TRAJSrS ACTIONS OJ<' SECTION G. 645 



The magnetising current is obtained by using a Poulsen-arn generator con- 

 nected acrosb' the town mains (240 volts). In order to demonstrate the extent of 

 the losses, a magnetic heater or boiler, consisting of a solenoid wound on a glass 

 tube containing water and a number of iron strips or wires, is inserted in the 

 bhunt circuit of the generator. A short time after the shunt circuit is closed 

 the water begins to boil. 



An experiment was also arranged to show the variation in permeability of iron 

 ■with variation in temperature. A ring of Lohys (mild steel) is insulated with 

 asbestos and wound with a number of turns of copper wire. This is connected 

 in the shunt circuit of the generator. By passing a large current through the 

 vvinding of the ring, thereby obtaining a strong magnetising force, the magnetisa- 

 tion losses are such that a rapid rise in temperature is produced, causing the 

 iron to attain a bright red heat. The variation in permeability corresponding to 

 the rise in temperature may be followed by observing the current in the sluint 

 circuit and the voltage across the terminals of the ring. Data, including the 

 watts lost per kilogram, are given which enable the changes to be followed. 



2. A Self -Adjusting Commuiailiig Device. 

 By Professor Miles Walker. ^ 



It is well known that a continuous current generator fitted with commutating 

 poles (sometimes called inter-poles) will not commutate very heavy over-loads" 

 owing to the magnetic saturation which occurs in the iron of the inter-pole. 

 When the excitation becomes very great the saturation of the iron interferes 

 with the proportionality that should exist between the current to be com- 

 mutate-d and the conmiutating flux. 



The paper d«serib.ed a device tried at the iNlanchester !-'chool of Technology 

 which automatically produces the right amount of commutating flux over very 

 wide ranges of load. So effective is the correcting influence, even where the 

 diverter in parallel with the commutating pole winding is deliberately put out 

 of adjustment, that, whether it be short-circuited or given a resistance equal 

 to five times the proper resistance, the flux will be preserved at approximately 

 the correct value. 



Each brush (say, the positive brush) consists of two parts, one part, A, 

 being slightly in advance of the other on the commutator, and lightly insulated 

 from it. The winding on the commutating pole has twice as many turns as 

 one would ordinarily have, and the cross-section of the copper conductor is 

 only sufficient to carry one-half of the armature current. The part of the 

 brush B may be connected through a diverter D, having the same resistance as 

 the winding W, or the diverter may be dispensed with altogether. The direc- 

 tion of rotation of the machine is such that a point on the commutator passes 

 first under B, then under A. 



Assuming that the number of turns on the commutating pole is such as to 

 give the right commutating flux when half current is passing through them, 

 then it will be found that the current will divide evenly between brushes 

 A and B, because the correct excitation of the commutating pole gives an even 

 distribution of current over the face of the compound brush AB. If, however, 

 the number of turns in W is too great, there is a tendency to over-commutation; 

 that is to say, the current in B becomes greater and the current in A less. 

 This has the effect of automatically weakening the commutating pole, and 

 the pole remains of just such a strength as to bring about the required distribu- 

 tion of current between A and B. 



It will be found on machines of ordinary rating, and more particularly on 

 machines of very large rating, that the 'correcting"' voltage between A and B 

 necessary to bring about a suitable division of current between the two 

 brushes is normally very small, being of the order of 05 of a volt, whereas the 

 voltage which can be generated between A and B by a pole being too weak 

 or too strong may easily amount to 5 volts or more. Thus a very little dis- 

 turbance in the equality of loading on brushes A and B is sufficient to yield 



' See T/ie Elecirician, vol. Ixxv., p. 872. 



