Computer history: Electronic and mechanical engineering (1986-1989)


Computers for electronic and mechanical engineering

In the early seventies, the Physics Laboratory TNO started with computer tools for the design of electronic circuits and the layout of printed circuit boards with a CALMA system. Between 1984 and 1986, a large amount of money was reserved for the renewal of Electronic Computer Aided Engineering (ECAE) facilities.

The CALMA system: server and work station
The CALMA system: server and workstation

To produce those PCBs, TNO-FEL operated its own complete PCB factory which produced many kinds of PCBs with different dimensions. However, designing an electronic circuit and designing the corresponding PCB requires two different kinds of expertise. The electronic designer is only interested in the function of the circuit, while the PCB designer is mainly interested in the realisation of the circuit on a PCB: the correct placement of the components on the PCB and the generation of the printed wiring.
Until 1986, electronic design was purely manual work. Sometimes an experimental circuit was built on a breadboard, which operations were then verified by measuring the signals. The design was adjusted until the circuit worked well.
The design of the PCBs was largely manual work, although use was made of computer support from a Computer Aided Design and Computer Aided Manufacturing (CAD/CAM) system. The Physics Laboratory used the aforementioned CALMA sign system; the LEOK operated two digitisers connected to a PDP 11/44 system. These two systems provided support to the PCB designers and controlled all kinds of peripherals (printer, plotter, photoplotter for negatives and a milling/boring machine); see, among other things, the story of our self-manufactured COBRA.

Sieb & Meyer NC controller with a Schmoll Machinen drilling and milling unit (1977)
Sieb & Meyer NC controller with a Schmoll Machinen drilling and milling unit (1977)

Computer Aided Electrical Engineering supported the designer in making a logical design, ‘the electronic circuit’ using a library with existing and self-conceived electronic parts. The operation of digital and analogue building blocks could be simulated on the computer in a way that the design could be ‘tested’ and improved without using a breadboard. As soon as the design met the set specifications, a physical design could be made in which the logical building blocks were converted into existing components such as ICs, resistors, transistors, capacitors, and so on. Thereafter, a suitable PCB in size – sometimes even round or with a hole – was chosen to place and wire the components. The designer then efficiently placed the components with computer help on the PCB. Thereafter, the printed wiring between the components was generated. Problems with the wiring had to be solved by adjusting the placement of components. The PCB could then be produced.

Around 1983, bottlenecks in the electronic design process occurred more frequently: the computer systems and software became obsolete, and the designs became increasingly complex. Modern systems offered more possibilities and were more user-friendly, faster and more efficient, especially for the electronic designer. Plans were made to replace the whole process and the supporting computer systems. Between 1984 and 1986, a large sum was reserved for the replacement of the laboratory’s Electronic Computer Aided Engineering (ECAE) facilities.
In a memorandum from the project group “FEL-CAE” (Computer Aided Engineering), it was proposed to purchase CADNETIX systems. Those systems were also at two other TNO institutes, the TPD and the Product Centre. An organisational structure was also proposed. The system experts were placed within the newly established group System Management and Support (Group 2-7). In 1987, the first CADNETIX workstations and a server were installed for schematic and PCB design. The Electronic Computer Aided Engineering (ECAE) system consisted of two CADNETIX (CDX) servers and four CDX workstations distributed over the ‘open shop’ where the researchers could make their basic designs and the ‘closed shop’ where the print production was prepared.

The Cadnetix systems supported the dynamics of the electronic circuits, such as checking whether the intended functions still operated correctly at higher clock speeds, whether the PCB had no crossed tracks and complied with the design rules, such as minimum track width and minimum distances. The system also provided the automatic generation of documentation of the designed PCB such as drawings of component
layouts, bills of materials, etc. For each design project, there was a database in which all information about the design was stored: scheme, simulation data, print layout and documentation.
A designer had to follow a course of a short week, after which a not-too-large design could be made within a couple of days. Simulating them could cost more or less time, depending on the simulated design complexity and any problems that occur. Making a PCB layout required about two days for an average design if the scheme designer had complied with the set rules. The production of a PCB on average took about three weeks.

The CADNETIX workstations had an MC 86020 processor, 40 MB disk storage, and 4 MB RAM, and ran under the BSD-UNIX operating system. The Electronic Computer-Aided Design (ECAD) systems for the development of Printed Circuits Boards (PCB) had proprietary graphics accelerator cards. For that time nice systems that were pretty fast! Very noisy as well. For Electronic Computer Aided Engineering (ECAE), a simulator engine was purchased that was little used due to the lack of symbols with specifications that the engine required.

CADNETIX later joined forces with SUN Systems. TNO acquired a SUN 3/50 and a SUN 3/60 (MC 86030 processor systems) for ECAE purposes. Because the CADNETIX systems were installed before the entire IEEE 802.3 local area network was installed at TNO-FEL (the FELLAN), a first separate Ethernet segment was installed for the CADNETIX systems. That network was later connected to the FELLAN using a bridge.

Although there were many problems with the operation of the application software, this user interface was very well thought out and implemented at all stages of electronic design. In 1990, CADNETIX ran into financial troubles. On December 24, 1990, CADNETIX was acquired by Daisy Systems. The new, merged company name became DAZIX.

ECAD work increasingly required more performance. Thus a SUN 3/60 was acquired having an enlarged chassis for accommodating the CDX proprietary graphical accelerator card. Although there were many problems with that software, the user interface was very good and supported the user well in all phases of his electronic design.

Controlling the Excellon drilling system for the printed circuit boards and the Gerber (acquired by Barco Graphics in 1998), and later the SECMAI photoplotter was a challenge for the system maintenance engineers. Output files from the ECAD systems were converted and adapted in various ways. When this link eventually worked, the own PCB production facility was shut down and moved to an external supplier.

The Excellon CNC system
The Excellon CNC system