bit airborne computer - using Alex Trumble Relay (ATR) chassis
mechanical packaging similar to the 1830 Phoenix. These units had a 2 microsecond memory, three index registers and a real time clock. Two
of these units were sold to Sperry Flight Systems, the first delivered July 2nd of 1973.
Project start was August 1968, first delivery in February 1969. [lab]
This was/is the Minuteman missile launch computer.
These were mounted in radiation hardened operations rooms and connected to missiles. These machines ran continuously running test programs
to assure the readiness of the missiles for launch. Because of their design and benign environment, some of these units operated for over
5 years seven days a week, 24 hours per day without experiencing a hardware failure. This computer model had a unique red/black separation
Input/Output section design to separate classified targeting information from status reporting channels.
This unit used the same basic 18-bit instruction set, however it was mapped onto a 36 bit hardware structure. It also used a
plated wire memory because there was a radiation hardening requirement that core memory and mated film memory couldn't
meet. There is more about plated wire in the Engineering, Memory section. Project start was May 1969,
first delivery October 5th 1971. [lab]
From: Wes Peters - 19 January 2009 via web site e-mail. Message: The 18-bit
computer page doesn't mention the FORTRAN IV compiler that was used to develop the Minuteman Operational Targeting
Program as a part of the Rapid Retargeting upgrade for Minuteman II and III, and later used for the Peacekeeper
Operational Targeting Program when Peacekeeper missiles were installed in Minuteman silos at F. E. Warren AFB. The
Fortran compiler was hosted first on Univac 1108, and later on DEC VAX/VMS hosts. I worked on a static analysis
package for this compiler as part of the MOTP ARSIP upgrade when I graduated college in 1984, and later was the
developer test lead for Peacekeeper Operational Executive Program, the operating system that ran on the AN/UYK-11 in
the Peacekeeper Launch Control Facilities.
Feel free to contact me if you'd like more details about the swan song
of the AN/UYK-11, or as we affectionately knew it, the "Wicked Slow Computer."
3.7 The 418 Computer Series
We received 418 and 1219 computer comments from Montana, New Jersey, and Sweden then put together an 18bit computer '
Article for the Month'. When he read the article, Ben Manning wrote "
Hey Lowell, I worked on 418 III's for 4 or 5 years, in the UK on a system for Joseph Lucas and then in Bluebell on the LEAP Law
Enforcement system. I think I finally junked all my archival stuff during our last move on Dec 2014. I had all
the system documentation, plus the instruction card for the 418 III but it is probably gone.