3. Storages.
(i) The storage problem. As was explained in § 1 it is necessary for the calculator to have a memory or information storage. Actually this appears to be the main limitation in the design of a calculator, i.e. if the storage problem can be solved all the rest is comparatively straightforward. In the past it has not been possible to store very large quantities of information economically in such a way that the information is readily accessible. There were economical methods such as storage on five-unit tape, but with these the information was not readily accessible, especially if one wishes to jump from point to point. There were also forms with good accessibility, such as storage on relays and valves, but these were quite prohibitively uneconomical. There are now several possibilities for combining economy with accessibility which have been developed, or are being developed. In this section we describe the one which will most probably be used in the calculator.
(ii) Delay line storage. All forms of storage depend on modifying in some way the physical state of some storage medium. In the case of ‘delay line storage’ the medium consists of mercury, water, or some other liquid in a tube or tank, and we modify its state of compression at various points along the tube. This is done by forcing supersonic waves into the tube from one end. The state of the storage medium is not constant as it would be for instance if the storage medium were paper or magnetic tape. The information moves along the tube with the speed of sound. Unless we take some precautions the sound carrying the information will pass out of the end of the tube and be lost. We can effectively prevent this by detecting the sound in some way (some form of microphone) as it comes out, and amplifying it and putting it back at the beginning. The amplifying device must correct for the attenuation of the tube, and must also correct for any distortion of form caused by the transmission through the tube, otherwise after many passages through the tube the form will be eventually completely lost. We can only restore the form of the signal satisfactorily if the various possible ideal signal forms are quite distinct, for otherwise it will not be possible to distinguish between the undistorted form of one signal and a distorted form of another. The scheme actually proposed only recognizes 21024 distinct states of compression of the water medium, these being sequences of 1024 pulses of two different sizes, one of which will probably be zero. The amplifier at the end of the line always reshapes the signal to bring it back to the nearest ideal signal.
Alternatively we may consider the delay line simply as providing a delay, as its name implies. We may put a signal into the line, and it is returned to us after a certain definite delay. If we wish to make use of the information contained in it when it comes back after being delayed we do so. Otherwise we just delay it again, and repeat until we do require it. This aspect loses sight of the fact that there is still a storage medium of some kind, with a variety of states according to the information stored.
There are, of course, other forms of delay line than those using acoustic waves.
(iii) Technical proposals for delay line. Let us now be more specific. It is proposed to build ‘delay line’ units consisting of mercury or water tubes about 5’ long and 1” diameter in contact with a quartz crystal at each end. The velocity of sound in either mercury or water is such that the delay will be 1.024 ms. The information to be stored may be considered to be a sequence of 1024 ‘digits’ (0 or 1), or ‘modulation elements’ (mark or space). These digits will be represented by a corresponding sequence of pulses. The digit 0 (or space) will be represented by the absence of a pulse at the appropriate time, the digit 1 (or mark) by its presence. This series of pulses is impressed on the end of the line by one piece-crystal, it is transmitted down the line in the form of supersonic waves, and is reconverted into a varying voltage by the crystal at the far end. This voltage is amplified sufficiently to give an output of the order of 10 volts peak to peak and is used to gate a standard pulse generated by the clock. This pulse may be again fed into the line by means of the transmitting crystal, or we may feed in some altogether different signal. We also have the possibility of leading the gated pulse to some other part of the calculator, if we have need of that information at the time. Making use of the information does not of course preclude keeping it also. The figures above imply of course that the interval between digits is 1 μs.
It is probable that the pulses will be sent down the line as modulation on a carrier, possibly at a frequency of 15 Mc/s.
(iv) Effects of temperature variations.–The temperature coefficient of the velocity of sound in mercury is quite small at high frequencies. If we keep the temperatures of the tanks correct to within one degree Fahrenheit it will be sufficient. It is only necessary to keep the tanks nearly at equal temperatures. We do not need to keep them all at a definite temperature: variations in the temperature of the room as a whole may be corrected by altering the clock frequency.