We have mentioned the provisions made to provide back-up power systems involving generators and banks of batteries (see history of the computer – redundancy). You can probably tell from this that we are talking about a lot of power here.
Power means heat, and in the exacting world of the computer, excessive heat can cause all sorts of problems. Manufacturers had very close tolerances for operating temperature and humidity.
In the early days of vacuum tubes, electronic circuits were more tolerant of heat, most components could be designed to operate at high temperature. For instance a resistor designed for one quarter watt dissipation in a circuit today could have been replaced by a 10 watt resistor of the same value in the tube design, and have the same effect in the circuit. Of course the 10 watt resistor is physically much larger, but space was not as critical as in today’s circuits.
The problems of heat generation were, however, very apparent in the vacuum tube computer. The vacuum tube works by heating the Cathode so that it will emit electrons. The Cathode has a negative charge, while the Anode has a positive charge.
Due to the difference in voltage, of some hundreds of volts, the electrons are attracted to the Anode and can flow freely in the vacuum. This flow is regulated by one or more grids placed between the Cathode and Anode. The grid has a slightly negative bias compared to the Cathode, and can be modulated, or altered, to control the electron flow, and thus the current.
For example, in an amplifier, a triode (a tube with three electrodes, Anode, Cathode and Grid) can be made by applying a varying signal, perhaps from a vinyl disk, to the grid. Its small variations in amplitude, or size, are amplified in the current flowing through the tube, usually measured across a resistor in the Anode circuit.
But we digress! Back to the computer. In the computers using vacuum tubes, they were usually used as a switch, on or off, 0 or 1, in tune with the binary system. This was easily achieved by applying a negative voltage to the cathode to turn off the tube, or a more positive one to turn it on. This arrangement works well in circuits like flip-flops and their derivatives.
But – there’s always a but – due to the vast number of circuits required, with each tube generating heat to work, the cooling problems were huge. Large blowers and cooling fans around the tubes, as well as room air conditioning were standard. Also liquid cooling was used.
When transistors came along, in the 1960s, there was less heat generated per circuit. However, with the new solid state technology came new requirements for more sophisticated designs and capacities. The number of individual circuits multiplied.
Added to this was the narrow tolerance to temperature variations. A transistor, also normally used as a switch, could turn on when supposed to be off, when overheated, causing chaos in the system.
A transistor was not a perfectly predictable device at that time. They would behave as required within tolerances, and they were individually selected for this. If close to the tolerance limit, and in a susceptible position in the circuit or physically in the machine, a problem could occur. Room air conditioning became very important.
In part 2 we will look at conditions in the computer room.