]\gt ; SLIP .
iresis of Iron and Steel in a Rotating Trans. , A. , 1896 , pp. 716-746 .
, Verification of Law of , Effect of Values of at Critical Point .
Trans. , , vol. 187 , 1896 , pp. 715-746 .
of Variation of Hysteresis thence Trans. , , vol. 187 , 1896 , pp. 716-746 .
[ 715 ] XVIII .
The Hysteresis of Iron and Steel in a Magnetic Field .
FRANCIS GIBSON BAILEY , , Professor of Electrical Engineer , ng , HeriotCollege , Edinburgh .
Communicated by Professor O. LODGE , , F.R. Received Apri19 , \mdash ; Read June 4 , \mdash ; Revised September 4 , 1896 .
IN a paper on Dynamo Electric Machinery by Dr. JOHN HOPKINSON ( ' Phil. TraIlS .
' 1896 ) , the suggestion is made that the value of the hysteresis of the iron core of a rotating dynamo armature need not be identical with the value obtained when the magnetising foroe is reversed by through a value .
It was subsequently pointed out by .
SWINBURNE that as a necessary deduction frolu Professor EWING'S molecular theory of the hysteresis of iron in a rotating field , or of iron rotating in a constant field , should show a distinct diminution in value below that in an alternating field , when the magnetic condition of the iron approaches saturation .
According to EWING'S theory hysteresis is due to the formation of stable magnetic combinations between adjacent molecules which tend to resist any movement of the molecular magnets caused by change of direction or magnitude in the magnet.ising force .
On the breaking up of these combinations by such a change in the magnetising force , new arrangements are formed , and the potential energy of position is transformed into kinetic energy of partial rotational movement round the fixed axis of , which is dampod out with or without osci]lations above the axis of rotation .
It has been further suggested that the damping process may be due to eddy currents induced by the movement of the magnets , but the precise nature of these eddy , or the extent to which other retarding influences akin to mechanical fr or viscosity may act has not been determined .
In the alternating field there is no continuity of position of the molecules ; the magnetising force passes through a zero value at each alternation , and the molecular combinations may vary in successive alternations , although the average value of the induction and hysteresis remains sensibly constant .
Increasing magnetising force and induction result in increased movement of the molecules with increased momentum aild consequent dissipation of energy .
But in a rotating field of constant value there is no of the strength of the magnetising force .
The molecules are.always under the same restraint , and the movements impressed upon them will be uniform unidirectional .
It is true that with small magnetising forces , 13,1.97 when but few combinations are dissolved and the movement is of a quasi-elastic nabure , the amount of energy dissipated will not be greatly different from that dissipated in an alternating field of equal strength .
In fact , since the molecular changes ill be forced to take place in one direction only , there will be- on the average a greater resistance to movement , and in consequence an increased dis- sipation of energy .
When , however , the field becomes stronger , each molecule will develop a tendency to rotate in synchronism with the field and be less affected by the magnetic influences of surrounding molecules .
At this point the hysteresis ill show a considerable diminution which will become more marked as the ] increases in strength , until finally every molecule will Yotate in unison with the field with complete absence of oscillatory movement .
The value of the hysteresis under these last itions will furnish an important clue to the nature of hysteresis .
If the hysteresis sensibly vanishes at this point it will be strong evidence that the damping of the movemenl , of the molecules is not due even partially to mechanical friction , but must be produced by some action which is called into play by the rapid lations of the molecular nets , but not by the comparatively slow motion of their rotation with the field .
Since the difference between these speeds must be enormous , the damping may be due either to some form of eddy currents or possibly to some form of fluid friction .
The first experimental work upon hysteresis in a rotating field was carried out by Professor FERRARIS ( ' Atti .
R. Acc .
di Torino , ' No. 23 , .
Produoing a rotating field } ) means of two coils at right angles supplied with alternating currents of approximately one quarter difference of phase , he showed that a laminated iron core would rotate by reason of its hysteresis .
Beyond proving that at low speeds the hysteresis was independent of the speed , his results were not quantitative , owing to irregularities caused by vibra ion in the apparatus .
In order to investigate matter the following apparatus was de@igned .
powerful electro-magnet is caused to rotate on fixed spindles arranged so the axis of rotation passes between the pole pieces , which are bored out to a cylindrical form concentric with the axis .
A cylindrical armature is held on pivot in the fixed spindles concentrically between the poles .
The direction of magnetisation rotates in a plane at right angles to the axis of the armature and concentric with its axis .
The armature , though free to rotate in bearings , is prevented ffom continuous rotation by a spring attached at one end to its spindle , and at the other end to the fixed spindles of the magnet .
Movement of the armature is indicated by a beam of light reflected from a small mirror attached to the armature on to a circular ] concentric with the armature axis .
The apparatus is shown in fig. 1 .
The magnet is of Swedish iron , 8 sq .
centims. .-section inside the coils .
The pole pieces are bored out to a diameter of centims. , and subtend an angle of .
They are axially of the same length as the armature .
The excitation is produced by two coils each of 316 IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
ilrns of No. ] wire , one end of the exciting circuit being attached to an insulated ring , the other to the magnet .
The current is led in by brush contacts .
of the bearings of the electro-magnet is in the yoke piece , the other is in a gun-metal bracket bolted to the ends of the pole pieces .
This bracket carries the insuIated collecting ring .
A third bearing is arranged in a bracket placed between the pole pieces to prevent any bending of extended end of the spindle .
The spindles are of steel , and are held by set screws in upright gun-metal standards bolted to the bed plate .
The rigidity of the electro-magnet was ensured by strong bolts passing through the yoke piece , by the bracket across the ends of the pole piece , and by additional brass brackets of -section across the sides of the pole pieces .
This was found to be satisfactory , not the slightest movement of the pole pieces being detected .
The armatures were cylindrical in form , and were made of very thin plates strung on a manganese steel pindle 2 millims. diameter , and held in place by ebonite washers at each end .
Some difficulty was experienced in obtaining satisfactory sheets of charcoal iron , but finally some exceptionally thin plates , 081 millim. in thickness , were obtained from Messrs. KNIGHT and CROWTHER , who very kindly hffi them specially rolled for the purpose .
rrhe metal was almost pure iron , containing per cent. of carbon and only traces of silicon .
Its specific resistance was ohm .
The plates were insulated from each other by a layer of tissue paper .
The armature was centims. in length and consisted of 250 plates .
The diameter was centims. A second armature of hard cold rolled high carbon steel was made with plates 141 millim. in thickness , consisting of 147 plates similarly insulated with tissue paper .
Its specific resistance was ohm .
The springs used were of non-magnetic material in the form of a flat helix .
To avoid disturbances the spring was enclosed in an ebonite box through which the armature spindle passed .
The armature was supported by the pointed and hardened ends of the steel spindle , which rested on hollow coned bearings of hardened steel , giving sufficient strength.while causing a minimum of friction .
The magnet was driven by a small leather belt from an electric motor .
Owing to the construction of the apparatus it was impossible to read the speed directly at high speeds , and the value was accordingly calculated from that of the driving motor .
The magnetising current was read on a SIEMENS ' electro-dynamometer , which had been calibrated from a Kelvin balance .
In order to ascertain whether the presence of windage or eddy currents in the spring attachment or spring or steel spindle produced any appreciable action , preliminary test was performed with a dummy armature in which the lron was replaced by .
ebonite , everything else remaining the same .
At the highest speeds and with the strongest the effect was inappreciabIe , and errors from these causes were certainly absent .
When the magnet rotates , the armature revolves with it until the torque exerted by the spring becomes equal to the torque due to the hysteresis and eddy currents in armature .
The rate at which energy is dissipated in the armature is then easily measured , for Power absorbed torque speed of rotation , or Energy absorbed per revolution2torque .
Since the restoring force of the spring was found to be proportional to the angle of rotation , the hysteresis and eddy current loss per revolution is proportional to deflexions of the armature as shown by the reflected beam of light .
The reflected beam of light was focussed on to a semicircular translucent scale concentric with , and perpendicular to , the axis of revolution of the electro-magnet .
The readings of the movement of the spot of light are , therefore , proportional to th energy absorbed per revolution in the armature .
The springs were calibrated by measuring the deflexion produced by a small weight at the end of a light steel arm attached to .
the armature .
They were calibrated frequentl .
, and were found to change only very slightly , even after considerable damage by being wound up , as once or twice happened .
The hysteresis per revolution is thus measured directly in terms of the force of gravity acting through a known space .
If is the length of the arm , is the mass of the weight and equivalent weight of the arm , is the volume of the iron , is the deflexion caused by the weight .
Then a , of 1 division against the spring ergs per cub. centim , per revolution of energy loss in the armature .
Determination of Eddy Currents .
The only correction of any importance that is required is the loss due to eddy currents .
In most tests on hysteresis at a high speed of alternation the eddy current loss as an element of some uncertainty , and therefore great attention was paid to this point .
There are two ways in which error may be cause Eddy currents will be set up in the armature plates by the movement of the lines of force through them , which will oduce a torque on the armature in the same direction as that due to hysteresis .
There may also be eddy currents set up in the framework of the apparatus which will modify the magnetic field through the so that it will have a value slightly from that , produced by magnetising current alone when the magnet is standing still .
This , however , -may be sh-.own to be very small .
In the rotating parts there will be no eddy currents , since the direction of the lines of force will travel round-with the electro- magnet .
In the fixed spindles there will be currents- .
set up due to the leakage AND A ROTATING MAGNETtC FIELD .
lines of force which pass through them , which currents will flow along the sides of the spindles and across the ends .
The latter portion will be parallel to the lines of force of the field , and hence will produce a magnetic field at right angles to this , causing a slight distortion , but not altering the total magnetic flux between the pole pieces and through the armature .
Since the spindles were some distance from the pole pieces , and since no perceptible heating was produced , it may be assumed that no distortion was produced of sufficient magnitude to produce an appreciable change in the of the hysteresis in different parts of the armature .
The eddy currents in the armature plates may be divided into two portions , ( 1 ) those due to the lines of force which pass through the plates parallel to their planes ; ( 2 ) those which are produced by the fringe of lines of force at the ends of the pole pieces , passing into the armature at the ends in a direction perpendicular to the plane of the plates .
The first set affect all the plates equally , and can be calculated with fair accuracy , but the second set only affect the end plates and are not easy to culate .
The distribution of these latter lines of force and currents are shown in fig. 1 .
It is clear that these lines of force not penetrate far into the armature , owing to the numerous small air gaps between the plates , and hence the ourrents will be limited to only a few plates at each end .
Fig. 1 .
Fig. 2 .
In order to detelmine the magnitude of this error , an armature was prepared , consisting of only 20 steel plates , insulated with ) as before , and centim. diameter ( fig. 2 ) .
It was placed in the middle of the polar area .
If the effect , of the eddy currents in end plates were appreciable , it would be proportionally much increased in this mature , and , at a high increase of speed , there should be an appreciable increase in the deflexion .
( It will subsequently shown that the hysteresis in the steel is independent of the speed .
) On experiment , it was found that there-was no normal increase in the deflexion .
It may therefore be concluded that these end eddy currents may be .
neglected .
rlhe test was made witvarious exciting currents , from a low induction up to the maxlmum obtainabIe .
Calculation of Eddy Currents .
As the armature consists of a row of plates insulated from each other , in which the lines of induction lie in the plane of each plate , the problem is reduced to the estimation of the eddy currents induced in a thin plate rotating on a central axis perpendicular to the plane of the plates in a magnetic field , the direction of which is at all points parallel to the plane of the plate .
The distribution of the lines of induction in the plate is a matter of some uncertainty , and will vary slightly with different degrees of magnetisation , owing to the variations in the permeability .
Since it will be shown , however , that the effect of eddy currents is but small , an approxiaccurate distribution will be assumed that will allow of mathematical treatment .
As it has been shown that even at the ends the lines of force do not pass from plate to plate , it will be assumed that their direction is entirely in the plane of the plate .
Fig. 3 .
Since the magnetic reluctance of the air gap is considerably higher than that of the armature or pole pieces , the distribution of the lines of force in the air gap will be sensibly uniform and radial .
It will be assumed that the whole of the lines of force pass in this way to the armature , and the fringe from the sides of the poles will be neglected .
This assumption wilJ somewhat increase the value of the eddy currents , as it gives a density in the outer layers slightly in excess of the actual value .
Inside the plate the paths of the lines will be curved , owing to the curve.of the circumference and the space occupied by spindle .
Let a distribution be assumed such that the rate of cutting lines of force is proportional to the distance from the centre ( neglecting the hole ) , and such that the portion of each revolution during which any point is cutting lines of force is the same , viz. , of each revolution , and IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
72 ] .
that the rate of cutting lines of force during that time is uniform .
For the remaining revolution , the point is travelling along the lines of force .
A reference to the fig. 3 , illustrating this distribution , will show that such a distribution is very approximately the actual condition .
Fig. 4 .
Let QP be a plate in the armature , fi , and let ABCD be a section of this plate at some point between and , this part bein covered by the pole piece and PK being between the pole pieces .
Let the of the plate AB ) , and let the diameter , and let the angle subtended by the pole piece .
In the samples used , has a value of 150 to 200 .
In any section QQ ' the path of the current induced in the plate by its revolution in a magnetic field parallel to the plane of the plate , will be approximately rectangular and similar to ABCD .
Let ON , and let abed be the path of the current through a point in OF at a distance from O. Then the length of the path .
Let the thickness of the element be , and the breadth along .
Then breadth along .
Let specific resistance of the metal in ohms .
Then the resistance of element .
Let the induction across diameter of plate .
Then the induction in the air gap assuming a distribution of magn etisation as above .
And according to the previous assumption induction at right angles to the tangent at any point in the path as far as the line OP will be also PROFESSOR 1 G. BAILEY ON THE HYSTERESIS OF Let number of revolutions per second of the armature .
Then EMF in the circuit Bnmz2 volts .
Then the energy spent in eddy currents per second ( from OP ' to OP ) in the elementary circuit Therefore the loss in the whole section Over the whole plate the loss is obtained by taking sections from to ; i.e. , integrating from to The integral when it becomes , but since is a large number and is small , it may be simplified to the form This Therefore the eddy curren losses in each plate atts .
The volume of each plate , therefore the eddy current loss per cubic centimetre watts .
For calculation it is more convenient to replace the half thickness of the plate by the thickness .
Then the loss per cub. centim. The expression given by ofessor J. J. THOMSON for the loss due to eddy currents in iron plates in an IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
alternating magnetic field , the difference in the constant being due to the fact that the distribution of E.M.F. is different in the two cases , the average being larger for the rotating field , and also that in the latter the average length of path and resistance of the eddy current circuits are smaller .
It is more convenient to estimate the loss due to eddy currents in ergs per revolution .
This ergs per cub. centim. per revolution .
For the soft iron and .
eddy losses For the hard steel and ; eddy losses of Induction in The induction in the armature was measured by winding a few turns of round the iron , through holes in the ebonite washers .
The electro-magnet was excited by a measured current .
The coil was placed in series with a ballistic galvanometer , and the throw observed when the coil was suddenly moved half round .
By this means all effects of the stray field of the magnet on the galvanometer and leading-in wires were avoided .
These were , however , very small , since the galvanometer was 30 feet distant , and the connecting wires were carefully stranded and insulated .
The galvanometer was standardised by a Clark 's standard cell and a Muirhead standard condenser .
The resistance of the galvanometer circuit and coil was made large , to avoid damping due to currents induced in the coils by the of the needle .
Other causes of damping bein the same both for the induction throw and the condenser throw , it was not necessary to allow for this .
In the soft iron armature , and to a less extent in the steel armature , the paper occupies some space , and with a strong field the lines of force through the paper are an appreciable quantity .
The strength of the field was determined by winding an air coil round the ebonite washers at the end , and measuring the throw .
This gives a fairly accurate value for the strength of the field , as it is just outside the iron .
The area occupied by paper and .
in the armature was calculated , and the proportional correction subtracted from the total number of lines of force as given by the coil round the iron .
It did not amount to more than 5 per cent. in the stro1Jgest field .
The results of the calibrations are given in the Diagrams 2 and 3 .
from which the values for the induction in the tables of readings are taken .
In the value of the area of the iron , the small hole in the centre was subtracted from the total cross-section of the armature , so that over a part of the armature the value of will be really somewhat less than that given .
But for the part under the pole-pieces the value will be nearly correct .
PROFESSOR F. G. BAILEY ON THE HySTERESis OF Test o.f Permeability and Hysteresis in Reversing Field .
The method adopted was that used by EWING in his more recent experiments and described by him in his paper ( ' Phil. Trans 1894 ) , in which a continuous ring of iron is wound uniformly round with a magnetising coil , and a secondary coil is connected to a ballistic galvanometer .
The current is always brought to the maximum value , and the throw of the galvanometer taken when the current is reduced ( counting the sign algebraically ) , until the value of the reversed initial current is reached .
change being practically instantaneous , is probable that this method gives much the same value for the hysteresis as a rapidly alternating current does , but the poin has not been experimentally proved .
Dr. JOHN HOPXINSON'S tests*on hardened steel show close similarity in the permeability with an alternating culrent and with a step-by-step method , but his experiments are not conclusive for the hysteresis value .
As the iron was well laminated , there would be , according to EWING , very little creeping effect in the magnetisation of the ring .
For this purpose rings were made of the iron and steel , consisting of a narrow strip wound in several layers , the Qamples being taken from portions of the sheets of material used for making the armatures , and the physical conditions being kept as nearly as possible unchanged .
Primary and secondary windings were wound on , the secondary being of fine wire placed immediately on the iron , so that no correction for air space was necessary .
The dimensions of the rings were : Soft iron.\mdash ; Area of metal . . . .
centim. Mean circumference of ring centims. Steel.\mdash ; Area of metal . .
0366 sq .
centim. Mean circumference . .
centims. The method being well known requires no further description .
The hysteresis is given by the expression , where is the area of the curve .
and being in absolute measure , the hysteresis is expressed in ergs per cubic centimetre per cycle .
By measuring the areas of the closed curves and dividing by the following values were obtained:\mdash ; Soft Iron .
3,060 Hysteresis 1,280 ergs .
* ' Electrician , ' September , 1892 , and May , 1893 .
IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
Hord Steel .
2,720 Hysteresis 3,580 ergs .
These values are shown in the curves on Diagrams 4 and 5 .
Test of Variation of Hysteresis with Induction in Rotating Field .
The following tests were taken to determine the nature of the variation of hysteresis as the induction is increased .
Both hard cold-rolled steel , and soft charcoal iron were used .
The machine was kept running continuously through the test , and bhe current was kepb on continuously and gradually increased , except when very heavy currents were used .
Some 10 or 15 secs .
were allowed to pass before the reading was taken to give the deflection time to become steady .
The effect of not complying with these conditions will be discussed later .
The shape of the pole pieces was varied twice to obtain a stronger field , and the armature was reduced in length .
The first change was to cut away the poles , making the length of the polar area centim. instead of centims. , and later coned ends were fitted to give a stronger field , the latter shape being shown in the Diagram ( 1 ) .
The calibration curves for the three shapes are shown in Diagram 2 .
The hard steel armature was used .
Readings were taken at 28 revolutions and at 56 revolutions per second , the agreement being fairly close .
With a shortened armature and stronger field , a set was taken at 34 revolutions per second .
With the shortened armature and cone-shaped pole pieces a set was taken at 26 revolutions per second .
This was the strongest field that could be obtained .
The results are given in the adjoining table , and the curves are plotted in Diagram 4 , together with the hysteresis curve , by the method of reversals for a piece of the same sample of steel .
The whole curve is shown , from very low values of up to nearly 20,000 .
It will be seen that all four sets of readings are very fairly concordant .
A maximum value of 44,000 ergs is reached at an induction which varies between 14,000 and 15,000 .
At the beginning the curve rises very slowly and then turns up sharply alon an almost straight line .
Between 9,000 and 10,000 there is a slight flexure in all of them and another straight piece until the maximum is reached .
Here there is an abrupt bend and a more rapid descent on the other side , again almost straight , the MSASURMNA APPARATUS .
Diagram 2 .
IRON AND STEEL IN A ROTATING MAGNETIC FIILD .
PROFESSOR F. G. bailey ON THE HYSTERESIS OF lowest value , 8,000 ergs , less than one-fifth of the maximum , being reached without any sign of another flexure .
This corresponds perfectly to the three stages in the curve of hard steel , the rapid rise occurring during the rapid rise of the induction , and the maximum value being reached just at the bend before the approach to saturation .
As saturation is more nearly reached , the hysteresis becomes rapidly smaller , the ourve wing no sign of bending off The values given in the tables for hysteresis are corrected for eddy currents , the magnitude of the correction being given at the foot .
The maximum value of the correction at 56 revolutions per second , when and the hysteresis is small , is only 4 per cent. of the hysteresis , and about the same for the smaller value of the hysteresis in the stronger field at the slower speed .
It will be noted that the ballistic curve lies considerably below the rotating field curve up to a point close to the maximum , after which it rises far above the other .
The meaning of this will be discussed later .
TABLE I.\mdash ; Hard Steel Armature of 105 Plates of thickness centim. Diagram 4 .
Square pole pieces .
Speed , 37 revs .
per sec. IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
TABLE II.\mdash ; Hard Steel Armature of 147 Plates of thickness centim. Variation of Hysteresis with induction .
Diagram 4 .
urrenb -evcub.centimHyst.inper r TABLE III.\mdash ; Hard Steel Armature of 105 Plates of thickness centim. Diagram 4 .
Coned pole pieces .
Speed , 26 revs .
per sec. centi B Hysteresis Charcoal Iron. .
The steel armature was replaced by the soft iron armature of plates , 0081 centim. thick .
This was subsequently reduced to 156 plates , the calibration curves being given in Diagram S. Curve 2 was carried further in experiments , the maximum induction attained being 20,200 .
Several sets of readings were taken at different speeds with both sizes of armature , some of the results being shown in Diagram 5 , together with the ballistic curve .
The general shape of the curve is similar to that of the steel , the three stages being Diagram 5 .
clearly marked\mdash ; the slow beginning , then a long straight rise , with a less distinctly shown flexure in the middle , and a sharply defined maximum , followed by a sudden drop .
The maximum occurs at a higher induction than in the steel , and the drop afterwards is more rapid .
IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
It may be noted that , while the first and last parts of the curve are in good agreement , there is considerable divergence near the maximum , and repeated experiment failed to remove this .
In fact , when kept running at the maximum value , the deflection was by no means steady , showing that it is a genuine phenomenon , and not due to errors in the apparatus .
Occasionally , in the latter half of the curve , the same phenomenon occurred , the value altering slightly for a few readings , and then coming bac.k on to the original curve , without any alteration or stoppage of the apparatus .
In these experiments the .
induction was pushed as high as possible , and a consistent continuous curve was obtained down to a very low value of the hystel.esis .
The minimum value obtained at an induction of 20,200 is only one-thirteenth of the maximum value , and the curve shows no signs of turning off again .
It is therefore highly probable that the hysteresis vanishes altogether at a slightly higher induction , although the saturation point has hardly been reached .
These last readings are very difficult to obtain , as the smallest irregularity or incidental error entirely vitiates them , and for some time the results were not good .
However , the curve given was otained several times , and the points marked are the mean of several readings , all in good agreement , so that it may be taken as correct .
The comparison with the ballistic curve exhibits the same features as in the case of hard steel , the relation between the two curves being singularly alike .
It may therefore be ( joncluded that the positions will be the same for other samples of iron and steel , since these two occupy extreme positions among the various types .
That the hysteresis in a rotating armature at low induction should be greater than in an alternating field is quite intelligible , since the movement is more gradual , and is free from sudden shocks .
There is also not so much choice in the direction of movement , and hence some of the molecular combinations will offer more resistance to dissociation .
The point is of considerable importance in the design of large dyoamo armatures , which are usually worked at an induction of about 10,000 or 8,000 C.G.S. At this part of the curve the value the hysteresis is some 50 per cent. higher than that given by a ballistic test , and allowance must be made accordingly for the larger amount of heating .
On the other hand , in ring armatures , which are worked at a high induction , the hysteresis will be considerably lower than the value given by the ballistic method .
732 HiSSOR I G. BAILEY ON THE .
OF Diagram 7 .
IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
Diagram 8 .
' 2 *5 6 7 JQoeO\ldquo ; ' / ' lnciuction in c.as .
units .
Diagram 9 .
TABLE lV.\mdash ; Soft Iron Armature of 250 Plates of thickness centim. Diagram .
Speed , 35 revs .
per .
sec. TABLE V.\mdash ; Hysteresis in Soft Iron Armature of 156 Plates of thickness .
centim. Square pole pieces .
Diagram 5 .
Speed , 21 revs .
per sec. ergs IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
LE- VI.\mdash ; Hysteresis of Soft Iron Armature with 156 Plates of thickness :0081 centim. Speed , 18 revs .
per sec. Diagram 5 .
B. 11,600 13,300 14,700 15,300 15,900 16,600 16,800 17,300 17,800 18,300 18,800 TABLE \mdash ; Soft Iron Armature of 250 Plates of thickness centim. Variation of Hysteresis with Excessive Vibration .
Diagram 7 .
Speed , 26 revs .
per sec. Eddy currents Higher values of were not possible as the vibrations threatened to break the pivots of the armature.736 PROFESSOR F. G. ILY ON THE HYSTERESIS OF TABLE VIIL\mdash ; Soft Iron Armature of 250 Plates of thickness 0081 centim. Speed , 31- ) revs .
per second .
Diagram 8 .
Starting with maximum current and decreasing .
TABLE IX.\mdash ; Hysteresis and Induction in Soft Iron Armature of 250 Plates of tluckness centim. The Armature slowly turned and then left at rest and the reading taken .
Diagram 9 .
In iagram 6 is given the first part of the curve plotted on a magnified ] and determined by means of a very weak spring , so that the deflexions were large .
The initial values are small , but the increase is rapid up to an induction of 1,400 , after which the curve becomes more nearly straight .
The effect of violent vibration was tested with this armature by comparing the readings taken with the armature almost tight in its bearings , and the magnet running very smoothly , with the readings taken when the armature was very loose .
This has the effect of causing excessive vibration , the armature rattling to and fro at every revolution .
As might be expected , the hysteresis is diminished ( Diagram 7 ) , but not by much , and it may therefore bs concluded that the small vibration present , under normal working , does not influence the values to any appreciable extent .
A similar experience of the small effect produced by violent mechanical vibration is mentioned by EVERSHED and VIGNOLLmS ( ' EIeetrician , ' 15th and May , 1891 ) .
IRON AND STEE IN A ROTATING MAGNETi .
C ] D. The effect .
of passing from a high induction to a low one , allowing a short time between each reading , was examined , the result being shown in Diagram 8 .
There was no definite change in the hysteresis , the curves of rising and falling currents being practically identical .
This point will be referred to below , when more rapid changes are taken .
The effect of speed on hysteresis will be examined later , when experiments will be described in which the induction is kept constant and the speed varied .
Finally , a test was made at an indefinitely slow speed by slowly turning the armature through a short distance and then stopping it .
The deflexion was then read , while it was standing still .
Since the smallest error of centring produced an additional pull either forwards or backwards , there are some irregularities in the readings , and a high induction could not be reached .
The result is in complete accordance with the test at higher speeds .
but it will be noted that the maximum value , though occurring at the same induction , viz. , between 16,000 and 17,000 , is not so high as before .
This is due to a small backward qvement of the armature immediately the magnet stopped , perhaps a kind of readjustment , which does not have time to take place when ruIlning .
It is only a small effect , about 10 per cent. of the value of the hysteresis at that point , and is less marked at points beyond the maximum .
The smallest actual rotation was sufficient to prevent its occurrence , the values then being closely in accordance with those at higher speeds .
The curve is shown in Diagram 9 .
[ As it was found that the speed of rotation was without influence on the hysteresis , and that the eddy currents were almost negligible at moderate speeds , the greater number of readings were taken at speeds of 15 to 30 revolutions per second , since the deflexions were very much steadier than at low speeds .
With the latter there wasx tendency for the spot of light to oscillate , especially at high inductions , owing to small errors in centring , which it was almost impossible to eliminate altogether .
The results obtained at low speeds , down to a speed of one revolution per second , were identical with those at high speeds.\mdash ; 2nd September , 1896 .
] The above experiments prove very clearly the nature of hysteresis in a rotating field , bear out most completely Mr. SWINBURNE'S deduction from the molecular theory of magnetism .
The agreement of the phenomena with the previously suggested dedu.ction forms a strong verification of the truth of this theory while it also easily explains various other points which have been noted .
The three stages of the curve correspond precisely with the three ageS of the molecular movement , and the difference between the curves for iron and steel readily follow from the their structure .
In the soft iron , as has been shown by EWING , the molecules move in larger combinations than in steel , and hence it is not until a induction is reached that uniformity of movement is produced .
When once past the maximum the decrease is very rapid , and both metals appear to give curves reaching approxizero value at an induction between 20,000 and 21,000 .
ROFESSOR F. G. BAILEY ON THE HYSTERESIS Ib is an important part of Professor theory of magnetism that hysteresis is caused solely by intermolecular magnetic control , and not by any friction on the moving molecule , although there are great temptations to assume some such friction Diagram 10 .
in certain phenomena .
But the almost complete disappearance of hysteresis in the soft iron proves that friction must be almost entirely absent , since an individual rotary motion would not avoid this .
The curve for hard steel is not so conclusive on this point , as the induction has not IRON AND STEEL IN A ROTATING MAGNETIC FIELD .
been pushed far enough , but so far there is no sign of any large amount of frictional control .
rfhis is noteworthy , as we are accustomed to consider the steel molecules as subject to some form of mechanical constriction .
In Diagram 10 are given the mean values of the curves obtained for iron and steel , the latter being plotted with a smaller vertical scale to admit of easier comparison .
With them are also plotted the curves of hysteresis in an alternating field , that for soft iron being obtained by a calorimetric method with iron from the same sample .
( ' Electrician , ' November22 , 1895 .
) PART II .
Effect of Speed on Hysteresis .
The question of the relation between the speed of reversal of magnetization and the value of the hysteresis per reversal has been investigated by several experimentalists , by the employment of an alternating current of varying periodicity .
The experiments of TANAKADATfi ( ' Phil. Mag September , 1889 ) by the calorimetric method , indicate that there is no change in the hysteresis between the values 27 and 400 cycles per second .
His method , however , is liable to error , and he does not measure the induction at the same time as the hysteresis , but finds it by a ballistic test .
As his magnetic circuit has only a small reluctance and no air gap , this assumption of identity is doubtful .
The experiments of Mr. A. SIEMENS ( ' Proc. I.E.E. , ' February , 1892 ) and Mr. C. P. STEINMETZ ( ' E.T.Z. , ' 1891 and 1892 , and 'Electrician , ' February , 189 indicate a change of hysteresis in this respect , but their results are not consistent .
By the magnetic curve Professor EWING has made some tests at low speeds , bnt the presence , of eddy currents in the pole pieces obscure the results , and mechanical lag and vibration in the moving parts prevent the use of any but very moderate speeds .
Dr. JoHN HOPXJNSON ( ' Electrician , ' loc. cit. ) finds that in hard steel wires , while the permeability is not changed , there is a small increase in the hysteresis as the speed is increased .
The experiment is , however , obscured in regard to this point by the changes in the form of the current consequent upon change of speed .
Mr. T. GRAY Proc finds no variation in the hysteresis between the speeds of 3 and 8,000 cycles per minute , but his experiments are somewhat contradictory .
Recently the author has shown Electrician , that the change of hysteresis , if it exists at all , must be very slight and that on theoretical grounds it is probably independent of speed .
In the foregoing experiments with a rotating field it has been indicated .
there is at most very little change of teresis at speeds .
The point was more carefully worked out as follows:\mdash ; The machine run at a increasing speed , and the current was kept perfectly consbant .
As been pointed out , this ensures a practically perfect constancy in the value of in the armature .
At first with the steel at.matul.e , a distinct variation was found , the sign of the change depending on the induction ; but the effect was traced to errors in the truth of the armature .
When these were eliminated the result of a large number of tests is to show that up to a speed of 70 revolutions per second there is no regular or definite change in the hysteresis when tested in the rotating magnet machine .
All the experiments were porformed by the current constant and unbroken , and increasing the speed step by step , titing .
at each reading until the reading had become steady .
The tests of the steel are shown in Diagram 11 .
There are irregularities which could not be eliminated , and which frequently repeated themselves at the same points , but on the whole there is a clear indication of no sensible change .
The readings are corrected for eddy currents .
The values at the maximum , or near to it , tend to be irregular , as was seen also in the previous curves .
This may well be expected , the metal being to some extent in a critical condition .
The values at the highest speeds were rendered slightly uncertain owin to vibration , which shook the spot oflight , but there is scarcely a definite change in one direction more than the other .
The soft iron shows the same kind ofcurves , and the same kind ofsmall variations , but it may be noted that the readings given in Tables XI .
to XIV .
are not so regular as in the case of hard steel , and repetition of the series does not reproduce the same irregu]arities .
At the value , four series were taken at different times ; two of them show slightly rising values , another falling values , while a fourth gives an approximately constant value .
An example of a rising curve is shown ab , but on repetition this gave a practically horizontal line .
On repeating the tests with a smaller armatule , the same results appear .
At the maximum value of the hysteresis when , the readings were very irregular , as was anticipated , since the iron was in a critical condition .
The same effect is noticed to a lesser degree near the maximum , when and 17,200 .
A large number of series have been given because the actual readings show irregularities which might prevent a generalization ; but after constant repetition it is seen that the variations obey no regular law , and are probably due partially to errors in the machine , though to some extent there is strong ground for concluding that they are due to actual variations in the value of the hysteresis .
The method is more accurate than those which entail the accumulatin of the waste of energy , since in them the amount ured is proportional to the speed , and the conditions are changed .
The deflections here are quite unaffeoted by the sPeed , IRON AND STEEL IN A ROTATING MAGNETIC IELD .
and depend only on the value of the hysteresis at that moment .
Hence the small evanescent variations may be readily detected , whereas by the cumulative methods they would be merged in the gerleral average .
There are a few points concerning the eflect of time and dden changes which are interesting .
As has been mentioned , the curves were taken with steadily increasing current or speed , a short time being allowed before a reading was taken .
When , however , a sudden increase is made in the current , the value of the hysteresis is higher than the normal for that current and the value slowly decreases -until it reaches the normal steady condition .
The effect is not large , about 3 or 4 per cent. , but is by no means regular .
The time of its disappearance is usually a few seconds and the effect is greater at medium and high induction than at low induction .
The reverse is shown in suddenly decreasing the current .
the hysteresis having too low a value , and not rising up to its normal for upwards of a the arrival at the normal being slower than with a rising current .
A similar effect is observed in changes of speed .
A rapid rise in speed gives an increased value of the hysteresis , which dies away in a few seconds .
But a rapid reduction in speed gives a diminished value of the hysteresis which lasts often more than a minute , during which time the deflexion is very irregular , mping up to the nolmal and going back again .
change is also greater than any of the preceding ones , and is frequently as much as per cent. The effect of rapid makes and breaks in the magnetising current in quick succession is to increaso the resis .
It recovers its normal value after a time but the readings are very irregular during recovery .
In all these effects , the machine is kept constantly running , and they occur more or less at all speeds .
They are , however , not regular enough in amount to allow of any connection with the speed bein established .
Effects somewhat similar to these have been noticed with alternating currents .
TOMLINSON mentions ( ' Proc. Roy .
Soc Dec. 1889 ) , that repetition of the cycle reduces the hysteresis , and he names it the " " accommodation\ldquo ; of the molecules , supposing that the molecules become arranged in a manner in which it is more easy for them to reverse , and it is not surprising that with the more uniform motion of a rotating field , the same effect should be observed .
The effect of suddenly diminishing the speed is more obscure ; but in these experiments it seems to be due to a disarrangement and an absence of recombination for a short , giving rise to great irregularity of hysteresis .
The length of time over which the effect lasts is renlarkable , as it often extended to over 1,000 cycles .
It was compJetely stopped by a temporary stoppage of the current , while TOMLINSON found that even slight mechanicaJ shocks were sufficient to stop the effect of ' ' accommodation The effect of reducing the current is probably due to the same cause .
The molecules are arranged in regular order , and ib is some time before they disarrange themselves , while during the process the hysteresis is lower than the normal .
An increase the current is not immediately followed by complete arrangement of the molecules , and during the process the hysteresis is too high .
On breaking and remaking the circuit after a sudden reduction in the current , the deflexion of bhe armature , previously lower than the normal , rises to a value higher than the normal and then drops to its normal value , showing that by breaking the circuit , the regular arrangement is completely disturbed .
Professor EWING in his book Magnetism in Iron \amp ; c. ) mentions that rapid makes and breaks increase the permeability , but increase still more the residual magnetism , Diagram 11 .
so Spoed ln Second .
but he does not give the hysteresis effect , which may or may not have beeIJ increased .
This is clearly a case of\ldquo ; accommodation When , however , the current is reversed , the increased amount of combinations renders the iron less permeable with more " " coercive force necessary , and therefore with presumably more hysteresis .
The result does not seem to agree very well with that of TOMLINSON'S .
In the rotating field it is possible that rapid and breaks tend to disarrange the symmetrical arrangement produced by co1ltinued rotation , and to the IRON AND EL IN A ROTATING MAGNETIC FIELD .
number of combinations .
This wouId account for the increase in the hysteresis , and also for its irregular value before it again reached its normal valne .
[ It will be noted tlmt while the average values of hysteresis obtained in the Tables X. to XIV .
agre for the most part very closely with those in the previous tables ; the values at the maximum are distinctly lower , both in iron and steel , indicat , ing a greater " " accommodation\ldquo ; under the long continued running in the critical condition.\mdash ; 2nd September , 1896 .
] In the hard steel these effects are very much diminished , the deflexions being more uniform and consistent .
The same difference is shown in the constancy of the hysteresis at the maximum point .
In the steel , all the curves agree very closely in the maximum value attained , Diagram 4 , and the speed curve at that point , Diagram 11 , although somewhat irregular , is much so than the curve from soft iron .
This foll ) from the supposition that in steel the aggregations of ] ecules are smaller and less dependent on each other , and hence the average behaviour is more regular .
ABLE X.\mdash ; Hard Steel Armature of 147 Plates , of thickness 0141 centim. Variation of Hysteresis with Speed ; Diagram 11 ; Deflexions on Scale : 1 division ergs per cub. centim. per rey .
*Value of spring 1 division ergs per cub. centim. per division .
TABLE XI.\mdash ; Soft Iron Armature of 250 Plates , of thickness 0081 centim. Variation of Hysteresis with Speed .
TABLE XII.\mdash ; Soft Iron Armature of 250 Plates , of thickness centim. Variation of Hysteresis with Speed .
* This series was repeated later , and the spring was not calibrated .
It had evidently changed in strength or method of applying restoring force .
IRON AND IN A ROTATING MA .
GNETIC FIELD .
TABLE XITI.\mdash ; Soft Iron Armature of ) Plates , of thickness.0081 centim. tion of Hysteresis with Speed ; Deflexions .
NDCCCXCVI .
\mdash ; A. ABLE \mdash ; Soft Iron Armature of 156 of thickness centim. Variation of Hysteresis with Speed ; Diagram 17 ; Deflexions : 1 division ergs per cub. centim. per revolution .
The foregoing experiments were carried out in the electro-technical laboratory in University College , Liverpool .
My best thanks are due to Mr. E. H. MORGAN , my assistant , for his help in the constluction of the apparatus and the performance of the experiments .

