The Recalibration
1952 · Drift Correction · Work 674 of 1,871
A technician's checklist on a clipboard hung on the front of an analogue computer, the sheet ruled into lines, most lines carrying a pencil tick in the margin, the last two lines still open, a pencil on a string, the writing seen at a shallow angle so it reads as a texture of marks, potentiometer knobs and a graduated dial with fine tick marks on the panel behind the clipboard, in a computing laboratory of the early 1950s, one lamp lit over the clipboard, the laboratory beyond dark.
- What it was used for
- An analogue computer had to be balanced, and every newly patched program had to be checked. Every operational amplifier had its own balance control; EAI describes an individual balance adjustment on each amplifier of the TR-48 and, on the 231R-V, a dynamic balancing circuit to minimize inherent drift of the direct-coupled amplifier stages. Then came the static check. For a fixed initial condition the expected values were computed by hand and compared with the voltages at the outputs, in EAI's words comparing calculated values of the problem variables with voltages measured in the computer program. Only when the two sets of figures agreed was the setup held to be correctly wired.
- What was new
- There is no innovation here in the sense of a device. What is documented is how the problem was shifted: according to Rubin, components initially had tolerances of about one per cent, from around 1954 temperature-controlled resistors and capacitors reached 0.01 per cent, and on the TR-48, according to EAI, chopper-stabilised amplifiers kept drift small. Later the checking was automated, on the EAI 231R-V by the punched-tape-controlled ADIOS unit, which performs the static problem check automatically. The barb stands in Rubin's own text: he admits having tuned installations, when necessary, until they reproduced the comparison solution. The check meant to find the error could thus become the target the machine was adjusted towards.
What the sources leave open: The sources consulted record no event for the year 1952; Rubin's figure of 0.01 per cent refers to the EAI 31R of about 1954, that of one per cent to the machines before it and the EAI quotations to the TR-48 and 231R-V of the 1960s. Rubin was an employee of Electronic Associates and was writing about the market in which his own firm operated.