The name "Dornier" is normally linked with advanced aircraft design - with airplanes like the DO-31 VTOL, etc. Not many people know that the very same company built wonderful analog computers which outperform most systems of other manufacturers by far. The system shown in the following, the DO-960 was Dornier's top system.
Dornier started the business of building analog computers when they got into VTOL design (namely the already mentionedDO-31) - the mathematical problems posed by these aircraft made the development of advanced computer systems necessary. Since no digital processor then (and maybe even today) was able to solve all of the differential equations necessary to describe the airplane in detail, Dornier started building analog computers.
The DO-960 was the last system manufactured by Dornier (as far as I know). It is a hybrid system, i.e. it consists of a complex analog computer as well as of a digital computer system which is interfaced with the analog part by lots of AD- and DA-converters, etc. The digital system controls the timing and read out bus selection of the analog computer as well.
When I got the system it consisted of two racks mounted next to each other. Unfortunately, only the rack on the right contained analog computing modules while the digital parts were spread all over both racks. Worse, the digital control system was designed to be controlled by a Data General NOVA computer system which was replaced by a PC based solution many years ago. Since I did not want to bother with a flaky PC with a 20 MB hard drive and with lots of undocumented software I sought a way to get around this PC solution.
After reading lots of the documentation that came with the system it became clear to me that Dornier had two main versions of the DO-960 on the market: An entry level system with a digital processor with ROM based BASIC interpreter and a high end system with was controlled by a NOVA (or a PC as in my case). Fortunately I found everything which was necessary to "downgrade" my machine to the ROM based version in a large card board box and started to lay out a plan for the conversion.
First of all I decided to get rid of the second rack - since it did not contain any analog electronics I was sure I could fit all of the necessary components into a single rack. This task was more difficult than I had anticipated. The first problem was the central grounding block - a massive piece of copper mounted between both frames. There were literally dozens of ground cables ending there from all over the system. Disconnecting these and separating the two racks physically took nearly a day.
The next problem to be solved was the fact that most of the connections between the digital and the analog part of the computer had been removed years ago and the documentation was not very clear about the cabling of the overall system. To figure out which module got its signal from where took another two days. I started to feel quite familiar with the system after having it taken apart to the last screw. The next task I got into was getting the digital processor up and running. The digital processor is based on a CAMAC-crate (these were very popular in the instrumentation business some years ago).
Since I stripped the machine down from two frames to only one frame I had two CAMAC crates so I had one for spares which proved to be a gift from heaven since one of the bulky CAMAC power supplies was broken and I did not want to debug this, too (it has an intermittent error - the + 6 V line has drop outs which I could not find easily). Getting all boards in the proper places and making a cable for the current loop serial line interface to the control terminal took another two days. After this was accomplished, I had a CAMAC based computer with a BASIC interpreter in ROM which was missing its analog counterpart.
Getting the analog part of the system up and running was quite difficult, too. Some of the modules were broken, I had a small fire in the backplane due to two shorting PINs in a VG plug, and I had a mean short circuit in the +/- 10 V lines - more about this later.
After nearly two weeks, the system was ready to solve its first problem after many years in storage - some pictures of this may be found later. The following pictures give an impression of the system and its components.