PROBLEMS RELATIVE TO DESIGN OF CONTROLS FOR HUMAN USE 



141 



The practice of studying the effects of 

 modifications in existing equipment has 

 directed attention away from the basic ques- 

 tion of whether or not the control system 

 being used is the best one for the task. For 

 illustration we may consider the task of 

 making settings on a linear scale. This is 

 a Type B task. A study of optimal values 

 for a rotating knob device indicates that 

 by the use of an optimal knob diameter, 

 an optimal ratio between control movement 

 and pointer movement, and an optimal error 

 tolerance, subjects were able on the average 

 to complete an adjustment in from 2.0 to 

 2.5 seconds depending on the distance to be 

 traveled to the required setting. However, 

 studies of position adjustments using a slid- 

 ing device consistently report total adjust- 

 ment times under 1.0 second. Thus, in a 

 Type B task in which minimum time for a 

 total adjustment is desired, a shding device 

 for positioning the pointer will result in faster 

 adjustments than an optimally designed 

 rotating knob device. The results of the 

 former study will furnish the designer with 

 information concerning the speed of per- 

 formance to be expected ^vith a rotating 

 knob for making linear scale settings and will 

 furnish him with optimal values for various 

 design variables involved in such a device. 

 On the other hand, it will not tell him the 

 optimal design factors for "making settings 

 on a linear scale." Information of the 

 former sort is typical of the information now 

 available. Information of the latter sort 

 is typical of the information which is lacking. 

 Device manipulation rather than the adjust- 

 ment of the control system has been con- 

 sidered "the task." 



A second general feature of studies on 

 control design is the failure to control other 

 variables in the system while modifying the 

 variable about which information is desired. 

 Studies of compensatory tracking (a Type C 

 adjustment) have reported optimal gear 

 ratios in terms of required winding speeds 

 when handwheel cranks are used. The 

 optimal gear ratios are those which allow 



the fastest winding speeds over the range of 

 target speeds expected. Tracking accuracy 

 increases with winding speed, but the latter 

 is limited by fatigue factors and the max- 

 imum speed attainable. In this task the 

 operator receives most of his control data 

 from the observation of the position and 

 movement of a variable pointer relative to 

 a fixed pointer. The smoothness of the 

 movement of the variable pointer and the 

 rate at which it moves affect the nature of the 

 data which the operator receives. These 

 two factors are also related to the gear ratio 

 of the handwheel device. The question 

 therefore must be asked as to whether the 

 fast winding speeds are optimal because they 

 accomplish optimal movement characteris- 

 tics or whether they are optimal because they 

 improve the display characteristics of the 

 system. Fast winding speeds have their 

 disadvantages. If they merely improve dis- 

 play characteristics, then they are optimal 

 only insofar as no other means of improving 

 the display is available. Further research 

 is called for. 



A third unfortunate feature of studies on 

 control design variables lies in the interpre- 

 tation of the results. Conclusions are stated 

 in terms of operator performance, although 

 the data are obtained from system perform- 

 ance. Consequently, the differences be- 

 tween control tasks as well as the differences 

 between control systems tend to be neg- 

 lected. The effects of neglecting these 

 differences may be illustrated by the reported 

 effects of adding friction to manipulative 

 devices. In the literature the results of 

 a number of different studies have been sum- 

 marized as the effects of friction on operator 

 performance and have been considered con- 

 tradictory. However, the results are ac- 

 tually the effects of introducing or in- 

 creasing friction at a certain point in different 

 control systems. The situation is analogous 

 to an attempt to state the effects of increas- 

 ing the octane rating of the fuel on the 

 performance of gasoline engines. If no ac- 

 count is taken of the differences between 



