ILL - Summary of control systems evolution

This is a subject which has always been open to management decisions based on politically sensitivities rather than practicalities.

Two early PDP11/20s were acquired for the nuclear physics group for the BILL and GAMS spectrometers, with a system (RTS) developed at the ILL by Ledebt. The principal control computers were chosen by a committee of one German and one Frenchman; each computer served up to 6 instruments, the CARINE (Télémécanique T2000) and NICOLE (Telefunken TR-86) systems.

The TR86 system was enormously expensive in hardware and programming effort, and very fragile. CARINE was reliable but very slow as a shared resource. Already by 1973 the inefficiency of existing ILL systems was clearly evident. E.g. the D8 diffractometer typically only counted for 25-40% of time, awaiting controlled positioning, nonetheless the instrument had to wait until 1979 for a PDP11/40.

This choice was odd and probably dictated by national preference since it was clear, even in 1970, that shared systems posed problems of fragility and maintenance.  At Harwell even in 1968 individual DEC PDP8 computers were being introduced after experiences with shared systems for Time of Flight (CASSANDRA) and 4-circle instruments (ANDROMACHIE brought to ILL for D15/D16). The Pluto triple axis hence had its own computer, since costs were falling.  A few years later at Saclay the instruments were controlled by individual Honeywell Mini-6 computers. Elsewhere individual DEC PDP8 and PDP11 computers were the norm eliminating many conflicts evident on shared systems.

As part of the "Deuxième Souffle" (1979) the CARINE and NICOLE systems would be replaced, and the Central Computer was to be upgraded. For instrument control there was strong support for the European industry, and seven SEMS Solar 16-40 computers (SEMS: Société Européenne de Mini-informatique et Systèmes, France) were acquired for D1B, IN1, IN2, IN3 etc. While the hardware outperformed the PDP11/34 systems chosen for the NICOLE instruments, IN4, IN4, D7, and PN1, the poor software support and specialised electronics interfacing required resulted in years of delay for the Solars. Only four of the seven systems acquired were ever put into service. There was little feedback from the scientists since at that time there were few data treatment programs running on minicomputers. The task of creating control functions required programming in close collaboration with the electronics engineers.

By 1983 at RAL (Rutherford Appleton Laboratory) individual DEC VAX11/730 32-bit computers were used from the start on the 8 or so instruments, all with the same acquisition electronics first on the computers UNIBUS; later this was attached through an SCSI interface on the microVAX. The greater number and diversity of instruments at the ILL always precluded a single solution.

In time these systems in turn were replaced by fully networked computers, and electronics too was configured and accessed through the network.  The simplistic description of instrument control and data treatment devices became blurred.