DIRECT CURRENT GENERATORS 73 



and bear on the commutator so hard as to score it 

 badly. One carbon brush should be removed from its 

 holder and the fibre brush inserted in its place, with 

 the two wires in the brush connected to a low reading, 

 or milli voltmeter. With normal volts no-load on the 

 generator, the brushes should be shifted till the instru- 

 ment needle has passed through the zero point, and 

 then back again until the instrument again indicates 

 zero, to make sure that the actual zero has been found. 

 Pencil mark the rocker arm for this shift and then move 

 the fibre brush to each of the other studs successively, 

 shifting the brushes, if necessary, till zero reading is 

 obtained, and pencil-mark the rocker arm for each 

 stud. If a different shift is required to locate the neutral 

 of the different studs, shift the brushes to a position 

 which is the mean of all the different positions. With 

 the brushes set in the mean position and the inductive 

 shunt properly connected, put on normal load and note 

 the commutation. If commutation is not practically 

 sparkless at normal load and rated overload, take off 

 the load and field excitation, and connect a german 

 silver shunt across the commutating field terminals. 

 If the machine requires an inductive shunt, the german 

 silver and inductive shunts are connected in series. 

 With the total shunt resistance great enough to shunt 

 not more than 10 per cent of normal load current, full 

 load is applied and commutation noted. The length 

 of the german silver is changed and the commutation 

 is tested until an adjustment has been obtained which 

 gives the best commutation throughout the range of 

 load required. An ammeter is then connected in and 

 the number of amperes flowing through the shunt 

 circuit read and recorded. In case satisfactory com- 

 mutation cannot be secured, the wiring, spool assembly, 

 pole and brush spacing, air gap, polarity, spacing of 

 equalizing rings, etc., should be checked. If these are 

 all found to be correct, the fibre brush should be used 

 again and the full load neutral of each stud tested. If 

 an appreciable voltage is obtained between adjacent 

 bars, the brushes should be carefully shifted until 

 zero voltage is obtained, and the shunt across the 



