In order to find the characteristics of the shield control of the adiabatic calorimeter, two different cases are analyzed and a general formula for selecting the best settings of controller (parameters) is established. In one case, a conventional PID controller and in the other case an additional integrator cascaded to it, are used to control the calorimeter system. The temperature range is 0-600°C and an accuracy to be attained is ±0.01%.
Some of the major non-linearities of the system being controlled are taken into consideration. They are the non-linear variation of the heat capacity (
C1) of test piece and the temperature change of heat loss (σ23) of the shield.
The major factors which cause an error are large outer disturbances, the drastic temperature change in
C1 and sharp rate of change in σ23 at elevated temperature. If the total static gain
k of the part of system between the detecting and final control elements can be made big by using a high gain DC amplifier and a magnetic amplifier, it is possible, except in the case where d
C1/
dθ
1 has large value while
C1 itself stays small, to neglect such non-linear terms as being smaller by one order than the terms containing
k throughout the range of operation. Then the system is approximately linear and the influence of temperature dependency of
C1 and σ23 becomes equivalent to the effect of disturbances in the heat input to test piece.
The transition of the error caused by these temperature changes in
C1 and σ23 is tracked with an analogue computer and it is checked whether or not the controller settings programed for linear cases are adequate. It is found that with the PID+I type controller, a satisfactory control is achieved throughout the range of temperature variation whereas with the mere PID type controller, it is impossible to eliminate the stationary error of considerable magnitude at higher temperature.
抄録全体を表示