Abstract :
The paper is concerned with the design of apparatus for maintaining a chamber at a nearly constant temperature, above the ambient temperature, despite fluctuations of its environment. After brief discussion of thermostats in general, attention is confined to electrically-operated devices in which the temperature-sensitive element enters into a bridge configuration: departure of the temperature from the assigned working value, by throwing the bridge out of balance, effects a corrective change in the heat supply. By analysis and experiment published elsewhere, the author has shown that in such systems hunting, necessarily present if the heat supply is controlled in discrete quantities, will occur also if there is a continuous relation between temperature and heat supply, provided that the control sensitivity exceeds a critical threshold value. The bearing of this result on the design of fine temperature-regulating apparatus is examined, and the frequency and amplitude of hunting are considered. The relations of the factors determining the residual inconstancy of temperature are analysed. Amongst the conclusions reached are (roughly stated): (a) hunting does not depend on the heat capacity of the chamber; (b) the residual inconstancy does not depend on thoroughness of thermal insulation of the chamber; (c) there is a single figure of merit measuring the intrinsic effectiveness in resisting changes of ambient temperature and of supply voltage; (d) of these two factors, the former is of less practical importance than the latter; (e) the effect of temperature-difference between the chamber and the temperature-sensitive element must not be ignored; (f) identification of the temperature-sensitive resistor with the resistor carrying the controlled current, advocated by someprevious workers, is wrong in principle. The author´s apparatus is described: it is the outcome of an attempt to obtain the smallest possible inconstancy consistent with derivation of electric supplies fr- om the a.c. mains alone, and is in accord with the principles reached in the foregoing study. Difficulties experienced in measuring small temperature-changes are reported, and the significance of readings of mercury thermometers is assessed. Finally, it is shown that the author´s design could be modified for large size and high temperature. A term not in common use has, for the sake of brevity, been employed throughout the paper: a ?millideg.? (to be pronounced as written) stands for a thousandth of a degree Centigrade.
