The amount of power as ascertained from the calibration curve of the motor for the voltage and current used therein when driving the generator as just explained, is a measure of these two losses. The power thus used is practically constant at all loads and is about 2 per cent. of that necessary to drive the generator at full load.
Fig. 2,912.—Characteristic curves for a compound dynamo. If the machine be over compounded, the characteristic curve has the form of the curve A B, which curve was obtained from a machine over-compounded from 118 to 123 volts, and designed to give 203 amperes at full load. The preliminary arrangements for testing a compound dynamo are similar to those for a shunt generator, and if the shunt across the series field winding be already made up and in position, the readings are taken precisely in the same manner. It is generally considered sufficient if observations be recorded at zero, ¼, ½, ¾ and full load. If it be desired to ascertain the effect which residual magnetism has upon the field magnets the current is decreased after the full load point is reached without opening the circuit, and readings are taken in succession at ¾, ½, ¼ and zero load giving in this case the curve B C D E S. It is thus seen that residual magnetism exerts no small effect upon the voltage obtained at the different loads, for had there been no residual magnetism in the field magnets the curve B C D E S would have coincided with the curve A B. The curve A B, and the straight line A X drawn through the points A and B, are almost identical, and as A X represents the theoretical characteristic curve for the machine, it is seen that compounding is practically perfect. In order to insure such accurate results being obtained, providing the machinery be correctly designed, requires considerable care in taking the readings; for example, each step or load on the ascending curve should not be exceeded before the corresponding deflection is taken, else the residual magnetism will cause the pressure reading to be higher than it actually should be, and the following pressure readings will also be affected in the same manner. In case the shunt to be employed across the series field has not been made up, it is advisable to perform a trial test before taking the readings for the curve as previously described. The trial test consists in taking two readings,—one at no load and the other at full load, the shunt being so adjusted as to length and section that the desired amount of compounding will be obtained in the latter reading with normal voltage at no load. If the first trial fail to produce the desired result by giving too low a voltage at full load, the length of the shunt across the series field should be increased, or its section should be reduced by employing a less number of strips in its makeup; again, if the voltage at full load be higher than that desired, there must be made a decrease in length or an increase of section in the shunt employed.
The friction of the brushes can very conveniently be determined next by lowering them on the commutator and giving them the proper tension.
The increase in power resulting from the greater current that will now be taken by the motor to run the dynamo at its rated speed, will be a measure of this loss. In general, its value will be about .5 per cent. of the total power required to drive the dynamo at full load, and this also will remain constant at all loads.
The friction of the air upon the moving armature of the dynamo cannot be determined experimentally, but theoretically this loss is small and may be estimated as .5 per cent.; it is also constant at all loads.
The core loss may be determined experimentally by exciting the field magnets of the dynamo with the normal full load field current through the magnet coils, and noting the increase of power required by the motor to maintain the rated speed of the dynamo thus excited under no load, over that necessary under the same conditions with no field excitation. This increase of power will be the value of the core loss. The core loss is approximately 3 per cent. of the power required to operate the dynamo at full load, and it is constant at varying loads. If it be desired to divide the core loss into its component parts, it is necessary also to run the dynamo under the same conditions as before with field excitation but at half its rated speed. If, then,
H = the power lost in hysteresis at rated speed,
E = the power lost in eddy currents at rated speed,
T = the power lost in hysteresis and eddy currents at rated speed,
S = the power lost in hysteresis and eddy currents at half speed.