Computers at the Laboratory for Electronic Developments for the Armed Forces (LEOK) until 1974
The initial work on digital techniques started at the Laboratorium Elektronische Ontwikkelingen Krijgsmacht (LEOK) in the System group ‘Data Handling’. The ‘state of the art’ of electronics in the world was characterised by valves and relays until the introduction of the transistor in 1958. The transistor could be used as a switching element.
In 1960, a number of project activities were completed at the LEOK. Research time became available to increase one’s own knowledge in this new field. In 1961, a start was made on researching the possibilities to apply digital techniques. Experiments were made with various versions of gate circuits, adder circuits, shift registers, and the like. Circuit components were used that were made available by other laboratories (e.g. Physics laboratory TNO (PhL-TNO of the RVO-TNO) and the Shape Technical Center (currently the NATO Communications and Information Agency (NCIA)).
In 1962, the experience gained in this way was used in a project for the Royal Netherlands Army. A ‘vehicle rolling road’ was developed for this purpose, which was suitable for measuring and recording different vehicle characteristics (speed, acceleration time, braking distance) of a military vehicle which could be used for calibrating other distance and speed meters. The information obtained from a pulse generator was processed in digital circuits composed of Philips circuitry. Many lessons were learned about ‘timing’-problems in digital circuits and how designs had to take these effects into account.
Improvements regarding the circuitry timing were fully implemented in a subsequent project, intended for the Royal Netherlands Air Force (KLu), which started in January 1963. The aim of this project was to automate the altimetry for Navigation station North-Holland, codenamed NSN, at Den Helder (note that the term navigation station was used for the deception of the actual operational objective of the station). The scale of this project, called Radar Information Processing Equipment (or RIVA in Dutch), was quite large for the laboratory, as it involved multiple system and technology groups. The processing of the radar information obtained by means of a video-integrator and the generation of control data for the VI antenna took place in a digital calculator of own design. This digital calculator had the following characteristics:
- wire programmed; special purpose
- one-address instruction code
- parallel processing
- interrupt facilities with priority levels
- core memory:
- 512 program words of 17 bits each
- 128 words ‘live’ memory
- 64 words ‘dead’ memory.
- clock frequency ca. 100 kcycles/s
- instruction times:
- add/subtract: 60 microsec
- multiply/divide: 420 microsec
The toroidal core memory was manufactured completely by hand. Fixed instructions and constants were hardwired. The circuitry used in the digital calculator was designed by the Physics Laboratory RVO-TNO. Variable program instructions were entered or modified using switches on the main control panel.
The RIVA project period can be seen as the start of software programming at LEOK. In that period, the first ‘general purpose’ digital computer was bought, namely the DICON (‘digital controller’). DICON was designed by Hollandse SignaalApparaten (HSA). The system had 192 address locations of 18 bits each. Each program had to be entered instruction by instruction using toggle switches. The coding was purely binary with a fixed comma after the main bit. Negative numbers were represented by their complement to 2. The machine supported the automatic conversion of Gray-binary for entered numbers in Gray-code.
The memory was variable: both a number (constant) and an instruction could be registered at any memory address (stored program). The program had to read instruction by instruction entered manually through keys.
The first research using the DICON had to do with the analysis of problems with digital servos. For that reason, the DICON was equipped with some synchro analog-digital (A/D)- and digital-analog (D/A)-converters.
The follow-up project of the RIVA project was the 3D-simulator project for the Royal Netherlands Navy (from 1965 to 1970). The simulator had to serve for the injection of simulated targets and clutter into the 3D-radar. The simulator was in use until the beginning of 1975. For the realisation of the simulator, a Ferranti computer was rented; this Hermes computer (‘germanium-logic’) was installed at the beginning of 1966 and was in use till the mid of 1967. This system can be regarded as the first real general-purpose computer on which LEOK programs were executed.
In March 1967, the SIMulation REKenaar (SIMREK), a Ferranti F2208, arrived as a project computer at the LEOK. The design was based on Ferranti’s F1600 series of computers, a modern ‘military, especially Navy’ computer at the time. The system had a memory of 8k words by 24 bits; 2 microseconds cycle time, and 16 interrupt channels. Delivery went according to plan, except for a few minor details, including a missing screwdriver. The SIMREK was equiped with a fast papertape reader for 5 and 8 holes papertape, a papertape punch for 5 and 8 holes papertape, and a Datamec magnetic tape unit (556 characters/inch; 45 inch/s).
This shortcoming prompted the Dutch Ministry of Defence to postpone the final instalment payment. When the misunderstanding was cleared up, Ferranti supplied with a smile a ‘golden’ screwdriver with a price tag of Dfl 152,927.–.
The Royal Netherlands Navy bought three systems and the (UK) Royal Navy had also purchased several systems including the SIMREK. The Dutch SIMREK system was modified by LEOK for use in the 3D-simulator. SIMREK was able to generate up to 31 simultaneous artificial targets for the 3D-radar. Each target had an adjustable course and speed.
At the end of the sixties, LEOK started to reserve funds to buy a computer system for a new to establish computer centre as part of the LEOK. The primary objective was to purchase a process control computer with the additional condition that there were sufficient support software and peripherals for program development and scientific calculations. After a market analysis, a system configuration based upon the Ferranti FM 1600B-computer was recommended on grounds of price, delivery time, quality of the instruction set, the organisation of the input/output, and the already available knowledge about the system based on the experiences with the SIMREK. After some delay, the contract was signed at the end of 1969. A computerless gap occurred between the end of the 3D-simulator project in 1970 and the acceptance of the Ferranti FM 1600B system. However, limited work could still be carried out at the SIMREK computer of the Navy in Den Helder.
Thus, the newly Computation/programming group had to limit their activities to the preparation of the installation of LEOK’s own Ferranti FM1600B computer. Fortunately, the RAREK (Radar computer) computer became available earlier than expected as the 3D-project ended and the system was made available to the LEOK.
In the summer of 1971, the LEOK’s own FM1600B was installed. The photographs below, taken during the formal acceptance of the Ferranti system, give an impression of the system.
The computer system consisted of the central processing unit, 16 Kbyte memory (later 40 Kbyte) and 22 interrupt channels. The following standard peripherals were connected to those interrupt channels:
- papertape punch reader and puncher, resp. 300 and 100 characters/s
- teletype ASR 33 (10 characters/s)
- line printer, 300 lines/minute
- quad magnetic tape unit, 9 tracks, 556/800 bpi, 75 inches/s
- plotter, 100 increments/sec
- operator panel
- (later) a Tektronix 4010 graphics terminal.
Compilers for ALGOL 60, FORTRAN II and CORAL 64 were available, as well as an assembler for FIXPAC, a subroutine library, and application programs. CORAL 64 was the NATO programming standard at the time, block-structured like Algol 60, for applications in real-time environments. The translator (compiler) has six passes! Calculations could be done in integer, floating-point, or fixed point (where the comma always ended up at another point). The advantage was optimal accuracy at maximum speed.
FIXPAC (FixedPoint Autocode) was the assembler. The instructions were based on three addresses, eg Va = Vb + Vc. The compiler had six passes! Calculations could be performed in integer, floating-point, or fixed point (where the comma moved around). The advantage was optimum precision and maximum speed.
At first, programs were entered from punch tape. The different compilation passes resulted in intermediate code which was output on punched tape. The system operation was done by means of the operator panel. Soon, the LEOK developed its own operating system to work from magnetic tape. After this adjustment, the bootstrap recognised magnetic tape as a boot device. We called that ‘BOS’. Then a larger operating system called EOS was loaded from the magnetic tape. EOS had a simple command structure and a very universal I/O interface. A self-written text editor, like that of the PDP 8, completed the whole operating system environment. Compilers were decompiled and changed to use the I/O interface (“I remember that Pim O. has been sweating for months on the Algol compiler, who then finally produced the legendary line “it is finished guys“). The whole system was operated from a Tektronix display station with a phosphor memory screen so that we got rid of the noisy teletype. We sold the operating system back to Ferranti: we got an extra block of memory. EOS was also applied with a hard disk drive instead of a magnetic tape within the Mech Lua Trainer project.
An important project which used the Ferranti was Torpeval (2D-phase). The information recorded at ships of the Royal Netherlands Navy was processed by the LEOK. The outcome was presented in the form of tables and plots. Printing and plotting of the Torpeval results required the writing of Torpeval in assembler code. Later, the ALGOL and FORTRAN compilers were adapted in such a way that the output devices could be addressed in a more high-level way causing programming to become quicker and easier.
The hardware was extended with the needed core memory to 32K (total) and a display terminal.
The Ferranti system was used for:
- Processing measurement data recorded at (remote) locations (e.g. calibration corrections for the submarines and mine hunting)
- System simulations
- Laboratory automation (technical documentation, production of printed circuit boards)
- Programming projects
- Software development of operating systems
Another project was to analyse the recorded tapes with ‘noise’ for the Royal Dutch Navy. The data of the analog recordings were first digitised and then sent via the Display Console to the SIMREK to save the data. As the SIMREK had only one magnetic tape unit, it was quite a task to ensure that the digitised data was well-organised on the magnetic tape in bits and pieces.
End of 1969, a 2400 baud synchronous terminal connection was established with the Control Data CDC 6400 of the Physics Laboratory RVO-TNO. Two TNO-colleagues of the Prins Maurits Laboratory had implemented protocol Mode4A (UT-200) software that simulated a card reader and printer. The package was implemented in Basic Plus, the standard language for the Resource Time-Sharing System (RSTS / E) operating system for PDP-11s. Basic Plus, incidentally, was a completely unstructured Basic derivate. It had statement modifiers: additional clauses placed at the end of a Basic statement. The kick was to write a program in only a limited number of statements, with one command spanning more than half a page. At the same time, plans were developed to couple the RAREK computer to the CDC 6400 system and to fit the Ferranti FM1600B with a time-sharing operating multi-user system.
Before that took place, the Ferranti was replaced by a DEC PDP 11/60. This configuration was extended later with a Digital Equipment Corp. PDP 11/44 and a PDP 11/34.
The largest LEOK project at that time was the Mech Lua (Luchtafweer = Anti-air) Trainer (MLT) (MLT) based on Ferranti FM1600B computers. In 1998, these training systems were still in use in Ede (they are ‘retired’ now). A TNO employee noted in early 1998: “It is incredible to see such a computer still working. It does not make a too retro system impression“.
Anecdote: Navy blue papertape
An important project required the processing of many hundreds of meters of papertape recorded on board of Dutch Navy ships. The Ferranti was equipped with an ‘ultramodern’ papertape reader, one with light cells. The problem was that the Royal Netherlands Navy used mechanical papertape equipment. They had bought cheap thin white papertape. The light produced in the Ferranti papertape reader, however, was so strong that the light passed through the thin white paper tape where it was picked up by the light cells. In short, the full width of the paper tape was recognised as ‘holes’. In the absence of papertape duplication equipment, there was only one solution: a few bottles of ink were emptied into a trash can, after which the paper tapes were manually fed through by the head of the system group. After drying (‘washing lines for papertape’), the Navy blue paper tapes could easily be processed.
Information on these pages is partially based on the publication ‘Gedenkboek “LEOK 1950 – 1975”.