PREFACE, This work is igteaded primarily as 4 texthook for graduate students in electrical engineering and 23 a reference bovk for engineers concerned with Une development or design ef new mayuetic devices. A fauiliac knowledge of the subject matter generally taught in undergraduate , enurses in electrical engincering is assumed. The book is divided into two parts: the first cight chapters in which the fundancntel background theory and inethods applicable to all types of magaolic and non-rotary electromagnetic devieos are developed, anci the fast six chapters in which these principles are applied to the solution of 9 variety of probioms. ‘These solutions have been developed in general terms, and each is followed by Uhe detailed nuneri- ez] solution for © particular eet of data. It has been my experience that such apecific and detailed solutions arc invaluable to students whe must study by themselves or away from an instructor, The accurate solution of practical electromagnetic problems’ his always been dificult, owing to the affects of magnetic leakage and the nonlinear relelionship between magnetic ux and magnetic intensity in ferromegnetie materiuls. Special methods have been developed to handle magnetic lealage, and great emphasis has been placed on graphi- cal and step methods fo: the: practical solution of problems mvalving the nox-lincurity caused by hysteresis und saturation in iron, and the variation of wagnetic leakage with motion, The material of this book has heen presented to graduate students at the Stevens Institute of fechnology fo. soveral yeurs, both in the day sehool and in an evening iveture course cavering a school year (one two-hour lecture and feur hours of tiene preparation and problems per week). A inrge amount of quantitative datu is presented, both on magnetic materials and clectromegnets, many of which are original and have been obtained by painstaking taboratory work done at Siavens hy graduate and undergraduate students -working on these projects, and also by me. Thanks are due Stevens for providing the apparatas, facilities, and specie) inugactic equipment required. ‘This work was undertaken in 1932 at the suggestion of Dr. AMan Hareltine, formerly head of the Department of Electricat Engineering + :vi PREFACE, and at the present time Professor of Physical Mathematics at Stevens. Many of the methods presented represent an outgrowth from some original work on rational methods of magnet design done by Dr. Hazel- tine. I wish to acknowledge the scheme of “estimating permeance” of Chapter V, and the design methods of Chapter X, as coming from. this source. My thanks are also due Dr. Hazeltine for the many helpful suggestions and frequent time spent with me in consultation and check- ing some of the more difficult parts of the manuscript. . Many of the original problems and inethods used. in the latter chapters are an outgrowth of my consulting practice. I am indebted to the Fairchild Aviation Corporation for the origin of many of the problems on dynamic characteristics and the models built. and tested to check the computed results. My generous thanks are due the many instructors in the Electrical Engineering Department at Stevens who from time to time have assisted by making drawings, computations, and suggestions; to my colleague, Professor William [.. Sullivan, for his criticism and helpful suggestions in connection with the development of the step-by-step methods of Chapter XII; to Mr. Harry W. Phair and Mr. Frank W. Stellwagen of the Fairchild Aviation Corporation for their assistance in correcting and criticizing the manuseript;- and to my wife, Mary D.- Roters, for her assistance in correcting the proof and preparing the index. Hexerar ©. Rorens Jauares, N.Y. January, 1941CONTENTS I. Invzrouneran . 4 TT. Maonerse Proserris or Inox ano Some ur das Auuors 2000 A AML. Tae ‘Tupory ce Ovematio of Ecctanmacnets ane tx TAcTons. Esromna nero Toe Hperewoa Desay 20.2.2... 0020 ceeeeee 3 TY. Cancutarios oe Macwerer Cieewirs Cantarnexu Teor snp Ant Caws op Kxown Praacaxen . cee AE VY. Causnanon ex vie Pemeraxee or FLox Paris tovcr Ar m- ween Suaeaces of Hreu-Peunccuuity Manenat .... YE. Cons VIL. Heerenc of Macwer Cars, we - VIIL, Magiatie Fonces EX. Caanieneosenss ar Teactive Magnets: Setection or Bust Tire 708 4 Srecme Dury fees . 8 Xf. Draran or Tracrve Macvers: Dusten Puocunone: Tustisraarive De. aInns. - : . 2s XT. Tr Draven Mauwers : cece eee 96 XI. Amu-Grean Msarars wees ce Pike XTEL, Atreewarine-Connmy Macnera : eT) XIV, Reaves 0. ceceeeceeeee APPENDIX... a Trem ees wees 533ELECTROMAGNETIC DEVICES CHAPTER I INTRODUCTION 1. Genet ‘This book is intended primarily for those who will be regpoasible for the design and development of new magnetic equipment. Developments in this field have been very rapid, and the present indications ere that: they will continue to remain so. The sizpianc alone, particularly the military plane, hes required tho development of mauy sew elvetro- devices for ins\ruments and control. Two reasons fur the sity of control deviecs ure Uke general inaccessibility of the equip- ment, which makes remute control desirable, and the severe reqairenvsnts of fuictional operation of this type of equipment. Mthough the averape electical cugineer is not necessarily interested ign. This is said becouse an intelligent, undesstancing of the operatiun und limitations of electromagnetic machmery and devices must be hared upon a knowledge of the fundamental laws of electromag netisn, elcetromapnetic force relations, and the limitations of commer- cially available imuterials, Successfirl development work in any Held cannot be predicated upon. snpaicial knowledge, but in every justance must depend upon a sound. understanding of the fundamentals of the field, coupled to an imagination toned down by practienl expericuce, Though itis true that experieave is of inestimable value, it can sometimes Le averrated. When dealing with the mathomatically ampredictablu, sech as human emotions and rue tions, experience, mellowed by maturity, is the essential vequisite. Wher dealing with av inanimete object which fumetions in direct response to the laws of mechsnies and electricity, an exact, quantitative knowledge is a requisite if'on cnelysis or design from ouly factional point of view is required. If the particular object ix to be Practically and e¢o- nomically sppled, experienes does become un casential factor. An exacl quantiiative knowledge of a subject can be gained ouly by breaking it dewn inta ifs cascatial component parts, studying aud ias- i2 INTRODUCTION {Cnar. 1 tering these, and then putting them together in their proper relation and perspective. This book has been written with this in mind. In the following articles the subject matter has been broken down by chapters to show the essential parts and their correlation. 2. Properties of Magnetic Materials On this basis the first important subdivision is a knowledge of the materials available for building magnetic devices. Tremendous advances have been made in the last few years both in the number of types of magnetic materials available and in the improvement of their properties. ‘These recent advances ate discussed in the latter part of this chapter. ‘A full discussion of all the properties of magnetic materials which are important in predetermining the complete magnetic performance of direct- and alternating-current electromagnets, transformers, polarized chokes and transformers, permanent’ magnets, ete., is presented in Chapter TI. A very complete set of data, many of which were takén especially for the purpose at the Stevens’ Laboratory, covering all the materials commercially available, is also presented. 3. Factors Entering into the Efficient Design of an EFlectromagnet In actual: operation, because of the transient and unsymmetrical nature of the magnetic cycle, the electromagnet is very complex. ‘The energy changes occurring during a complete cycle include a storage of magnetic energy, mechanical work done, hysteresis energy loss, disgorge- ment of magnetic energy in the form of a spark, and finally the energy associated with the residual force. The efficiency of a magnetic device in any particular application depends on how skillfully the designer can control those energy changes which are desirable and mitigate those which are undesirable. Chapter III presents a complete discussion of all these energy changes, based on theoretical considerations and actual experimental data. 4. Magnetic Circuit Calculations Predetermination of the magnetic saturation occurring in an electro- magnetic device, or the available magnetic potential across an air gap, depends upon an accurate analysis of the magnetic circuit. Magnetic circuit calculations range from a simple series magnetic circuit without Ieakage, to the more involved cireuits with distributed leakage, unsym- metrical parallel cireuits, polarized cores involving anhysteretic mag- netization and incremental permeabilities, and residual-flux predeter-Are 7] TEMPERATURE HISE 3 mivation fer both suft and permanent miegnet cores. All he various types of maguetic circuit calculations are covered in Chapter LV. &. Permeance of Air Paths ir, or the ‘The calculation of the permeance of dux paths through y more general pwhlem of the gelerminatioz of solencidal or lamellar fields, is a problem which, except for the more simple configurations, has defied the efforts of mathemaliziaus for canturies, A two-dimensional field emanating from continuous cylindrical studaces ean be evaluated analytically by # sudlicienUly astute mathematician. A two-dimensional field with discontinuities bas been evaluated for the speeial ease of inter- secting, plane surfaces. In practice, however, simple two-dimensional fields do not occur. The actual field is always thrce-dimensionul with several discontinuities in the form of comers and edges, With patience, two-dimensional fields cam be handled by the method of field mapping; but three-dimensional fields, unless they con be considered bo vary in only twa dimensions, are difficult, if not imposible, to hunlle. Accuraey in the predefermination of the force of a maguet is, to a very great messare, dependent upon the accurate evaluation of airpath per- meances. [y Chapter V, » heuristic method of evaluating these per- meances is develuped. The effieacy of any meLhod of computation must. ultimately depend on the acvuracy of the revulls obtained and the rela~ live effort required. Numerous examples of computed results checked by experiment au 6. Coils for, Magaets ‘The heart of any cleelrosmagnetic. deview is the exciting coik The eonstruction of a coil, aside from the choice of wire size, is purely & mechanical mutter, Saiicieat space must he provided for the requised munnber of turns, sufficient insulation to withsland the highest prebable surge voltage, and 2 mechanical construction whieh will insure rugged- ness, strength, high space factor, aud low cost. Commercial winding pariiee dictate: cofl construction, winding tolerances, insulation, et Chapter VI pives complete data om winding practice, and also the method of computing coil performance. 7. Temperature Rise Heating, as applied to average-sized electrouagnets ur obher stall non-moving cleetromagnetic deviers, is discussed in Chapter VIL The" actual Lemperature 1sc of a body as a function of time due to the internal4 INTRODUCTION [Cnap. 1 evolution of heat is complicated, depending upon its thermal capacity, thermal conductivity, and the heat-dissipating capacity of its external surfaces. The latter depends upon the nature of the radiating surface, its temperature, and whether or not the air surrounding it is still. The problem is first treated from a theoretical point of view under ideal conditions, and the results so derived are correlated with experimental data, In this manner, not only have the proper empirical constants been derived, but also the limits of validity of the various theoretical formulas have been determined. 8. Electromagnetic Force Formulas ‘The actual manner in which the work of a magnet will be obtained, that is, a large force through a short stroke or a small force through long stroke, depends on the shape of the working pole faces. Force formulas, with all their limitations and corrections, and for all manner of pole-face shapes, and types of action, are derived in Chapter VIII. 9. Magnet-Pole-Face Types ‘The proper type of pole face for any particular maghet depends on the relative values of the force and the stroke. The use of the right pole face results in the least.weight for a given work, Chapter. IX discusses the force-stroke characteristics of the various types of pole faces, and develops complete data for determining the most economical pole-face type for u given force-stroke characteristic. 10. Design of Direct-Current Electromagnets ‘The actual procedure of designing a magnet to meet given specifiea- tions is, like all machine design problems, an essentially heuristie pro- cedure. All types of magnet, while essentially the samo in fundamentals, differ in the detailed procedure necessary to get a workirfg design. The amount of labor necessary to get a satisfactory and cconomicul design depends on how elose the first choice of basic design factors is to the optimum values. ‘These optimum values, for any particular type, vary with the ratio of the foree to the stroke, In Chapter X the criteria for an optimum design, the procedure of design, and basic design factors are ‘detailed in such a manner that an optimum design can bd obtained. with rapidity for any one of six basic types of electromagnet.a Aer tah ALTERNATING CURRENT MAGNETS U, TimeDelayed Magnets In many instances it is necessary to provide for a time-delsy action inaneleetromaguet. Thus, if is sometimes desired to delay the exeeation of the work of the maguet au appreciable time after the energizing impulse has been initiated. Whea the time delay is short this can be incorporated in the clectrical design of the magnet ilself. When lung tine delays are necessary the delay mechauista must be external to the aaguct. The problem of the predetermination of time delay can be handled mathematically with good results, provided that the inductance of the magnet remains substantially coostaot when the current varies, and there is nv uotion of the armature during the time-delay period. When these ennditions do nat ocenr, the solution must he abtained by step-by-slep methed applied to the differential cqaation for the cireuit. Coniplete data on time-delay methods, with their mathematical solution, and the conlation of the mathematical solution with experimental results are given in Chapter Xf. 12. Quick-Acting ar High-Speed Magnets The problem of predetecmining the time of action of a moguct, though probably one of the most difficult problems in megnes design, is.also the mast fascinating. it is of particular interest in high-spoed magnets, where the shortness of the thue required for the magnet to complete its work is the yardstick of mezit. Such magnets ate used in many modern cantrol devices. ‘The problema ig one of electrodynamics, involving two variables, pace aud time. The partial differential equa- tions that result, because of their now-caustant coufGelents, must be solved hy actép-hy-step integration. Complete details of the method of sulution, sud tle correlation of the compnted and experimental results. for an actual magnet, are given in Chapter AIL In addition, a rational method of developing a preluminary design for short-stroke high-speed magnety to mevt required epecibeationa is developed. 43, Alteruating-Current Magnets ‘Vhe aliernsting-current maguet differs from the direct-current mag- uct, priicipally because the current ‘through the ecil is not determined by the resistance of the cvil, us it is in a direct-current maguet, but instéad is limited by the self-induecd voltage produced by tha alternating flux. As the ikix linkage does not vary with the pouition of Uhe plunger, the curnéit de6 INTRODUCTION [Cran 1 current magnet to have an odd-shaped foree-stroke characteristic. Aside from this, the fact that the flux is alternating produces other effects. Thus the force pulsates from zero to a maximum twice during each cycle of supply current. The average force, however, bears the same relation to the current as with direct current. In some applications where it is undesirable to allow the force to fall to zero it is necessary to split the working flux into two parts which are out of phase. This is generally accomplished by means of a shading coil. Alternating-current power magnets are, in general, undesirable. Their volt-ampere consumption, which bears a definite minimum rela- tionship to the frequeney and the work, is high compared to that of the corresponding direct-current one. ‘The iron cireuil must be thoroughly laminated to prevent excessive eddy-current losses. Their one redeem- ing feature is that they ate inherently high-speed magnets. They are extensively used as relays and for applications, such as electromagnetic hammers, which depend 6n the alternating-current characteristies. For power magnets it is common to use a direct-current magnet powered from a rectifier. Chapter XIII presents a complete discussion of the alternat- ing-current magnet from.the above point of view and complete details regarding their design. 14. Relays ‘The relay, es its name implies, is a device intended to repeat or relay: In an electromagnetic sense it is a small-work magnet of low power consumption for closing contacts. However, from the point of view of design, the relay is very general and embraces all the problems of special magnets. Thus a relay must always be designed with respect to its residual foree which determines the current at which it willvelease; in addition, it must often be designed with regard to its speed of action, or for a definite time delay. When operating on single-phase alternating current, shading coils must be used to prevent chattering. Power con- sumption, rather than heating, is the general design limitation which distinguishes it from a tractive magnet. Relays are discussed in Chapter XIV. RECENT ADVANCES IN FERROMAGNETIC MATERIALS It is cafe to say that the only advance in the design of magnets in the last few years has been made because of new ferromagnetic. materials * ‘Extremely interesting discussions of new developments in this Geld and their commercial applications may be found in the following references: V.E. Leao, “Survey of Magnetic Materials end Applicatiousin the Telephone System,” Bell System Techaioal Journal, July, 1939.Awe 1h SOCT MAGNETIC MATHIUALS 7 whigh have been developed. In sarna instanres, with Permalloy and Alnicn, Uhe effeci has been so great as te change ihe whole course of design. 16. Soft Magnetic Materials Tn the field of the “soft magnetic” materials 2 complete exploration af all tha possible alloys of iron, nickel, and enbalt has resulted in the production of many uscful alloys. These alluys inay be divided into three gramps: 1. Ferronickels: iron-nickel alloys which have very high initial and maximum permeabilities aud very low hysteresis loss. 2. Herrocobalts: iron-cobalt alloys which have a high permea- bility that enduzes well bayond the saturation limit of ordinary iron. 3. Constunt-permenbikty alloys: iron-nickel-cobalt alloys which have permeabilities that are invariant within limited renges of Amr density, and 2 hysteresis loss whieh is zero or negligchle. In the field of electromagnet design the first two of these are of spacial interest. Ferronielel (17 per cent) is of particular use in dhe design of sensitive relays. Tn this application it affords two important properties: a relatively high flux density with a very. sual! magnetizing » force, and a very small residual effect. In other applications the low luysleresis loss and high resislivity of ferronickel are very useful a3 they allow the construction af special devices, such as voltage generators and transformers, which must have neyligible losses. Ferrocubalt is chiefly useful in tractive magnets where a high density of force is desirable and in other applications where its high saturation density results in a sub- slautial reduelion in size and weiglit. It also has a high ineromeutal permeability at high polarising flux densitics. Constant-permeability alloys are particulatly suitable for nse in eireuit elements in which dis tortion aud energy locs must be a minimum. Jn the telephone industry, thesy waterials ace known as Permaliny, Permendur, and Perminvar, respectively. ‘ED. Yersen, “Magnetic Materials and Preparation,” MaGrae- 331 Book Co., 1937. This is the fourth chopter of a book catitled “lntroduction to Ferro- soaguotiom” by Fraxers Brozea. 0. E. Wren, “Rrseat Developments in Magnetic Meteriats," Jounal of the Institution of Mledrical Beuyinerrs, March, 1938. ‘These artickes 4 ery extensive and have comptete bibliographies of ali the iugpertant references in the literature.8 INTRODUCTION (Curae. I 16. Tractive and Residual Effects in Commercial Soft Magnetic Materials ‘As tractive force is a very important consideration in the design of eleclromagnets and relays, Fig. 1 is presented in order to give a definite picture of the relative tractive effort which can be obtained from the various special alloys coramercially obtainable? From these curves it is apparent that ferrocobalt is superior to ordinary iron in producing a tractive effect. Unfortunateiy, except for special applications, the price of ferrocobalt is prohibitive. ‘The 34.5 per cent. variety is much less expensive than the vanadium variety and is 400, - = Bi. 47% Ferronickel — Sample 11 Chap. i lz nighSiticon Steel — = 9 = “| J3. Swedish Chatcoal ton—= 1 | la. 35x Fervocobalt —- » aw =| Fal Tt] 5. BEV, 49% Fenocobalt—" 21 ~~ Lo at LY et 2 Lett Lr lett L LW LAS | Its Fi [ t oes 1 fl 1 i 100 1680 Magnetic intensity, ampere-tuens per Inch Fia. 1. Curves of tractive effort as a function of magnetic intensity for various magaetic materials eommercially available. Data for Curve 5 taken from Bell ‘System Pecknical Fournal, Vol. XVHJ, No. 3, page 438. such easier to fabricate. ‘The addition of vanadium is necessary if the product is to be rolled inte sheets and also is advantageous as it increases the resistivity. On the other hand, where it is desired to produce a Iarge foree with very low magnelizing currents the ferronickel is shown to be superior. In relays and sometimes in tractive magnets, the residual force pro- duced by the coercive intensity of the material is important. This effect can be compared only by an examination of the demagnetization curves of the materials. Figure 2 shows the demagnetization curves of the alloys of Fig. 1,? except that for sample 12a which has been replaced by sample 125, : > The samples referred to in Pigs. 1 and 2 are described in Art. 33, Chapter TT.Awe ay IMPORTANCE OF IMPURITIES 9 In order to shaw the effect of the ccercive inteusily in producing dual Cractive force, the foree curves, 4 and &, ha've been drawn to the Hight of the axis of ordinates. Curve A shows the residual praduecdd when there js an ideal gap of zero length in the magnetic cireuil. The faree values are obtained hy projecting, Ferizontally from the residual flux densities, on the axis of ordinates 19 Curve A as shown by the dashed Lines, In actual practice, however, there is always an air gap in the maguetic cixeuit. This airgap leagth being taken as one-one thou sandth park * of the iron lengeh, the uir-gup pormeance line OC may be drawn. ‘The intersection of this linc with the demagnetization euzves gives the actual air-gup flux densities with the air gap present. Projecting bso LAU betrovichel ~ Sar 2, High Silican Sleet— 3 Swecich Chateau Tea 135x Ferrocobatt S28 V, AOR Feetecobalt- fe 1 hsp. | 244 Y b AL 8 «3 = + oo 40 GF 8 109 170A Cocscive: Magnrtic Intensity, 0 1 2 28 ‘amparc-tums per inch’ ~ Force, Wb wersg. in Fic. 2. Demagnotisation carves and vesidual iisetive fort for various riugnetia materials commercially available, these intersection points on the furee curve B, by the borisontal Enes showa, the residual forces are obtained. To make the comparison more vivid these forces are tabulated below. | ‘he advanlage of ferroniekel and higb-silicon steel in producing 2 smull residual effect is st shown. 17. Importance of Impurities in Soft Magnetic Materiats ‘rhe beginning of Uke development of magnetic inuterials may. be daied at opproxitnately the year 1900 when invesligators attempted to find the cause of aging im tratisformer iron. In those days the hest iron +Thin ia about the smallest gap length ‘at is commercially practicable in the ucual design.10 INTRODUCTION [Cuan 1 Resrovat Force Foutowine Excitation ro SATURATION Residual Force, {b. per sq. in. No Air Gap | 0.5% Air Gap 47% ferronickel. High-silivon steel Swedish charcoal iron 35% ferrocobalt. ... 2% V-49% ferrocobal.. available was Swedish charcoal iron. Owing to the low resistivity of this iron, the eddy-current Joss was high, which when combined with the hysteresis loss produced a relatively high total core loss. I6 was found that the temperature rise of the iron due to its losses when used in transformers caused the core loss to increase rapidly, sometimes doubling. it in the course of a few months. The only cure was to disassemble the transformers in order that the iron might be reannealed. Attempts were made to discover the cause of this effect, known as aging, and while there were conflicting theories, the cause has remained obscure until recently. However, practical progress was made, and in 1900 Barrett, Brown, and Hadfeld brought forth their silicon steel. By alloying the charcoal iron with 25 per cent silicon they found that aging disappeared, hysteresis loss decreased 25 per cent, and the resistivity was increased fourfold, greatly reducing the total core loss. Since that time considerable study has been devoted to the iron-silicon alloys. ‘The commercial product of today is far better than Hadiield’s best laboratory alloy, ‘The interesting feature of the reduction of iron loss by alloying with - silicon is that the entire effect is now definitely traced to the removal of minute quantities of such impurities as oxygen and earbon by a process of absorption, which, though not actually removing the impurities im the physical sensc, renders them impotent. Research work carried out during the last few years has indicated - that absolutely pure iron is an ideal magnetic material, having zcro hysteresis loss and infinite permeability. Actual experiments on single crystals of very pure iron have shown relative permeabilities of several hundred thousand and extremely Iow hysteresis loss. Iron of this exceptional quality-has been produced only in the form of Single crystals. ‘The final removal of the last traces of impurities isApe 18] STRUCTURA AND GRAIN ORIENTATION i accomplished by long annealing at high temperatures in an alnosphere of hydrogen. This not only reduces the ordinary oxides but also climi- roviass Lace of exthon, sulphur, and phesphocus as gases (CH,, HS, PH). Iron specially prepared and treated in this way has chewa (otal Ungurilics of due order of only O41 per cent. Commercially such material is at present impractical, owing to its high cost end the diffoully of producing it in the form of rods and sheets. Riven if it conid be produeed it would not ba desirable for alter- nating-current apparatus beause of its low resistivity and consequently high eddy-current loss. For these reasons alloys using silicons, or some similar material sueh as aluminum or vanadium, which miiigate the effect of the residual impurities of commercial iron and at the same time produce a high resistivity, anest still be used. Other very important factors in dhe produetion of iron of super megnetio quulitics are grain size and grain oleutation. It has been definitely proved that as the grain size is nereased the hysteresis loss is deescescd. Furthermore, if the iron is xolled gu Ulat the axcs of the grains have a parillel orientation a3 compared to # naudom orieatation, a materist of exesedingly high permeability and low hysleccsis loss is ebtaived. This latter procass, known as “fibering,” has the effect of giving the sheet magnotie properties resembling those of a sfagle eryelal. Such materials are commercially svailable+ in the esse of ferronickel and ferreeobalt, prolonged annealing at high temperatures in an alinusphere of hydrogen has shown a remurkuble impzovement in their magnetic properties. Fifty per cent ferconickel when trested in this way has shown anaximuin permeabilities over 150,000 2s compared to about 40,000 for the usual product. his product is commercially available under the name of “Hipernil Likewise, 2 par cent vanadium Permendur when specially hydrogen- annealed shows « large inercase in permeability and slight increase in saturuticn density. Trad ances 18. Structure and Grain Orientation uf Magnetic Materials ‘The acccpled expianation of many of the effects just doserihed is based upon the existunce of a definite lattice structure of the erystals of the material whieh imparts to the erystal an axis of easy magnetien- bility. Any portion of the inatter having 2 continuous Inttice strneture and a definite orientation Uhroughout is called o gruiv. An ideal magnet saatonal will have regularity of the Iatlice sirmeture throughout. As the regularity of the lace struct is interrupted st the boundaries of the grains, anything which inereases the nmaber of grains or dheir randomrr INTRODUCTION . (Crap T ovientation is undesirable. Likewise, anything which inercascs the size of the grains and decreases the random orientation of these grains is desirable. Tmpurities in the metal tend to make the grain size small, as they constitute the focal points from which crystallization starts. Thus small particles of oxides and carbides, when present in solid solution in the metal, will cause the crystallization to start at many points simultane- ously, producing small grains. Consequently, the total absence of impurities is essential if large grains are to be produced. Another factor controlling the grain size is the manner in which the metal is worked and the subsequent heat treatment. Considerable suecess in producing large grains has been obtained with silicon steel by cold rolling followed by annealing. If this is done in a special way, a preferred orientation of the lattice structure of the grains ean be obtained which results in a magnetic material having properties approaching that of a single erystal.* Cold working of a material will introduce intemal strains or lattice distortions and’ tend {o make it hard. Annealing by heating to a high temperature with very slow cooling produces a homogeneous structure free from lattice distortion. Thus magnetic materials, which in the process of fabrication have been stressed beyond the elastic limit, should be carefully annealed. “ 19. Hard Magnetic Materials Advanees in this field also have been astounding in the last. decade. In contrast to what has just been said about soft magnetic materials, the problem in producing a hard magnetic material is to introduce lattice distortion in the erystal deliberately, Such a distortion or strain in the material can be produced by cold working, but where it is intended to produce a strong permanent magnet material it is best done by quench hardening or by dispersion hardening. Wither process depends on the existence of a heterogeneous structure involving the presence of two components, one of whieh is dispersed in # finely divided form through- out the matrix eausing severe lattice distortion. In all the earlier types of permanent-magnet steels, such as carbon, tungsten, chromium, and cobalt, the hardness depends on the formation of a martensitie structure on quenching. ‘These steels have the inherent disadvantage that the martensitic structure producing the lattice distortion is unstable and subject to deterioration from vibration or high temperature. “See Art. 30, Chapter IL.Ane. 19) WARD MAGNETIC MATERIALS B Yn the last feiy years new permaneut-magnet materisls have been developed in which the liitice distortion is produced by a dispersion of the geeond component in finely divided form throughout, the matrix during a carefully contiolled costing process. Such stecls are known us dispersion-hardened alloys. The best-known materials in this group are the nickel-iruealucinua and sickcl-iron-aluminum-cobalt alloys which appear under the names of Mipermag and Alnico TT, for the former group, and Alnico I, II, and IV for the latier group.§ ‘These imuterials are fabricated by casting or sintering. Sintering is a process whereby a properly proportioned mixture of the ingredients of the alloy in 4 Guely divided form ave malded into a solid inctal by a process involving the use of temperatures a few hundred dearces below the melting peint in combination with very high pressures. When the material is cust it must be Gnished by grinding, and when sintered it is formed into molds to the finished size, “Chess dispersion Inedaned ulloys Leve an exseedingly lncge bysleresis loss and henee form powerfal permanent magnets. ‘The enmmerrial Product averages about twice the available magnetic energy of the hest grade of cobalt steel and several times that of the best grade of tungsten, steel, ‘This improvement is so great that it has revolutionised the design of many pieces af electrical equipment. ‘hus many types of small apparatus which formerly depended upon an electromagnetic windiag to prodker « consénct fax ean néw be mare economically designed utilizing Uhese permenent-magact materials. ® Soe Art. 42, Chapicr IL.(CHAPTER I MAGNETIC PROPERTIES OF IRON AND SOME OF ITS ALLOYS 20. Normal Hysteresis Loop and Magnetization Curve The magnetic performance of a piece of iron is dependent on its previous magnetic history, so much so in fact that cerlain apparatus, such 25 some sensitive relays, ete., must be operated in a definite mag- netic sequence or eycle if they ure to give reliable results? This effect is due to a magnetic property of the iron termed hysteresis, meaning a lag; fo that is, the effect lags behind the Pl Le cause, and hence an immediate effect a wap ‘son, Te “|| (magnetic flux) may be due to an im- fe tHe} mediate cause (magnetomotive force) modified by some previous cause of which the observer has no knowledge. aol I It is for this reason that,. before co making any magnetic measurements ‘on iron, it must first be put through er, a very definite magnetic cycle so that Lioa the previous history is known. For Tra. 1. Hysteresis cycle in soft iron 5° Of comparison, all curves giving having no previous mnagnetic history the relation between the flux and (ample 3). magnetomotive force of a piece of iron are usually plotted with flux density and magnetic intensity as axes. Such a curve is ealled a meg- nelication curve, . Consider an iro ring sample which has been thoroughly demagne- tized,* Fig. 1. ‘The ring sample is chosen merely for its simplicity and LJ ? This is particularly true at flux densities weli below the knee of the magnetization curve, ia which range most sensitive relays operate. +A piece of iron is demagnetized by subjecting it to a slowly roversing magnetic intensity, which has an initial value at least es great as any previously applied value, and gradually decreases to zero. This process effectively removes the previous magnetic history of the iron. 4Ane. 20}, MORMAL HYSURRFSIS LOOP 13 convenience; for Use gukw of definiteness and proportion Fig. } has been drawn {rom actual test data taken on sample 3, the data for which are given on pages 47 and 69. As the eveiting magnctomatice force on the ring is gradually inereased the flux density rises on the nary shown by 0-1 of Fig. 1, reaching the Qux density Bp at a maynetic intensity of ffm. Such a curve is called a rising magnotizalion curve. i, after reaching point 1, the magnetic intensity is gredually decreased, the flux dersity will fall along the ; ie Hanes, sai. 3a bation Cee \ \ ftir ‘eemeabiity i 5-H + > cy b OH gat inde 9 409 eid ean te0g 7000 2Hon 2809 3209 3609 Actatiw: Peemoabiity Fig. 2. Normel hysierest: Feats ait earned ciagnetization eurve for soft iron (saraple 3). curve 1-2, reaching the value B, at H = 0. This Gux density By sebained by the pices of iron after being magactiaed ia called the rasaduad flux density, or tetentivity. ‘The eyele is continued hy increasing Hin. the ucgalive direction, causing the dux to deere ‘The value of muguctis intensity H, required to decrease Lhe flux density to zero is called the coercive magnetic ialensity or tha coercive forces. The eyele is completed by allowing the magnetic intensity to decrease to —H, and thence inerease to +H, casing the dux donsity bu full along the curve w+, and finally to rise alung the curve 4-5-6 7. Point 7 aloug the curve 2316 MAGNETIC PROPERTIES OF IRON {Crap 11 will not in general coincide with point 1, being usually slightly lower. If the maguetic intensity is now varied from +H to —H back to +H several times, points 1 and 7 of these succeeding loops will gradually coincide, forming what, is termed a normal or symmetrical hysteresis loop. When the magnetic intensity has been alternated sufficiently between the same positive and negative values to obtain a closed hysteresis loop, whether it be a symmetrical loop or not, the iron is said to be in the cyclic state. The locus of the extremities of the normal hysteresis loops of a material is called its normal magnelization curve. Such a curve is shown in Fig. 2, for the same sample (3) as Fig. 1, along with several of the normal hysteresis loops from which it was derived. It will be noticed that at the higher magnetizations the normal magnetization curve lies outside the hysteresis loop. If the rising magnetization curve of Fig. 1 is plotted on Fig. 2 it will be almost indistinguishable from the normal magnetization curve except in the region of sleep slope, where it is slightly lower. For all practical purposes we shall consider these two curves to coincide for mild steel. Figures 11 and 15 give normal magnetization curves for many commercial materials. 21, Permeability ‘The quotient of the flux density by the magnetic intensity as read from the normal magnetization curve is known as the absolute value of the normal permeability or merely the permeability of the material. If this permeability is divided by that. of a vacuum the relative pérmea- bility is obtained. In references to ferromagnetic materials the permea- bility is almost always given relative to that of a vacuum, and hence, when we refer to the permeability hereafter, we shall mean the relative permeability unless specifically stated otherwise. The permeability curve derived from the normal magnetization curve is shown in Fig. 2. ‘The value of the permeability at B = 0, which is the slope of the tangent to the magnetization curve at B = 0 divided by the permeability of air, is known ax the inilial permeability. It is equal to 300 for this sample, and generally is of the order of several hundred for the ordinary commereial material: Figure 15 shows normal permeability curves for several samples of commercial silicon steels. The mazimum permeability, shown as 3040 in Fig. 2, occurs at that point on the normal magnetization curve where the line OA through the origin is tangent to it. Maximum permeabilities of commercial materials range from a few hundred to several thousand.CNTAL, PERMEABILITY 7 shar 221 22. Incremental Permexhbility: It js very convenient, when dealing with certain Lypes of apparatus whore a small alternating magnetic intensity is superposed on a con- stant magnetic intensity, to speak of the incremental permeavility, which is the quotient of AM hy Al/, AI being in the reverse direction from the i 4 1 A vie Ty uo { sed LET Hy 2) sm osk ghee ee ' ' H F 1 b t t ' i ite q Sau 9b ap ay oo BHO TD hacmazwafle pet fi Bie (pr AB) pees en ot ghee ne Fro. Sa, Curves illustrating ineremeatst permeability and its manner of variztion: Fic. at. Exmpirisal constants for Spooner's formula for incremental permesbility. Repriated hom Spoonsr’s “Piopertion and Testing of Magucla: Materials,” by courieey of the McGraw-Hill Book Oo, change in H immodiately preceding. ‘here is no restriction as to the magnitude af AB or Af, or/as to the position on thé hysteresis loop ox magnetization curve at which they are taken Suppose that in Fig. Sa a direct polariring magnetic inteusity of1s, MAGNETIC PROPERTIES OF IRON ICnar. 11 Ho is impressed producing a constant flux density Bo 2s shown on the normal magnetization curve. Let there now be superposed upon Ho a sinusoidal magnetic intensity Hg such that the resultant magnetic intensity will vary between the limits of H, and H2asshown. This will cause the flux density to rise along the normal magnetization curve (saine as rising magnetization curve) from 0 until point 1 is reached, where the magnetic intensity is at its maximum value Hy. At 1, Ha commences to decrease, causing the flux density to fall long a normal hysteresis Joop until point 2 is reached, where the magnetic intensity is at its minimum value Hz. At 2, H, increases and the flux density rises along the dashed curve to 1 again, after which the minor hysteresis loop 1-2 is traced over and over. It is therefore apparent that as far as the change in flux density caused by Ha is concerned the apparent absolute permeability of the iron is given by the slope AB/AH of the axis of the minor Joop 1-2. This permeability is known as the incre- mental permeability, and it is usually given relative to a vacuum. It is of chief interest in the design of chokes and trafsformers carrying direct current, If Hz is increased so that the resultant magnetic intensity varies between Hs and H;, the incremental permeability will be proportional to the slope of the linc 3-4 as shown. In general for a given biasing magnetic intensity below the knee of the normal magnetization curve the ineremental permeability first increases and then decreases as the magnitude of He is incrcased.. If Hq is kept constant at some value which is not very small, the incremental permeability will decrease as the polarizing flux density increases. This is because the tops of the hysteresis loops become flatter with increasing magnetizations, as can be seen from Fig. 3a. If Hg is allowed to approach zero the incremental permeability is proportional to the initial slope of the descending branch of the normal hysteresis loop corresponding to Ho. This pxrticular incremental permeubility is called the reversible permeability. Because the deseend- ing branch of the hysteresis loop has less slope as the magnetization is d the reversible permeability is always less than the initial per- meability. The reversible permeability is of particular interest where the superposed magnetic intensity is very small, as oecurs in audio- frequency transformers. As AB is usually the independent variable in most applications dealing with incremental permeability, the values of incremental permeability are always plotted as a function of AB instead of as a function of AZ. When dealing with incremental permeabilities it makes considerable difference whether the variation in flux density or the variation in mag~Aree 2h TNCREMENTAL PERMUEAHTLEGY ce) petic intensity is Lhe independent variable. In Vig. 3a, Ef bas heen made the independen’ vannble, and it will be noticed that the impressed sinu- soidal vaniution Hy prodnces a flux density variation By which is aot sinusoidal Furthermore, 2, is not superposed upon the flux density Bo Laut would be produced by the polarizing magnetic intensity Ho, but upon @, which is greater than By, depending upon the magnitude uf Ho. ‘Thus, when Hi, is the independent variable the valuy of the polarizing flux density By is aot easily determinable, Likewise, the wave form of the induced voldage #, will he very camplex compared to that of He. TT, will be the independent variable in constant carrent cireuits and in some resonant, sircuiis. “Thus if an inductaace is connected ia the plate cironit. of a scrcen-grid tube, the alternating component of plate current: is determined almost entirely by fhe applied grid volts More often, however, the Gux-dumsily varintion is the independest variable. This is true in all constant- or substantially constant-vollage Gireuits; thus in a transformer supplying a half-wave rectifier the varia tion in flux density is forced to be sinusoidal because the supply voltage js sinusoidal; in a Biter choke epecating from a low-impedaace tube the wave form of voltage across the choke, and hence the Aux variation, is ako determined eatiruly by the wave form of the supply voltage. The discussion so far might lead one to believe that, the incremental permeability is defined if Lhe mapnitnde of AR or AH isspecilied, together with either the pelariziug magnetic intensiby Bo, or the polarizing {ux density Ry, This, howover, ig not true, and in order that the incrementat permeability be dcierminate it is necessary that the wave shape of 42 or 4H be defined in addition Referring to Lig. de, if @ sinusoidal variation in maguctie intensity, equal to AH and polarized by Me, is-impressed on the magnetization eucve shown it will produes the wave of flux density variation d by B, lnving « total pulsation of AB asshown, The mereinenial permes- Libity will then be designated as AB/ALL. Ii, however, the vile of pulsation AN, alternating, about the same polariing ini Hp, but net sinusoidal in shape, iy impressed us the maguetization curve itswill produce the wave of fncdensity variation desimnated as BY having, a lotal pulsation of AB’ a3 shown. The incremental pernicability yf the sample will be designated as AW'/AM, Ibis quite obvious from Me figure that these twa vaiuss of incremental permeability obtained for the sama specified value of SA and Hq are nol the same. The same eonchision would be reuehed if AB and B, were specified in magnitude. It therefore follaws thet if the incremental permeability és o be determinate the wave shape of either AB ar Af must be specified besides the value of polarizing flux density or magnetic intensity.20 MAGNETIC PROPERTIES OF IRON [Cane 10 The most useful definition * of incremental permeability is that where the pulsation in flux density is maintained sinuéoidal, and where Hg is specified, thus: Be = a (4B sinusoidal) ‘The sinusoidal wave shape for AB is the most common one, as most, apparatus operates on an impressed voltage of sinusoidal wave form. This will produce a sinusoidal flux wave if the circuit resistance is small compared to the reactance, as is usual. Likewise the value of polarizing -cur- rent, and hence Ho, are generally known, whereas the value of the polariz~ ing flux density is seldom known. Figure 3c also shows that the polarizing flux density is dependent on the wave shape. Thus B, and Bi, for the sinusoidal and distorted waves, re- spectively, are not the same even though AH and Fi. 3c. Curves illustrating the effect of ivave form /Ho are the same. ‘on incremental permeability. Wave-shape distortion is appreciable only for the higher values of AB. ‘Thus from 0 to 3 kilomaxwells per square inch, which covers the operating range of ‘audio-frequeney transformers, the ineremental permeability is independent of wave form. In Fig. 19 are shown curves of the incremental permeability plotted as a function of the maximum superposed sinusoidal alternating flux devsity AB/2, for various values of Ho. ‘The data are given for samples 7, 8, and 9 of low-, medium-, and high-silicon steel, respectively ‘These curves show that the incremental permeability increases with +L. G. A. Sim, “Incremental Permeability and Inductance: ‘The Role of Wave- form in Measurement,” Wireless Engincer, Vol. 12, 1935, Nos. 136, 137. . “The data shown in Fig. 19 were taken by Charles Rouault, and checked by Igor Bensin, as a thesis project at Stevens Institute under the supervision of the author. For a complete description of the experimentsl method and a critical discussion of the results see “An Investigation of Incremental Permeability,” by Charles Rousult, Stevens Institute, June, 1939, and ‘An Improved Method of Measuring IncrementalAwr. 22] FNCREMENTAL. PERMBARITITY 21 superposed nifernating Sux density for any eonstanL value of polarizing snagnetic intensity until (ke maximum dux density reached in the cycle pastes tho knee of the saturation curve. Vor any constant value af superposed aliernating ffux density the incremental permeability dearcases as the polarizing magnetic intensity ineceases. ‘Whore no experimental data are available, the incremental permea- bility ean he approximated by an empirical method if Use aormul magne tization or permeability curve for the material is lenown. ‘This method, developed by Spooner, is based on Lhe experimental inet Ubnt, if the incremental permeability is plotted as a function of the flax pulsation AB (the maximum tnx density of the tip of the minor loop being held constant}, the result will be very close to 1 straight line which con be represented by (he simple equation naa ath x AB The incremental permeabilily is iuther fouad to be # funetion uf the fux density of the ton of the minor loop farthest removed from zero. ‘Che resulting formula for incremental permeability is: Ka = #5, (a+b X AH), io) where 4p, is the noruml permeubility corresponding (o the top-of the suiuor foop farthest removed frum veru. AB is the amplitude of the total thnx pulsation of the minor loop expreased ic Kilomaxwells per square inch. © and bere constants which have beea determined experimen tally from tests of various magnetic materials. ‘Fhe values of these constants are given in Fig. 3b. Permeshility,” by Tear Densiu, Stevens Tostitate, Suse, 1940, Briefly, Mr. Rousult excited a theroughly laminsted ring sample With a vatiable sinusoidal voltage derived from a conmnersia power Kae through a varisble auto-transformer. in series witht the oxeilinys cirenit, a low-rceistance souree of direct curremt for polarizing and standard resirtance wore inserted. ‘The tutal nehtanes drop to aerating current waa never allowed ty exeued 10 per cent of the irepressod voltoge in order to avoid distortion of tbe itapressed sinucotdal wave form. ‘The zomzaitude of SB was com- puted from: the measured rant, value of voltage induced in un iaulated coil wonad on the sample, and the ¢ron: section of be munpie. 4H was detormunet by corn poring the murrunt wna, obtained by impressing the voltage drop acroas the standard resistance on a cathode-ray oscillogragh, with a ntandadl snr wave’ shape nf keen rues, vahwe. This eumparisen was based on making the cresi-to-trough valve of the standard wave the sane as that of the distorted exerting wave, aud Ubence computing AU rg 2VE Bem vi (Pals, where Bro, », 19 the root mean square value of the standard comparison wave, WV the tums on the exciling winding, R, the resistence of the stantard, and f the mean length of the iron riag sampla. § Thomas Spooner, ""Permeskility,” Jour. ALE, Vol. 42, p.425 January, 1923.22 MAGNETIC PROPERTIES OF 1RON (Guar. 1 Spooner states that this method gives results with fair accuracy for all classes of ferromagnetic materials if the amplitude of AB is considera- ble, but that for very small values of AB (reversible permeability) it is unreliable* One difficulty sometimes encountered in applying this method is finding the flux density B,, at the tip-of the minor loop if the polarizing magnetic intensity Hp is given instead of the polarizing flux density By. If anhysterctic magnetization curves, like those of Figs. 48a and 6, are available it may be calculated as By + (AB/2) if the even harmonic distortion of flux pulsation AB is small, If the maximum value of the alternating magnotic intensity H. is known, By, may be approximated from the normal magnetization curve corresponding to a magnetic intensity equal to Ho + Ho, provided that B,, so located is above the point of maximum permeability. For B,, below the point of maximum permeability Spooner states that the effective magnetiza- tion curve may be taken as a straight line between the origin and the point of the normal magnetization curve corresponding to maximum permeability. It will be noticed that another effect’ of superposing an alternating field on that produced by a direct current is to change the apparent permeability of the iron for direct current. Referring to Fig. 3a, the permeability for the direct magnetizing foree Hp alone is Bo/Ho, while with the alternating flux density B, superposed it becomes B;/Io. Spooner * states that. when AB is small the apparent permeability over the entire range of the magnetization curve up Lo fairly high densities is increased over the normal value, but when AB is large it is reduced, and at high densities is less than the normal permeability. If the ineremental permeability of the iron is known, the valuc of the polarizing flux density B, ean be approximately 7 determined for any value.of AB and Hy as follows: From AB and pa calculate AH. Then By, the maximum minor loop density, can be determined from the normal magnetization curve from Hy equal to Ho -+ (AH/2). Subtracting 4B/2 from By, the polarizing flux density B, can be found. ‘The value of By plotted against. Ho for constant values of AH or AH is sometimes called an anhysteretic curve (apparent magnetization curve with super- posed alternating flux). Such curves, obtained experimentally, are ** Data taken by Bensin (op. ed.) indicate that AB sbould be not less than about 20 kilomaxwells per square inch. T. Spooner, "“Esfect of a Superposed Alternating Field on Apparent Magnetic Permeability and Hysteresis Loss," Physical Review, 1925. 7 Owing to the even harmonic distortion prexent in either the alternating Aux density or ruagnetic intensity variation the axis of either cannot bé determined exactly by taking the mean ofthe maximum and minimum values,Aer. 241 SATURATION 23 given in Fig. 13a and & for samples 8 and 9 of medium- and high-silicon steel, respectively 23. Saturation As the flux density ofa piece of iron increases, the slope af the magne- tization curve equal to dH/dH7, sometimes called the dilfexeatial pecmea- bility, first inereases und then decreases. When the difterentaal permica- bility is only a few times greater than that of a vacuum Che iron is said to he saturated. The term saturation as ordinarily used is bread, ineinding, almost. the cnlire range of the magnetization curve above the sharp bend called ihe knee. However, there is a very debuite flux density, called tha saturation density, designated by B,, that the iron itself will carry. Any flax density in excess of this value is cor ered as heing carried by the void which would be left if the iron were removed. “Phus the Hux density in any piece of iron cun be considered as consisting of two parts: By, the so-celled fertic Mux density car- ried by the iran itself; and qlf, the fux de: sity that. would exist in Era, 4a. Separation of the flux of a piete of irom into the wbseuee of he iron its ferric component and sir nemsponaat. if the magnetic inten- - sity were maintained constant. ‘The actual fux density in the iron can then be expressed Ly the following equation: 63 1s hcg, find | | wo Om Toco Aenpesr Inns er tat 11 KoOOBp A pl "Chis is Uhtiated in Tig. do for a sample (8) of urediuwe-silicon steel: ‘The saturation density By can be extuapelated from data at lower densities hy means of the following well-known relation: # ay = ECB, -- By) * This relation, firsl discovered by Fréhlich, was later morified by Kennelly so az tabe in amore useful form. Kennelly, Trans. 4.1.0. Vol, p. 485, IROL; or see “Phecry and Calewlation of Flectria Cirouits!? First Edition, p. 44,24 MAGNETIC PROPERTIES OF IRON (Car. 1 which states that the permeability is proportional to the magnetiza- bility. If we introduce for ferrie permeability p, its equivalent B,/H and solve for By we will obtain: i B= E BB, Letting 1/B, = 9, and 1/kB, = , we have: H Brae ya gaat el where vy is the ferric reluctivity. “In other words, the law may be stated that the ferric reluctivity is a linear function of the magnetic intensity; if vy is plotted against H for any sample of steel the points should fall on astraight line, Actually, however, it is found that this law does nob hold until the knee of the saturation curve has been. passed. In Fig. 4b, the ferric reluctivity as computed from the normal rag. netization curve J of Fig. 1a, fora sample of anncated Swedish charcoal - = se Era onl Ser BO Toteaaterl we ech Tro fo ae ny Too 200 oo —700~—« 00 30040009, Amperesturne per neh Ht Fia. 4d, Ferrie reluctivity curves for extrapolating for the saturation density by Kennelly’s method. iron, is shown plotted against the magnetic intensity. It will be noticed that the straight part of the curve which can be extrapolated, begins at about 460 ampere-turns per inch; this point will vary with different materials, some such as cobalt steel requiring a much higher value of magnetic intensity before straightening out. ‘The slope of this curve is equal to 0.00717, corresponding to a value of B, equal to 139.4 kilomaxwells per square inch. - The intercept « isAgr 24} RESTOUAL FLUX DENSITY 25 considered to be a measure of the magnetic hardness of the matezial; that js, the greater the a, the harder the material. In Table TL Chere are given the saturation densities #, for various magnetic materials. Those for samples 1, 7,8, 9, U, Je, 126, and 20 have been extrapolated Trom their uormal rasgnetizalion curves by the above method. 24. Residual Elox Density and Coercive Intensity It is often desirable, when dealing with the foree produced. by the residual {ix of s magnet, or with permanent magnets, to know the value of the residual dux deusity B,, and the ecercive intensity #7, that will he produrad by various maximun-values of magnetic iuteasity. If the cuereive intensity and the residual flux density am plotted as fune- tions of the maximum imagnetizing intensity it will be found that these onrves have the sameshape as normal magnetization curves (sco Fig. £3). Sanford and Cheney? have found that these curves, beewuse of Uncie similarity to ordinary magnetization curves, may be extrapolated in an exactly sumiler manner. ‘fhus if H,/B, or H,/H, are plotted against H,,as the independent. variable, the resulting curves have the same shape as those of ry plotted against 77, discussed in Ant. 9. The following analytical expressions, which Umy have checked ta a high order of precision, result: Hr Bo + bl, 6) A, FE ot bath © a, and ap arc intercepts on the asis of ardinates, and by aud bz are the reciprosals of the saturation values of B, and H,, respectively. ‘These expressions, like (hal of Art. 23, cannot be usecl at very low values of Hyp, where the rctuctivily curve xy is not. a straight line. In Fig. fe are shown enzves of H,,/B, and H,,/H, plotted auainst If, lor sample 1 of annealed Swedish charcual iron. From these curves the values of by = 0.01275 and by ... 0.3775 are obtained, giving suturation values of By = 784 and H, = 2.65, respectively. In Wig. Sb are shown siinilar curves taken from data by Sanford and Cheney? fur 36 per cent cobslt Permanent magnet stecl. ‘These curves pive values of by = 0.01735 and by = 0.002285, corresponding to saturation values of B, = STG und B, = 436, wespectively. "Phe Variation ef Residual Inductioa and Coercive Foree with Maguetizing Force," Bur Standards Sei, Peper, 384, 1926.26 MAGNETIC PROPERTIES OF IRON [Curae 11 Figures 5c and Sd show H, and B, plotted against H, for both sam- ples: Note the exact resemblance to ordinary magnetization curves. In E8([ deaed swetsn nt Gee banal al ‘Graccoal iroh y Permanent Magnet Steet 4 i a M407 i Ha/He r Hm ABZ 120 6| + is eo} {|i i! ost | i 53 A Be hice wad a et 603) rT 13 30} ani e eal | TI | 7 +H ite Sp de he re Haat Zin, Hy act/in, Fie. 5a. Fra. 5b. Fras, 5a and $b. Hn/He and H,/B, plotted as a function of the maximum magnetizing intensity. xa = la He LT anhentey section |_| wf |Z, |e tnarebt om | he tobatl |/ Tecmaneot Mognef Stee) sol 0 fo Hel i Be 2a im sal 1» fool oo 0 a a a Haat /in, Hy at fia, Fig. 5¢. Fig. 5d. ros. 5e and 5d. Residual flux density and corecive intensity plotted as a function of the maximum magnetizing intensity. Figs. 18a and 120 curves are given for B, and He asa function of Hy, for various kinds of steel. ‘Table II also gives valnes of B, and H- for various steels,Q 8 Avr tA ENERGY CHANGES OCCURRING 26, Euergy Changes Occurring during Magnetic Cyctes In Fig. 6 is shown a normal hystereds loop for a saraple of commer cial mild cold-rolled steel, obtained after the iron was pnt in the eyelic state. Consider the energy changes as the loop is traversed. Starting at point 6and guing Luward 2 itis seen that the flux-is rising in the iron and, hence, Induclog a voltage act- ing against the current through the exciting coil. ‘This causes energy to be abstracted from the electric Girauit, the value of which is Br Wea f HB joules pec eubie ine and is evidently equal to the area 6-1-2-7-6 of the figure. Duriuy the next step of the cycle 4~3, energy is returned ta the electric circuit from the iron, as the indueod voltage due to the falling flux linkage is in the same dircetion as the falling cxeiting current. ‘This energy is given by tha integral: Fro. Energy changes aeeurring, dure ing a aormal hysteresis cycle. me Woa= fo HB joules per cubic inch +B nud is equal to the area 3 2-7-3 of the figure. The fast two steps of the eyele, namely, 3 to 5 and 5 to 6, are identical with the Gret two slepe beeause-of the syminetry of the loop about the origin. Hence the net. energy taken from the chreuit duriag the complete magnetic sycle will be equal lo 2/52 -- 21% 3, which evidently is the tatal area inside of the Joop. "Chis-enerey appears as beat in the iron, aad because it is abstracted from the electric circuit by the phenomenon of maguctie hysleresis it is walled the hysteresis energy loss. ‘his loss, in joules per cubic inch af ivon, is therefore equal te Ure ares inside of a continuous hysteresis loop, the coordinates of which ave webers per square inch aud sunpcre-turos per inch, The nermal hysteresis loop is of particular interest in apparatus where the finx goes through a cycle having equal positive and negative values,28 MAGNETIC PROPERTIES OF IRON (Corar, 7 as occurs in most alternating-current apparatus. ‘Then where the flux variations are periodic one can spenk of the hysteresis power loss per cubic inch, equal to the product of the area of the loop by the frequency of the supply in eycles per second. ‘Thisis evidently the average rate at which energy is being dissipated as heat, For such alternating-fux apparatus a magnetic material having small hysteresis loss is desirable from the point of view of increasing tbe efficiency and deercasing the size. For this purpose silicon is alloyed with steel producing the so-called “silicon steels,” which have a very low hysteresis loss. It is possible to evaluate the hysteresis loss only for a complete mag- netic eyele, that is, where the flux is brought back to its original value by purely electrical means. As an illustration of the lack of meaning of assigning definite energy forms to the various areas of the loop consider the following: Does the area 0-1-2-J-0 of Fig. 6 represent the energy stored in the iron due to the flux density B,,? If the iron had been originally demagnetized the flux density would have risen to By along the curve 0-2, and the energy abstracted from the electric circuit would have been equal to the area Q-2-7-0. ‘This area certainty does not equal area 0~1~2-7-0, but nevertheless the energy in the iron at a flux density of B,, is the same no matter whether one arrives along curve 0-2 or 1-2; hence, area 0-1-2-7-0 cannot be equal to the energy stored in the iron by B,,. However, neither does the area 0-2-7-0 represent the energy stored by By, because during the change 0-2 some hysteresis loss has occurred, and therefore the area 0-2-7-0 is greater than the available stored energy due to B,,; that is, some of the energy abstracted from the electric circuit, has already been dissipated as -heat in the iron, The fact of the matter is that the only way one could determine the energy stored by B,. would be to subtract, from the aren 0-2-7-0, the hysteresis loss occurring during 0-2, which would have to be measured by a calorimeter, Now consider the ares 8-4-0-3. Does this represent the energy stored in the iron by the residual flux density B,” Since this area repre- sents energy abstracted from the electric circuit when the flux density is decreased from B, to zero, it cannot be energy stored in the iron but must represent the energy input to the iron necessary to demagactize it. How- ever, at point 4 the iron is not completely demagnetized because it still has stored energy. Thus, if at 4 the exciting magnetic intensily H, is removed the flux density will rise to point 0’, returning to the electric cireuit the energy equal to the area 4-0-0’-4. Besides returning this energy the iron will still have some stored magnetic enerey due to the flux density 0’. ‘To actually demagnotize the iron it ig necessary to. apply 2 negative magnetic intensity greater than H_ which will decrease theArT. 265 NORMAL FYSTURESIS LOSS 29 flux density to 4‘, such that when Lbe magnetic intensity is removed the flux density will rise to point O along the curve 4'-(1* Tn other words, at point 3 the iron bas stored magnetic energy due to the residual density By whieh cuuld be evaluated by means of a calorimeter 05 previously mentioned, and in ordor ty remove this stored, cnergy, more energy equal to area 3 4’-0-3 mnst be put into the ion. Consequently, the stored energy at‘ due to 2, plus the area X~4'-0-3, must 30 into hysteresis Joss when the irou is cemagnetized. The area 3-4-4044, therefore, iy merely upproximately «qual to the eaergy input necessary 10 demagnetize it from the residual flux demity 8, In no event, though, can hysteresis loss be evaluated electrically until the magnetic eyele has been completed. 26. Normal Hysteresis Loss in Fig. 12 ara shown a series of normal hysteresis loops for various materials, ‘(he total symunctrical hysteresis Joes {complete Tuops) in joules per cubic inch per eyde is chown plotted as a function -of Ure maximum Isop density Br, on logdog paper in Fig. 14, by the curves labeled D. These curves have been obtained by measuring the aym- metrical loop areas of Fig. 12, In Fig. 176 ere shown simifar curves of normal hyzteresis loss for samples 8 and 9 of silicon steel. The data for these curves wore obtainad by exriting the sample with alteresting current and measuriag Uke dolal core loss with’a wattmeter, Correction was made for the eddy-current loss. onanrring, Is will be noticed Ehat these curves can, for a portion of their length, be represented by straight lines, the equstion af which will be: tog, Tn — log K + 2 log Bn where I¥s is the hysteresis energy loss in joules per cubic iach per cycle, log K is the intereapt an the axis of ordinates for B,, = 1, and nis the slope of the struight Siue. Or, tablag the antilog of hoth sides of the equation, we have @ ‘This is Steinmote’s equation for hysteresis loss: in general it ean only be applied aver a limited range of Sux density. In‘Pable I the values for Wy = KE, joules per cubic inch per 4 Tt zaust aot be thought, however, that the magnetic state 050 obtained will be adeatical with what would Weve heen obiined had the iron been demaguctized by successive reversals, Actually the slope of the curve 4’ Dat thearirin will be greater Uae Uhe slope of dhencrmal magnetization curve at this point. Lt is apparent, Lion, that, eveu thous the iron is demmenctized ae far ax sary of the ordinary tests would show, ithasnot bea retuinel to its virgin state; thabis,it retains 4 magnetic hislory.30 MAGNETIC PROPERTIES OF IRON {Cuar. IT K and n, evaluated from Figs. 14 and 17, are given for limited ranges of flux density. These constants can be evaluated only from experimental data for the particular iron in question, as they depend vot only on the composition of the iron, but also on the heat treatment, mechanical working, etc. For this reason the values given in the table should not be considered as necessarily representative. TABLE I K n | Range of Bu 0.010 X 10-3} 2.00 | 48 to 80 0.0068 “ | 1.63 | 48to 80 0.0083 | 1.62 | 32to 80 0017 | 1.50] 32 to 100 0.0037 “| 1.57] 20to 80 cong“ | 1.70] 25 to 75 27. Total Iron Loss Due to an Alternating Magnetic Field In the presence of a purely. alternating magnetic field there will be, besides the normal hysteresis loss discussed in the last article, a power loss due to the presence of circulating currents in the cross section of the iron flux path. These currents, known as eddy * | currents, are produced by the voltages induced in iad the perimeters of the cross sections of the iron path by the alternating magnetic field passing . through the cross sections. Thus, in Fig. 7a is shown the cross section of an iron lamination hav- = ing a thickness ¢ inches and a width large comn- id, pared to the thickness. ‘The magnetic fux passing Fie] Dorma to the cross section is represented hy the dots. The path for eddy currents is normal to Ae TE, EAGYCu™ the flux lines and parallol to the center line of thé lamination, ‘lamination as shown by the dashed arrows. The eddy-current density at any distance z from the center of the Inmination is equal to the voltage induced in one turn by the dux in area ated divided by the resistivity of the metal, or + «= Bm _ tifBut “pdt pAr. 27] YOTAL IRUN LOSS aT ‘The power toss rensity due to eddy currents will then be: and the loss due to eddy currents per unit volume of material is then _4 3p 2722 2 Saits per cubi @ where B,, is the maximum value (AB,,/2) of the cyclic loop density in webers per square inteh, é the thieknoss of the Laminations in inches, and a the res 5 It is thus seen that for any given Jamination the eddy will vary as the square oft the maximum cyclic flux deusity, the fcequen- ey, or the form factor of the induced wave of voll age producing the eddy eurents.. The constant 4/39 is usually deter mined hy aneasucing the eddy-mrrent loss experi- mentally, as il is found in actual praction that the 34] eddy-current loss depends not oaly upon the res yal aa tivity of Une sheet, hut alse Fregtenty -Cyntes ft See on the grain ay Vis. 7h. Eeparation of tota! eore Joss inte eddy erensing a5 the grain size _ Guten’ ard hysterods oomponcats, increases. in Fig, 160 are shown curves of total core Tas at two frequencies asa function of 2. for low-, medimem-, and high-silicon steel. Figure 1b gives the same data in terms of Iscquency for different values of Boy The eddy-current Jogg of a carpe of iran at any value of B,, can he easily delermincd from readings of the total core loss for two chiferant g | Cece Loe per Gycte ~ Feuies r 4 il Urges in \ ‘S60 “Properties en Testiag of Maguetic Mateials,” by Thormas Spooner, MeGraw-Hill Honk Co., First Edition, 1927, p. 25, alse Ctopier VIM,32 MAGNETIC PROPERTIES OF IRON {Cuar, IT frequencies by plotting the energy loss per cycle against the frequency. As the hysteresis loss per cycle does not depend on frequency, and the eddy-current loss per cycle varies directly as the frequency, the plot will be a straight line, the intereept of which on the axis of, ordinates will be the hysteresis loss per cycle. If this loss is subtracted from the total loss per cycle the difference will be the eddy-current loss per cycle. Figure 7 shows such a plot for sample 9a-of 29 gauge silicon steel, the data being taken from Fig. 16a or 16b. From Fig. 7b it can be calculated that of the total core loss of 0.65 watt per pound for sample 9a at By, = 64.5 lanax. per sq. in. and a frequency of 60 cycles per second; 0.1464 watt, equal to 0.00244 x 60, is the eddy-current loss; and 0.504 watt, equal to 0.00840 X 60, is hysteresis loss. ‘The data of Figs. 16a and 166, while given for 29 gauge sheet only, aay be extended to cover the iron loss for other thicknesses by correcting, the eddy-current component of the loss for the change in thickness. 28. Unsymmetrical Hysteresis Loss in Direct-Current Electromagnets, ete. In 2 direct-current electromagnet the normal magnetic cycle of the iron is quite complicated. For all practical purposes, however, this eycle is approximated quite closely by a loop having the normal mag- netization curve for the rising branch and the’ demagnetization curve of a normal hysteresis loop for the descending branch. Such a loop is shown in Fig. 6 by the rising line 4-0-2 and the falling line 2-3-4’. Each time an electromagnet gocs through a complete cycle of operation, energy equal to the area of this loop will be dissipated as heat, per unit volume of iron. Consider the simple ring magnet ‘illustrated in Fig. 8a. Let the air gap at the beginning of the working stroke be g; and at the end of the working stroke be g2; assume that the flux density throughout the iron is constant. When the coil circuit is closed (air gap held constant at g:) the fux in the iron will build up along the curve 0-1, Fig. 8¢, to ¢1, causing the flux density in the iron to reach the value By of Fig. 8b. If the air gap is now allowed to decrease from.g, to gz, the flux in the iron will increase from $1 to 2, causing the flux density to rise to Ba. Magnetically the same result would-have been obtained by decreasing the air gap first and then applying the magnetomotive force, in which case the curve 0-2 of Fig. 8c would have resulted. This curveis obtained by adding together the rising magnetization curve of the entire piece of iron, computed from eurve 0-1-2 of Fig. 86, and the magnetization curve for the air gap shown by line 0-7 of Fig. 8c. Likewise the demag-Ant. 28] UNSYMMETRICAL ILYSTERESIS LOSS 33 netizalion eurve 2-3—-4-5 for the entire magnet. (iron and air gap) can be determined in exactly the same manner. When the coil ci i broken the flux will deerease along this demagnetization curve to point 4, corresponding to the flux density By. The cycle of the maguck is completed by increasing the air gap from gz tom. ‘This will require mechanical energy. If the permeance of the air gop at go is large compared to that at g, the Bux ean, for all practical purposes, bt cou- lered to fall to sero along curve 4-0, Fig. 8c, Ltis now possible to evaluate tho effects of hysteresis from a practical point of view. First, if the cycle of the magnet were ideal there would be na losses of energy of any kind and the total area of the rectangle 0-1-2-8-0 would be available 2s mechanical work.'* Of this total area certain portions ure unavailable; area 4-3-2-7-8-4 is returned lo the 7 jy Hiteesn pte Mysteresis Oya (@) (a ecites aly fo tate tteznet Fro. 8 Unaymmetnical hyateresis cycle of direct-curreat electromagnets. electris cirenit. in the form of a spark when the coil cironit is interrupted, and area 0-2 3 4-6 0 is Jost owing to hysteresis, Area 3-2 sents the energy returned Lo the-olectric eirenit hy the fron end corre sponds to area 32-7 3 of Fig. 86. Area 0-28-60 represonts that portien of the hysteresis loss which is cupplied directly by the elect circuit and corresponds to sre O-1-2-3-6-0 of Big. 8 Area 0-6-3-7-8-0 represents the exergy stored in the air gap (length 92) Just before the cod cireuit is opened. Aflor the coil circuit, has been openedt the energy of the air gap is represented by atea O-€~4-0.° ‘The difference hetween these two energies is divided into kwo portions: part 437.8 4, which ix returned to the electric cirenit; and part 4-G-3-4, which is consumed in demogretizing the icon from Uhe residual fux density By (at zero airgap length) to the actual residual flux density B4, and corresponds to aren 4 6 3-4 of Fig. 8b. When the armature is maved mechanically from ge to g1 the flux will fall from 3? [his is explained in detail in Chapter LIZ.34 MAGNETIC PROPERTIES OF IRON {Cnar, I 4 to 0 if the permeance in position g, can be considered small compared to that in position go. ‘The energy changes involved during this latter process can be evaluated if the change from 4 to 0 is made clectrically. ‘This ean be done by increasing the coil current in the negative direction (air Bap = g2) until point 5 is reached and opening the coil circuit. This will reduce the Hux to zero, and the mechanical motion from go to gr can. now be carried out without expending any energy. In this cycle of operations the energy left in the air gap (area 0-6-4) will be dissipated, and, in addition, energy represented by area 0-5-4-0 will be taken from the electric circuit. These two energies evidently correspond to area 0-6-4-5-0 of Fig. 8) and constitute the energy necessary to demagnetize the iron from By to zero. Summing up: In a complete cycle of the magnet the hysteresis loss will be equal to area 0-2-3~4-5-0 of Fig. 8c, or the loss per cubic inch of iron will be the area of the loop of Fig. 8b. Of this total loss, part 0-6-3-4-0 is supplied by the stored energy of the air gap, and part 504-5 by mechanical means. Of the total work available from the ideal magnetic cycle, namely area $,.f',,, the portion 0-2-3-4-0 may be considered as lost as the result of hysteresis and portion 4-3~2-7-8-4 lost by its return to the electric circuit. To make possible the predetermination of hysteresis cycles for elec- tromagnets, the loops of Fig. 12 have been plotted from actual data taken on the sampled indicated. These loops may be used directly to construct the complete hysteresis cycle as shown in Fig. 8c. Loops for values of B,, other than those given on Fig. 12 may be approximated by finding the values H, B,, and H, for the particular desired value of Bm from the data of Figs. 11 and 13 and then drawing a curve through these points of the sae shape as the adjacent loops of Fig. 12. To facilitate evaluating the energy losses corresponding to the various loop areas the data given in Fig. 14 have been determined from the loops of Fig. 12. Curve A gives the energy loss of iron corre- sponding to the complete magnetic cycle, area 0-2-3-5-0 of Figs. 8b or 8c. This curve can be extrapolated to higher values of Bm if dosired. Curve C gives the energy required to demagnetize the iron from the residual density B,, area 0-3-5-0 of Figs. 8b or 8c, This energy approaches a definite maximum because the demagnetization curve quickly approaches a definite limit as saturation is reached. ‘This is shown by the relatively small area inclosed between the dashed magnetization curves (labeled B,, = B,) and those corresponding to Bm = 100 of Fig. 12. Curve B gives the energy returned to the electric circuit by the iron and corresponds to area 3-2-7-3 of Figs. 8b or 8c.Aven, 25 UNSYMMETRICAL HYSTERESIS LOOPS 35 This curve comet be safely extrapolated. Values for B,, greater than 100 ear: he approvimated by adding to ihe value ab My - 100, a value approximated fram the area hack of Uie uorinal magnetization curve for the increment over B,, = 100. Partial areas like 16-3-4 07 0-63-40 can be approxinated by taking a part of the value as given by Curve C, which part can be estsmated rom the residual density B, end the shape of the dernagnetizstion curve as given ou Lic loops of Fig. 12. Area 0-2-3 0 is obtained by subtracting the value given on Curve C fcom thot given by Curve 4. 29, Unsymmetrical Hysteresis Loops in Alternating-Current Apparatus Unaynunctrical hysteresis cyeles are also of interest in apparatus which earries an alternating flux superposed npan a constant faux. This occurs in tranafarmers or chokes earrying direel current, susk a3 cudio- frequency chokes, sudie frequeney Lransformers, and transformers sup- plying halt-wave rectifiers. in these cases the hysteresis lnss for a given alternaling fux density depends upon the displacement produced by the poladaing magnetic intensity, and may sometimee be as great as several times the normal hysteresis Inss produced when the polarizing megnelic intensity is zero. The ratio of the loss for the displaced loop to that of the normal Loop is called Uke displacement factor. A series of such leops the sume B amplitude but different displaccsaents are shown m big. Yo, while in Fig, 95" a series of such loops having the same B displacernent for a varying B amplitude are shown, In Fig. le the displacement factors for vacious values of polarizing magnetic intensities Hy a2 showa plotted es a fuuetion of the maxirum cyclic flux density (AU)/2 (sce Vig. 3) of the displaced hysteresis loops jor samples 8 and 9 of silicon steel. In Fig. 172 the normal hysteresis loss jor the wo samoles is shown plotted. The data of Fig. 17 were oblaiaed by moliing ailernuling current moasurements on the iron samples im the following manner: ‘he iron sanspies, in the form of riws with the lamina- tions well insulated with varnish, were provided with three insulated windings Que winding was vouucted to a GO-vyce alternating cunent source ta provide an alternating magnatie flux. A wattmeter wes used to sngasure the power input ty this winding. Another winding was cus nected to a higa-resistance divect-current voltmeter through a rectifying commutator in order to measure Une true value of the flux variation AB produced by Uhe alternating current. The third winding was connected havi 9 Figures a and #8 are (aken Irom data from ‘Properties and ‘Testing of Mag- netic Materials,” by Thomas Spooner, MeGaw: Mill Beak Co,36 MAGNETIC PROPERTIES OF TRON [Cuar: 11 in series with a choke coil to a direct-current supply in order to produce the polarizing magnetic intensity Ho. ‘The wattmeter reading was corrected for all extraneous power loss, that is, copper loss in the exciting winding, loss in the direct-current voltmeter, alternating-current power loss in the direct-current magnetizing circuit, and eddy-current loss in the iron. The eddy-current loss was obtained by measuring the core loss at two different frequencies with zero polarizing magnetic flux density and extrapolating in the usual manner. Flux variations were maintained sinusoidal by keeping the resistance in the exciting winding low and using a variable-voltage source of good wave form. = bly scala by 6.45) . 6 is 3 2 A. pet in, (multiply ale by 2.02) 12 3 4 5 6 A per in. (multioly scate by 2.02) Fia. Qa. Unsymmetrical hysteresis Fra. 95. Unsymmetrical hysteresis loops baving the same B ainplitude _ loops having different B amplitudes but different displacements. but the same displacement. Reprinted from ‘‘Properties and Testing of Magnetic Materials” by Thomas Spooner, by courtesy of the McGraw-Hill Book Co. The hysteresis loss for any value of Hy and AB/2 is obtained by multiplying the displacement factor obtained from Fig. 17a by the cor- responding normal hysteresis loss from Fig. 176. It will be noticed that the displacement factor increases as Ho increases for any constant value of AB/2, and decreases as 4B/2 increases for any constant value of Ho. In other words, the displacement factor increases as the center of the displaced loop is further removed from the origin. ‘This, however, is not true over the entire range of fux densities; it has been found by Edgar * that when the alternating flux density 4B/2 exceeds a certain critical value the displacement factor becomes negative, and that below this density the hysteresis loss for constant values of AB/2 first rises as Ho or By increases as shown in Fig, 17, then approaches a maximum, and finally, at high values of minor Joop tip densities, (B, -+ 4B/2),Aus. 20} UNSYMMETRICAL HYSTERESIS LOOPS 37 decreases. ‘The data of Fig. 17 do not show the latter iwa changes for the reason that they have not been carried to high enough values of (Bp + 8/2) ‘The data of Fig. 17 do not give any information regarding the value ol B,, the polarizieg dux density, due te any value of Hg. For any given valuz of Ho the polarizing Bux density 4, will vary, depending upou (aB/2), the maximum vale of the superposed magnetic flux density, In Figs. 18a and b, values of By plolted as 4 function of Hg for constant yalues of 43/2 are shown. ‘These data were obtained while taking these of Fig, 17 hy reversing Hg with the alternating dux density AB/2 super- posed and reading 4B, uu » ballistic galvanureler. By using these curves in conjunction with Kig. 17 the displacement factor im terms of dF} /2and By may be obtained" A, may alsa be obtained by the method outlined at the end of Art, 22, Ju Figs. We and 18d are given curves of alternating flux density AG/2 plotted as a function of the rms. alternating magnetic intensity H for various vonstaat values of Hy, These data were obtained while taking these of Big. 17. It ic welul for eslaudating thy required alecnat- ing-enrrent excitation fur apparatus having « sunerpased direct, excita liun, aud can Le used where (he alternating flax wave forin is sinusoidal. Where specific data like those of Figs. 18c and 18d are not available the alternating-current excitation in the presence of direct excitation aay be calculated from the data of Fig. 19 as follows: For any given value of AB/2 and Hg the incremental permeability 4, way be found from Vig. 19. The erest-to-crest value of AZZ will then be given by the equation AH = =. 8) If the minor loop ia such that the wave of alternating magnetic intensity cana be assumed fairly sinusoidal the rms. exciting magnetic intensity will be AH AB (Drma = 5 iin ampere-Luras per inch (18) A good many dota on this oubject are given in this my. Sco Chapler VE, “Dropation ond Tecting, of Moguelic Materials," Thomas Spooner, MoGrav-Hill Boole Co,5 “Magnetic Properties of Shrct Steel wader Superpased Alteraating Field and Unsymumebical Hysteresis lasses,” Yasujiro Niwa and Yosbihire Asami, Dept. of Communications, Tokyo, Japan, Researches of the Rlectrotechnical Laboruery, 124, June, 1923, “Tana Characteristics of Silicon Steel at 60 Cyeles with D-C Exeitation, RF. Edger, Trans, ALE-E., Vol. 52, p. 721, September, 1933.38 MAGNETIC PROPERTIES OF IRON [Crav. 11 where AB is the total alternating flux pulsation in kilomaxwells per square inch and ys is the absolute incremental permeability in kilo- maovells per ampere-turn in an inch cube, equal to the values given in Fig. 19 multiplied by 0.00319. 30. Effect of Grain Direction and Machining Strains on the Magnetic Properties of Steels ‘The permeability and iron loss vary considerably with the grain direction. Spooner states that, for electrical sheet, a flux direction perpendicular to the grain direction will cause the permeability through- out its entire range to be about 75 per cent of that obtained with the flux parallel to the grain, while the hysteresis loss is inereased about 14 per cent. Likewise the coercive intensity is about 25 per cent greater, It is generally desirable to use the iron parallel to the grain direction, especially when it is desired to build a sensitive relay where high per- meability and low coercive intensity are advantageous. In tractive magnets which usually operate at high flux densities it is not so impor- tant. . Recently silicon steels of exceptionally high permeability and low hysteresis loss at both low and high flux densities have been produced by a special method of cold rolling and annealing. When the method is properly carried out a fine-grained material having magnetic and elec- trical properties approaching those of a single crystal-are obtained. Losses as low as 0.46 watt per pound at 60 cycles per second and By, = 64.5 kmax. per sq. in., and maximum permeabilities as high as 22,000 measured in the direction of rolling, have been reported. Machining a piece of iron is decidedly detrimental to its magnetic propertics. It seems that the metal for a considerable depth behind the machined surface is strained, causing it to have a low permeability and a high coercive intensity. Cold working such as hammering or rolling has asimilar effect. These undesirable effects can be completely removed by annealing the iron. This is accomplished by heating the iron to a maxi- mum temperature of about 760°C. and then allowing it to’ cool very slswly. The iron should not be allowed to come in contact with oxygen during this process as the resulting oxidation is detrimental to the magnetic properties. The maximum annealing temperature varies slightly with different kinds of steel. 18 “New Development in Electrical Strip Steels Characterized by Fine Grain Structure Approaching the Properties of a Single Crystal," by Norman P. Goss, Trans. Am. Soc. Metals, Vol. 23, p. 511, 1935.Axc.3il MAGNETIC MATERIALS BOR CLECTRUMAGNETS 39 ‘The inpurlaoee of machining strains depends upon the size of the piece of iron. If the dimensions arc amall, then a considerable pereeut- age of the total volume will be affected and annealing is very necessary, whoress With a large piece it is relatively nnimportant. ‘he effect of machining strains can be seen by comparing curves 1 and 2 of Fig. Lia, both of which are for Swedish cburceal ion. Ring sample 1 wos rmachiaed fron a golid bar end has relatively small dimensions but was thoroughly anvesled. Ting sample 2 was machiaed out of ginch dead soft plate ay recived, and was given no heat treatment after umehining. Its dimensions are considerably larger than those of sample 3. For low flux densities the magnetization curve of sample? is much Lower than dhat of sample 1; wt the higher densities the difference iz not so inarked. Likewise by referring to Wigs, 12 and 13 it will be seen that the coercive intensity of sample 2 is much greater thaa that of sample 1. 31. Magnetic Materials for Electromagnets American Ingot Iron. ‘This is the purest form of iron eoramereially rofined in opea-hearih furnaces. The total impurities do not exceed 0.16 per cent, the earbon content being only of the order of 0.01 per eeab. It has high electrical conductivity, high permeability, aud low coercive intensity. It ean be obtained in the form of bars and plates (hot-rolled), colrt-rolled strip, and wire. Mapnetie data for this irou ae given in Figs. Via, 12, 13, 14, and ‘able 14. ‘lo bring out the best magnetic properties it, anust. be very carefully annealed after machining. A muxdiwus teo- perature of 1400? F. cr 760° C. followed by slow evoling is recommended. Generally speaking, for direet-current eleetromagnets this material is the best obtaitable. Unforbunately this iron, except for sheat and strip stock, can be purchased valy ia large quantities, usually rolled special to order, In sheet form this iron can be ohisined bydrozen-annealed, 2 forma having much higher permeability. Cold-Rolled Steel. From « practical point of view, when building. tractive magnets and other devices that operate at high flux densities, good-quality mild cokl-relled steal will be found satisfactory, Iu sisip stock, that is, strips having 9 thickness between 0.010 and 4-inch maximum, the neurl steel is a low-carbon steel with n bright. finish, designated as 8.A.B. 10-10 having between 0.05 and 0.15 per ecnt carbon. Lt is fnchiched in four degrees of hardness: dead soft, + hard, } hard, and full hard. The dead suft steel will take a 180° bend in either gram direction, the 3 Lard will take a 180° bend scruss the grain only, Ure ¥ hard will take a 80° bend across the grein only, aud the full have ‘cannot be bent in either grain direction without cracking. 1a bar stock, that i,40 MAGNETIC PROPERTIES OF IRON [Cnar. 11 bars and rods over } inch thick, the usual steel is one having slightly more carbon, between 0.15 and 0.25 per cent, designated as $.A.E. 10-20. . ‘This steel is not aanealed or heat-treated" after the final finish. . Its hardness corresponds to about that of the 4 hard strip. Data for both of these materials are shown in Figs. 11a, 12, 13, 14, and Table IT. It will be noticed that the $.A.E. 10-20 steel has a very low permeability at the low flux. densities, without annealing; it can be used without annealing for tractive magnets working at high flux densities. ‘The advantage of using this material is the case with whieh it can be obtained in a variety of sizes and shapes, the ease with which it can be machined, and its fine finish. For the most efficient results, the Snal machined magnet should be annealed. Where a free machining steel ia required, as, for instance, in parts to be made by ascrow machine, S.A.E. 1112 may he used. Magnetically this ateel is only slightly inferior to S.A. 10-10, Swedish Charcoal Iron. The best grade of Swedish charcoal iron is practically identical in its magnetic properties to pure American ingot iron. For that reason only one set of curves and data is given for these two irons in Figs. a, 32, 13, 14, and ‘able I. These curves have been obtained from a sample of Swedish charcoal iron and check very well for other published curves for American ingot iron, ‘These irons cannot generally be purchased from open stocks but must be specially ordered, Cast Steel. Whore the shell or some other part of a large magnet is + of intricate shape, it probably is of advantage to make it of cast steel. Magnetically this material is very superior to cast iron, having a satura~ tion density of 135.5 lanax. per sq. in., which is only slightly less than that of piure iron (see ‘Table II). Owing to the variation in composition and heat treatment, cast stecls differ greatly in their magnetic charac teristics. A representative magnetization curve for this material is shown in Fig. Ua. Cast Iron. Cast iron as a magnetic material is rather inferior. Its use can be justified only beeause of cheapness and ease of casting and machining. Its saturation density is about 90 kmax. per sq.in. This material like cast steel varies greatly, depending upon its composition. A representative magnetization curve is shown in Fig. 11a. Malleable Cast Iron, Recently there has been developed a highly magnetic form of cast iron which is sold under various trade names.! - The Newark Malleable iron Works manufactures this iron under the trade name of Magtiz: Other manufacturers are: Eastern Mallesble Tron Co., Delaware; National Malleable and Steel Castings Co., Cleveland. Yor additional magoetic data see “Symposium on Malleable Iron Castings,” June 26, 1931, A.S.T.M, and American Foundrymen’s Association,Aer SW MAGNETIG MATERIATS FOR ELECTROMAGHELS. AL "This iron is made from a white-iron bese, Lhe oxact composition and proces of manufacture being a seeret. It ean be east in intrieste shapes and requires gercful anwealing alter casting to develop its best mugactic properties. ‘Tho mechanical properties aud machinability of this iron are about the same as for matleable east iron. Its cost. is about twice that of ordinary pray-iron castings. Magnetically, up to densities of 80 kmax. persg. in, it issuperior to ordinary } hard machine steel, having a higher permeability and a lower coercive intensity; and up to 60 kmax, per eq. in. it is comparable & cast "This cast icon should be very useful far the yokes und pole cores of muguels where an it tricate shape is desirable or economical, !or detailed data oman annealed casting of this jaon coo Migs. La, 13a, 136, and Table Electrical Shect Steel. This materi ustd in the construction of alll clevtries! machinery in which the due is rapidly changing or alter nating. Tt is made of high-quality cpen-hearth steel wilh varying percentages of silicon. Munufaeturers usunlly grada there sheets on the basis of their core less (combined hysteresis and eddy-aurrent leases}, there being usually about six grades. Those grades having the lowest core loss have the highest silicon content. The silicon content ranges from about 42 per cent in the bighost-prade transformer steel to about U.5 per cent in so-called “armature " grade which is used for the arma- tures of sapull dirwcl curreul anachines. The tecm © silicon steel is usually opplied to only those steels having in excess of about 1 per ovat ‘licen. Our interest in these steels is merely confined to their use in ullerwating-current clectromagnots, choke coils, and transformers “carrying direct current. For choke coils and allernatiug-current electzumagnets a medium silicon (approximately for transformers carrying direst enrrent a B desirable. These sheets are manufactured in thicknesses ranging from 29 to 22 U.N. gauge, except Lhe high-silicon sheet which is not made in the heavier gauges, but is offen made in lighter gauges, 32, 36, aud 43 fer upecial radio applications, Data fer low, medium, and high silicon steels are given in Figs. 15, 16, 17, 18, 18, and Table IL. Electrical Bar Steol, Bar and strip stock of te and composition as electrical sheet steel is available in several per- centages of silicon. Tis law enereive intensity makes it desirable in a relay requiring small residual force, aad ils high resistivity is useful where it is ueceucury ty mitigate the effet of eddy currents as in high- speed diect-currenb magnets cr in aller oabingcurrent: magnets. In the latter application it is practical to make the plunger: of smell alternating-current magnets of solid silicon steel, while those of theAZ MAGNETIC PROPERTIES OF IRON (Cuar, IF larger maguets can also be made of the solid bar stock provided that radial slots are milled in to break up the eddy-current paths. Iron-Nickel Alloys. There are two rather important iron-nicket alloys, one having approximately 50 per cent nickel, known as “ Nicaloi” (General Electric Co.) or “Hipernik” (Westinghouse Electric and Manufacturing Co.), and the other having 78.5 per cent nickel known as “Permalloy ” (Western Electric Co.);-all these nickel-iron alloys require a very careful annealing process after machine working to develop their best magnetic properties. Permalloy. This is a very remarkable alloy distinguished particu- larly for its high initial and maximum permeabilities. ‘The initial per- meability is about 9,000, and the maximum permeability is as high as 100,000 at 32.2 kmax. per sq. in. Another property which makes it useful is its extremely low coercive intensity and hysteresis loss. It should therefore be particularly usefu! for sensitive relays which are to have very low residual forces. At present it is used extensively in the telephone industry to load cables. Its disadvantages are that it requires a very careful heat treatment and is very susceptible to mechanical strains; likewise it is-difficult to obtain. Data for this material are given in Fig. 110. Fifty Per Cent Nickel-Iron. This alloy, though having lower initial and maximum permeabilities than those of Permalloy, about 5,000 and 32,000, respectively, is in many ways more of a practical commercial material for general use. It does not require such an exacting heat treatment, nor is it so affected by mechanical strains; also its saturation density is higher, being about 100 kmax. per sq. Likewise the material can be more readily purchased. It is available in all the usual rolled forms, including sheets, plates, bars, rods, and strips. It, like Permalloy, has a very low coercive intensity and small hysteresis loss. ‘The main use of this metal at present is for the cores of high-quality audio-frequency transformers and chokes, where its high incremental permeability is essential. It is also useful for the cores of particularly sensitive relays where low coercive intensity and high permeability are essential. Data for an alloy of this type (47 per cent Ni) manufactured by the Allegheny Stecl Co., under the name of “Allegheny Electrie Metal,” are given in Figs. 116, 12g, and Table 1. Tron-Cobalt Alloys. These alloys are unusual because their normal permeability remains high up to high values of flux density and the incremental permeability is much higher than that of other materials in the presence of strong polarizing fiux densities, 75 kmax. per sq. in. and up. The saturation density of these alloys is about 12 per cent higher than that of iron. This makes them particularly useful forAge Sth MAGNETIG MATERIALS POR ELECTROMAGNETS as clectromagnets where the apace is definitely limited or where high force densities or Mux densities are destred. The hivk incremental permea~ bility al high polwizations is nsciul im polarized devices hike telophoue where it enhances tite sensitivity. There are twa iron-cobalt alloys thai are used: an alloy containing 84.5 per cent cobalt corresponding 1a the compound He.Co, and another copluining 50 por cent cobalt, This material cannot readily be cast direetiy inte its final shape as the molten metal is very viscous and has 2 tendency tu fora blow holes. It is generally cast in the form ol a billet, then hammmer-forged into a bar, and Ube rolled if necessary. Me- chanical working is quite essential to develop structure. “The process of heat treatment is very iinportant if the alloy is to develop its best mag- netic propertirs. At the present time, owing to the cost of cobalt and. alse to the difficulty of uifacture, ferrocobalé is relatively expensive. Dato for twn commercial samples (12a snd 125) enntaining 34.5 per cent cobalt are given in Figs. 11b and $24 and ‘Pable IT. Dats for a sample of Parmendur containing 50 per eent eobalt and 20 ger cont vanadium are given in Fig. 11 and Table TT, Chapter I, and Figs. J sud 2, Chapter £ Tron-Nickel-Cobalt Alloys. There has been developed recently a group of iron-nickel-cobalt alloys, known as “Parminvar,” © having with certain heat treatments a very constant permeabilily and unusually staali hysteresis losses at low flux densilies. Data for this material are given in Table TL. Tou-Nickel-Chromium-Siticon Alloys. These alloys sre char- acterized by having, their rritical teraperatures 8 depresaed into the low temperature range. Thay are usefil in the eanstenetion of temperature controlled apparatussuch as therma! relays or evalaclers, reactors whee reactance varies with temperataze, and special devices tike transformers whose output can be made to vary with temperature. Sn alloy of 45 per cent Mi, 5 per cont Cr, and 50 per cent iron bss a Curie point of 925° Co cae of 45 per coat Ni, 16 per cent Cr, and 40 per cent iron has a Curie point of 59° CE receiver 1’ Mapnetic Properties of Pecminvars’ G. W. Kimen, Bell Sys. Yeck. J, Jann ary, i979. \The critical temperatitre or Curie poiat is the temperature at which » ferro- megnetic miatesial heccitnes now-ravnetie. “Tha critical temperatures for nickel, zen, and opbalt, are 36%, 780, and IL20°C., respectively. ‘Tae parrmenhitity of these materials resins substantially easstanl with terpersture until about 50° C. below the Curie point and then rapidly desrensea to unity st tha Curia point 19 Por adiscuscioa of these matecials sce “Temperature Seuitive Magnétic ANeys and Their Uses," by L. R. Jackson and H.W. Russell, Zastremeads, Vol. Li, Novem- ‘ber, 1928.44 MAGNETIC PROPERTIES OF IRON [Cray TD Non-Magnetic Steels. In many instances it is desirable to use a steel which is non-magnetic.” Stainless steel, S.AE. 18-8, fulfils this qualification and may be used in place of the non-ferrous alloys. 32. Magnetic Materials for Permanent Magnets ‘The prime requisites for a permanent magnet steel are high coercive intensity, high residual flux density, and magnetic permanency. Their relative importance depends upon the application, in some cases it being economical to use a magnetically inferior material such as cast iron while in others the use of expensive cobalt steel, or Alnico is justified. ‘A permanent magnet is useful only because it can produce magnetic flux in an air gap outside of the magnet. The usefulness of a magnet is measured by the quantity of flux it can produce in the gap and the mag- netomotive force it can maintain across the gap. +8) Onc-balf the product of these two quantities is the by energy stored in the gap. ‘The maximum possible energy of the gap per cubic inch of iron is, there- By fore, a logical way -of evaluating the magnetic efliciency of 2 permanent magnet steel. Figure 10 shows the portion of a hysteresis loop between the residual flux density B, and the coer- . cive intensity H.. ‘This section of the loop is called 7 . the demagnetization curve and is iuseful in the dis- 1a. 10. Air-gap . . energy availablefrom cussion of permanent magnets. The residual flux apermanent magnet. density By can exist only in a closed iron sample such as a ring, the total coercive intensity H being required to overcome the reluctivity of the iron. If an air gap is intro- duocd into the magnetic circuit part of the available magnetomotive force is required to send the flux across the gap, thereby reducing the magnctomotive force available to overcome the iron reluctance. ‘Thus in Fig. 10 the introduction of a gap will reduce the flux density from B, to Bu, thereby reducing the reluctivity drop in the iron from H, to He — He and making available in the air gap a magnetomotive force equal to Hg X length of iron. The shaded rectangle having an area equal to Ballz will therefore be equal to twiee the energy of the gap per unit volume of iron. Obviously, then, the most efficient point of operation of the magnet steel will be where the area Bglfz is a maximum. For this reason the criterion for the comparison of various magnet steels has become the largeness of the product BaHfz. ‘The method of design- ing © permanent magnet to operate at this point will be discussed in Chapter TV. a erArr. 32) MATERIALS YOR PERMANENT MAGNETS, 45 Cast Iron. This material ean Le used for pecusunent magnets when properly heat-treated. Its advantages are its cheapness and ease of reachining, 14 has a coercive intensity of about 100 ampere-turns pet inch, which is almost as large as that of carbou steel, but it has a much Juwer residual flux density. Carbon Steel. Carbou sicel for permanent maguels coulains about 07 to 1 per sant carbon. ‘This steel when properly heat-trested will have a coprcive intensity hetween 100 and 120 ampere-turas per inch and a residual flax density between about 60 and 50 lonax. por sq. in, Commercially this steel has been almast entirely replaced by alloy stecls having belter properties, particularly less aging, See Fig. 20 and Table ILI for data. Chrome Steel. ‘This steel usually contains from 2 to 3 per cent chrominra with about 1 per ceut carbon, It is an oil-hardening steel with « fairly simple heat treatment, Data for a2 per cent chrome steet are given in Fig, 20 and also in Table HT. This steel ts aboub an stahle as tungsten steel but is less expensive und lence has replaced tungsten in many applications. Jt is reporied hy Gumlich® that an alloy having 6.24.per cent chromium and-1.14 per cent carbon is decidedly supnvior. See ‘Table IU for data. ‘This stecl is not generally available commercially. ~ Torigsten Steel. “This vtec] tontains about §4 per cont of tungsten, from 0.6 to 0.8 per cent of carbon, and sumatimesfrom 2 to 1 per cent of chromfum. It can be made either water or uil barduuing, and requires some care in heat treatment wing to the possibility of distorting and erucking it. Wherens the available energy of tungsten stéel in only n Kittle greater than that of cuzbou steel its chief advantage is that it is mure stuble; that is, it is not to subject Lo less of magnotie energy dua to mechanivrl shock or heating. Bee My. 20 end Table TH for data on this steel. Cobaltchrome Steels. ‘This name is given to a series of low- and medium-cobslt steels ranging from ahont # to 20 per cent cobalt. They also contain about 9 per cont chromimn and from 0.8 to 1 per cent carbon, They are air-hardening stecls auth contain a small amount of tungsten und weolybdenum to assist the air-bardeting properties. Their heat Lrewlnent is quite complex. Figure 20 gives data for a 15 per cent eobsltehrome steel. The maximum availuble energy of this steel ts not quite so great as that of cubalt'steal, but it is considerably less expensive. Cobalt Steel. “Uhis steel contains about 35 per ecul cobalt and cor- responds ta the alloy FesCo previously descrived as having remarkable bromiumCarbon Steela for Poonanent Magnets,” E, Gumtich, Bick u. Mash, Vol. 39, p. 50M, Nov. 20; and p. 886, Nov. 27, 1921,46 MAGNETIC PROPERTIES OF IRON [Cran, IT magnetic properties. Besides cobalt it contains tungsten, about 4 per cent; chromium, 2 per cent; and 0.8 per cent carbon. It is an oil- hardening steel and gives little trouble from distortion or cracking. It can either be cast into its final shape or rolled into bar stock. Data for this steel are given in Figs. 5b, 5d, and 20 and Table HL. ‘The cobalt, stcels are expensive because of the high cost of cobalt. and their use generally can only be justified where it is essential to decrease. the size or weight of the magnet. Magnetically, they are very stable. Dispersion-Hardened Alloys." ‘This is a relatively new group of permanent-magnet, steels developed in Japan and the United States. There are two widely used types, the aluminum-nickel-cobalt-iron alloy developed by the General Electric Co. and sold under the name of Alnico; and the aluminum-nickel-iron alloy sold under the names Alnic (General Fleetric Co.) and Nipermag (Cinaudagraph Corp). Alnico. ‘This alloy is made in two varieties, 5 and 12 per cent cobalt, and requires very careful heat treatment. The material is fabricated either by casting or sintering. When sintered it can be molded to its final dimensions, and when cast it must be finished by grinding. Any necessary holes must be cored into the casting; soft’steel inserts may be cast in for fastening or for any other purpose. It has a comparatively high coefficient of thermal expansion, and due care must be taken’ in designing the magnet and the mold to make adequate allowances for shrinkage in the casting. It is relatively weak and brittle as compared with other magnet alloys. . This alloy is remarkable because of its high coércive intensity, which is about 870 ampere-turns per inch forthe 5 per cent cobalt variety. ‘This is almost twice that of 36 per cent cobalt steel. Its residual flux density is only about 80 per cent that. of cobalt steel. Because of the higher maximum available energy of Alnico, a magnet of this material will be of smaller volume than one of other magnet materials for.a given amount of energy in the air gap. This reduction is so marked that Alnico is now being employed commercially for many applications formerly served by electromagnets. Itis very stable as regards decrease in magnetization due to vibration, superposed alternating fields, or high temperatures. Because of its high coercive intensity it is difficult to magnetize, requiring a magnetizing force of at least 4,000 amperc-turng per inch actually effective in the material. Figures 20, 12i, 12j, and Table ITI give data on both 5 and: 12 per cent Alnico. The 5 and 12 per cent Alnico’s are commercially designated as Alnico’s I and II, respectively. The 12 per cent cobalt variety has only 31 See Art. 19, Chapter I.ARr. 33] MAGNETIC DATA FOR TRON AND IRON ALLOYS wv slightly higher residual flux density but about 25 per cent higher evercive: intensity. ‘Table 111 alse gives data for Alnico II] and Alnica TY. Permanent magnets need not necessarily be made from bar stoek. Tivershed * points oui that tungsten and cobalt steels can be cest to form magnets slightly superior wiagnetieally to hose made fromm roiled stock. ‘These steels are very Uilficull to machine, and casting gives an easy Taeans of asanomicalty forming intriealy shapes. Alniw, ay pre~ viously mentioned, can only be cast Nipermeg. ‘This alloy which contains vo cobalt hae a highor enereive intensity than Alnieo, about 1400 umpere-tums per inch, but. a con- siderably lower vesidual flux density. Data fer this material are given in Fig, 20 and ‘Table fit 33. Magnetic Data for Iron and Iron Alloys? 11a and 1b} are given magnetization curves for the ordinary muagnelic materials used in electromagnets. The samples from whiel these data were taken are desenibed belo: (1) A carefully annealed rhig sample of Swedish chureoal iron, Grain direction oue-Lalf longitudinal; one-half transverse. Tis sample, by compurison, is magnetically identical with the best quality Amner- jean iagat icon, (2) A ring sample of Swedish charenal iron cul irom 2-in, annealed sheet. Machined with very light euis and not annealed efter machia- ing. Grain direetion one-half lengitadical; one-half transverse 3) A ving sample of high-quality, brightefinich, dead-yofl, aiild cold-raRed 4-in. steel sheet of analysis FE. 10-10. Machined with very light cuts and not annealed aftar machining. Geain direction ona hal€ longitudinal; oue-haif transvers @) A ding sample of high-quality, ty soiled g-in. sleet plate of anelysis SAF. 10-20, Machined with very light cats and not annealed alter machining. Grain direction one-half longitudinal; one-half transverse. , bright-finish, mild votd- #Pennasent Magnets in Theory and -Fractice."” $. Kvershed, Jour, EB August, 1925, This yaner is particularly recovemonded Lo anyone wicking i hensive discussion cf the properties of permanentamagnet stivls + 7 Tbmust be remembered that suet duta will vary depead: suple shed. No sach datain this chapter, athe: that Where specifically mentwned, arz tn be considered angcesurily as representa they arc merely date taken with eare on samples of standard compnereial matoials rilakly anagulacturer. you the particular of Figs, 15 and 16, and[Cnar, 11 MAGNETIC PROPERTIES OF IRON 48 ‘aiquypan Aypuyosotamod ,, 04, LOuMOD Bot oy 204 GOAN UONEZVOUSEUL SION “DTT “ory ou od suim-aredeay Aysuequy nsUBEH *900t ‘oot . oor ot vo 1d Aa we i 5 3 2 joe 03 lozt oer opt