OCT. 19, 1921 ROBERTS: FURNACE REGULATOR 405 



wound with a heating coil connected in series with the lamp. A 

 suitable contact is arranged to short-circuit the heating coil when the 

 strip has expanded to a certain point, whereupon the strip cools and 

 the contact opens. This particular model is mounted in a tube pro- 

 vided with a lamp base at one end and a lamp socket at the other. 

 For the present purpose, a 60-watt, 110-volt lamp flasher, with a lamp 

 in it, is placed in parallel with another lamp. The effect on the gal- 

 vanometer may be adjusted by changing from one size lamp to another. 

 A satisfactory combination of lamps in the present apparatus is one 

 which causes the flasher to complete about four cycles per minute. 



A satisfactory value for the resistance R3 is about V3 the resistance 

 of the furnace. The rest of the apparatus consists of switches and 

 resistance coils and need not be described. However, it may be per- 

 tinent to point out that the resistance wire used in the bridge should 

 be of constantan, advance, ideal, manganin, therlo or some other al- 

 loy having a very low temperature coefficient of resistance ; that the 

 resistance coils should have ample current-carrying capacity; and 

 that where copper connectors are used in the bridge, these should be 

 short and of large cross section. 



PRINCIPLES AFFECTING THE CHOICE OF A GALVANOMETER 



This discussion applies particularly to resistance furnaces having 

 a wire resistor wound on an insulating tube and imbedded in a thin 

 layer of insulating cement, both tube and cement, as well as the wire, 

 being good conductors of heat. The heating element as a whole is 

 surrounded by a thick layer of loose magnesia or other nonconductor 

 of heat. With this type of construction, the difference between the 

 average temperature of the wire and that of the tube amounts at high 

 temperatures to several degrees and adjusts itself much more quickly 

 to changes in the heating current than does the temperature of the 

 tube. Consequently, if the heating current is alternately increased 

 and decreased at short intervals the temperature of the heating wire 

 will oscillate through a considerable interv^al,^ while that of the tube 

 remains nearly constant. 



The average temperature of a point within the furnace depends, 

 among other things, upon the average rate at which energy is sup- 

 plied to the furnace. Since the regulator must choose between two 

 rates over whose value it has no control, it can only obtain the de- 



3 In the only case studied this interval amounted to at least 2° C. with a half period of 

 about one second. 



