GTE EA-22

The analog computer shown here is quite rare - it was built in the late 1960s/early 1970s by a small German company called "Goldmann Technische Elektronik", GTE for short, located in Ulm. Although next to nothing is known about this company, I recently found out that Mr. Goldmann once worked for Telefunken who produced analog computers as well (I always suspected this since there are some striking similarities between the machine shown in the following and some Telefunken made analog computers).

The machine is a very nice table-top analog computer (you need a quite sturdy table as the computer weighs in excess of 100 kg) with many design features making it even superior to my beloved Telefunken table top analog computers. The most noteworthy feature is that all modules are easily removed and are a joy to debug and repair.

The layout is classical: The top row of modules holds up to five modules, each holding four wire wound precision potentiometers. Underneath are ten computing modules which are completely interchangeable. Each such module corresponds to two precision, chopper-stabilized operational amplifiers (extremely similar to the Telefunken operational amplifiers). Below are the overload indicators for 22 operational amplifiers (two amplifiers are used for the timing unit). The bottom tow contains the necessary control elements for the computer.

The following pictures show details of the machine and were taken during restoration. The two pictures below show the right-hand side of the power supply and the two large connectors connecting the power supply with the rest of the machine.

The pictures below show the "backplane" of the power supply on the left and the connector to the power switch which is located on the front panel. This is a feature I particularly dislike: Although I fully understand that one does not want to route wires connected to mains anywhere near precious analog computer signals, I really don't like the idea of using a microphone plug/jack (!) to deliver 220 V to some part of a circuit. Since the original plastic wrapping has decayed subtantially, I rewired the jack and plug and covered everything in shrink-wrap tubing.

The circuit board shown below (front and back side) contains the two voltage regulators for the +/-15 V supply:

Associated with this board are two power transistor stages, one of which is shown below:

The high-precision +/-10 V power supply (these voltages are called "machine units" as all computations are related to this two levels) are derived from the board shown below. The strange black device in the middle of the left picture is a high-precision Z-diode which serves as the central reference in the computer. This stabilization circuit also requires two high-precision operational amplifiers which will be shown later.

This power transistor card drives the +/-10 V lines:

The circuit board shown below is the heart of the 400 Hz square wave generator which drives the electromechanical chopper relays and the synchronous demodulators which stabilize the operational amplifiers (and increase gain substantially at low frequencies - such an amplifier exhibits a typical open-loop-gain of about 10 ** 8 to 10 ** 9 - compare this to modern amplifiers... :-) ).

The rather big block shown below contains the push-pull drivers for the electromechanical chopper relays and the synchronous demodulators of the 22 operational amplifiers in the system.

The picture below shows an operational amplifier card. Before anyone starts laughing due to the sheer size, consider this: This amplifier is chopper-stabilized and exhibits a negligible drift over a wide range of temperatures (and time) and has an open-loop gain (at low frequencies) of 10 ** 8 to 10 ** 9 - not bad for a design which is nearly 50 years old. :-)

The picture below left shows the power supply on the workbench during its restoration. If you attempt this be aware of the fact that the all important +/-10 V stabilization circuitry won't work without the relay visible on the lower left energized. This closes the servo-loop so that the internal reference derived from the Z-diode shown above is used. When the relay is unenergized, the system can be remote controlled - this was normally used when two or more of these analog computers were tied together to tackle complex problems.

The picture on the right shows the operational amplifier drawer mounted in the top of the computer chassis. All in all, 22 operational amplifier cards can be inserted here. Two amplifiers each share a common electromechanical chopper made by "KACO", driven by a 400 Hz square wave signal. These choppers are visible on the bottom of the picture.

The following pictures show the computing modules. The pictures below show the front and back of the dual comparator module. Note the intricate mechanical structure holding the high-precision computing resistors at the bottom of the left picture.

Shown below is the dual integrator module. The large silver blocks are high-precision capacitors developed for use in analog computers (Telefunken used the very same model). The relays visible in the lower right of the left picture are used for the mode-control of the integrators (initical condition, operate, halt).

The pictures below show the front and back of the dual arbitary function generator. Functions (of a single variable) are approximated by polygons with equdistant supporting points. The slope of the polygon at each supporting point is set by a potentiometer. To set these function generators, an extender card is necessary which I, unfortunately, don't have.

Shown below is a double multiplier. The multiplication is based (as often) on the so-called "quarter square"-method: Instead of implementing a complicated function of two variables, this method only requires the computation of some square functions and relies on the fact that (x + y) ** 2 - (x - y) ** 2 = 4 * x * y.

To set the function generators one needs a voltage source generating output signals of +/- 1, 2, 3, 4, ... 10 V. This source is shown below: It is remarkably simple - since operational amplifiers which one would use today as an impedance converter were prohibitively expensive back than, a low resistance voltage divider chain was employed. These are the red resistors located behind the rotary switch.

The pictures below show the central control module. It is used to select the operating mode of the computer etc. Normally it would hold either an analog voltmeter or a flat DVM. Unfortunately this DVM is missing from this module, that's why there is a hole in the upper left corner of the front plate. The black and red jacks located above the row of switches on the left can be used to connect an external DVM which is necessary to set the (unbuffered) coefficient potentiometers of the computer.

The most complex single module is the timing generator. It controls the mode of operation when repetitive or iterative operation has been selected. As can be seen, this module employs yet another high-precision capacitor as those used in the integrators. In fact, it contains an integrator which generates a voltage ramp which drives a comparator which in turn generates the necessary control/reset signals for the computer and the timing generator itself.

A common problem with all analog computers is the (central) patch panel. In order to use the analog computer on various different problems with minimal setup time between two problems a removable patch panel is required. Naturally, this involves literally hundreds or even thousands of individual connections. GTE solved this problem quite elegant: The individual computing modules all have standard 4mm jacks mounted on their respective fronts. A basic configuration of the computer just works by using these jacks. If a customer wanted a removable patch panel (a feature especially important for users in universities where frequently different programs had to be set up), he could insert contact pins with attached springs into the jacks of the computing modules as shown in the pictures below. These springs then make contact with a removable patch panel containing the same complement of jacks and the same designations as the underlying computing modules.

The following two pictures give an impression of the restoration process. The left one shows my beloved wife Rikka who took care of the patch panel cleaning (a horrible task, to be honest). The right picture shows the various computing modules remove from the machine prior to cleaning and repair.

12-JUL-2015, ulmann@analogmuseum.org