statiotiary conducting loops (or sheets) over which the train levitates.

The interaction of thc induccd currcnh wi1h (hc inagnc~icCicld ciwics Ihc Lorces. There i s acrivc rcscnrcti in supct-condnctiog HDS Maglcv. Thcsc incliidc crlbrls at the Massachusetts Tnslilutc ol Technology and in Japan whcrc hill-scale tests arc hcing donc. Anollicr way lo tcs1 inagiictic levitntioii principlcs i s ii slntioiiary coil carrying a tiinc-varyi ng current, l c v i t a ~ l above a conducting sheet. A cui1 bc Icvitated i n a stable, hiiL undcrdanlped ecpdibriniu without feedback conti-01.Iliis experitiient deinunstrates, in ii simple way, eddy current Icvilatioti, B phenomenon that is not well understood. The iippronch we will show here is inore intuitive anrl less m~itthemalicnl lhari others, anrl lhc resalts are cotitlrmcd by R very simple eqieriment. A siimplc irictlioil is shown Ibr calculntitig the levitation height, suspensicin anrl lift-off power. rcsonanl Ir~cqucncy These scaling laws are very important to unclerstnntl for the design of electrical mnchitiery.
Analyzing l e l c w i L a l . i c x ~ h

e x y x - x itT1e rlt Ihe geonielry o f the expcri nicnlal

Q Composite 0. CantlllolExpert

Gallev

agnclic lcvitntion has numerous practical iipplicaiiotis in research atid i i i industry whem friction must be rcdmed or eliminated. Some of the inore promising :Ipplications i r e trailsportation (low and high speed Maglev), low friction bearings for gyroscopes a i d flywheel energy storage. Other applications have been proposed, such as levitation melting of conductive metals. Al)pIications--sucli a s cddy-cuircnt braking and induction 11cnting-th:rt involvc similai* physical pmccsscs as inagnclic lcvi Lalion can bc analyzed using similar or slightly modified solution kchiiiqtms. There are two tlavois of magnetic Icvilation: attractive and repulsive. I n attractive levitation, n ferromagnetic body is nllmclctl ro a source of magnetic flux, as R piece of steel is attracted to a perrnnncnt magnet. Levitation forces can hc crcafcd will1 n DC magnetic ficld crcatcil by DC currcnls, supercoriducting coils or pcrmnnent magnets.

This type of Icvitatir,n is iiinstablc withoul rcccdback control (by E N T shaws thctirem), but nutncrous analog and digilal control techniques are availithlc. A lull-wale clcctmmagiictic s w pcnsion (EMS) Milglcv S ~ S ~ C I using U ctippcr coils for gcticralion of magnelic is currcnlly being lcvlcd in Gcmiany. Ihe projectcd Icvcnuc-ploilucing train servicc would hcgin i n 2005 Irom Berlin tci Hainburg. In ~~cpiilsive levitntion or electrodynnmic or EDS levitation, eddy ciirrents w e generated in : conducting I body when the body is subjected to a time-varying magnctic [lux. Thc intcraction of the eddy currents with the magnetic flux gcnefittttcs forccs Icvitaks the body. In Ells Maglcv, thc changing magnetic flux is producctl by n supcrcnntlucting iriagnet on the inwing train. [his chiinging niagilctic Tlux gcncralcs circulating curients in slalioiiary conducting loops (or sheets) over which the train levitates. This changing riiagnetic flux generates circulating currents in

levitation system is shown it1 Fig, I . A circular cupper coil w s built by winda ing # I6 gauge insulated copper wirc 011 a nun-condactivc winding frirni. Thc resultant coil was imprcgnalcd w i l h cpoxy. Altct ctiring, ihc coil was placcd on top oF a conducting plate that is much wider than ilic coil. The copper coil was energized with 60 H7. AC with ntljustable voltage amplitude controlled by n variable transformer. When the voltage WRS of sufficicnt amplitude, the coil achieved lift-off and levitated in a slablc eqnilibrium nt height 11. By adjusting (he vollagc amplitude, the Icvihtion height can be adjusted. The tnechanistiu involved in electrodytiarnic ( E M ) levitation are identified throl1gh the usc uf M~IXWCIIS~ U L I CL tiuii s. Simplifying assnnipti o i l s arc idcntiiicd nncl tiscil 10 gcncrnk riiodels fur cvaluating thc levitation forcc ilnd Iirt-dr p m w .
Elernerimry thecxy

711ic qu;isistiitic or low Fi~ct~uciicy Ibrnis or Maxwell's cquatiotis we u s e d 10 analyzc lliis s1r~icluic. This i i i e a i i ~ that magnetic energy storagc is dominant (as cumpal-cd to energy storctl in lhc clcctric fictd) and wuvc plicooiiienn

40

0278-6648/00/$10.000 2000 IEEE

IEEE POTENTIALS

are small enough l be igriored. 'llic o ljrst law that is usehi1 is A I I ~ C 'LRw, S which simply states that a Llowing current cream a inagiiclic lield, or:

where il is the magnetic Eicld (Ampsltnetw) and ,i' is current tlcnsity

(Amps/inctcrz),In words, Aiiipcrc's I,aw mcans that the magiiet.ic Tlux dciisity intcgratcd aroiind ii cluscd coiiinur'
equids thc nct current through thc surf x e hounded by tlie contour. 1 'ildily's law shows the mcchnnisins by which ii
. . . . .. ,. , . , . .

liil force in iiii clcclrodyiiairiic Iwitslioti system: By Ainpere's law, thc +-directed current i n thc coils gcneratcs II timevarying inapnetic flux. This l'lux tins hoth axial (7.) i d d i i d (r)coinpotletits. Soruc o f this flux iinpingcs on the conducting plate bclow thc coil. 4 By Fariulay's Iliw, the changing tnagnetic f l u x iinpinging on the plate iiiduccs ail clcciric held (rind hencc, current Tlow) iii llic platc. * 'l'hc doiniiinnt current componen~ i n the plarc is i i i Llic $-direction. 'This

intlucctl ciurciits) is sinall coriip"red to the iriciilcril I'icld. Thus. the tield passcs through thc platc iis if il wcmn'l there at idl. Sincc h e r e is ~niiiitnalintlnced current, tlicrc is i n i i h d li rl I'otce. 1 1 Iig. 2b, wc sec the caw of high fic1 qiiency excilutioii whcrc the incitlcnt magnctic L'icld is shiclrlcd hotn passing through Ihc plalc. This is due to largc iiidtrccd circulating cumtits in the plalc. Nolc lhnr tlic flux lines are "sqtiishcrl" bciicntli the coil. The resultant iiiduce.cl cunents may be used to gencraic n lift L'orcc (as it1 Maglev) or may bc uscd to

current inlcracls with the r compoiicnt rri' tlie inag~ietic field I O gciierate -ti: lift

c.'

.:..

.
'

. . .

licat tlic conducting plate (as in induction he;iting). A siiiiplc calculation (showii shortly) finds this minimum plate thickness nt. :I given opcraliiig r'rcqucncy to reach the "high frequency" liinil.
C h i t.ics:%tl fixqciencies ];or an iipplicd niiignctic field kmrjiriitial to thc surl'acc 0 1 n wide flat plalc, thc cliarxteristic length ovcr which tlic I'iclrl dccays in the pliite is thc

'I
. .

'

.z

. . .

t..
. . . .

.
'

. . . .

. .. . . . . . . . . .. . . . .. . . .
'

-4

. . .

'

I 1,
.

' I h ' .:
.

- 4'' :
.:

:.

so-c;illccl skin depth 8. This is given by:

. . . . . . . . . . . . . . . . . ... . . . . . , . .. . . . . . . . . .. . .. . ... . . .. . . . . . . . . . . . con$iFttng.plate, :: . . .. . . . . . . .. . . . ~ .. . . . . . . . . .. . . .

;/;

:..

Fig. 7

Levifationexperlmenfshowing coil levlfutedelectrodynumicully above a conducting plate. The convention for hbeiing magnef/c windings is as foi/ow$:The cross indicutes current Infothe puper; #he dot Indicates current out of the paper. By the right-hand w/e,for this coil (lor DC currenf excituflon) fhe uxid mugnetic field at r=U is tho -z direction.

wlicrc~fi IIW r)pcrating frequciicy, s )-I is tlic inagiictic pcrineability OT thc platc arid ci is the electricill conduclivily 0 1 ilie plate. l'hc
iiicigiietjc field insiilc ihe plntc deciiys with th is chnrac ter i s tic

cliatiging tii;ignetic flux picrates circw lating (or eddy'') currents. The relationship i n II c o n d u c l i ~ ~ g "Ohmic" inaterial relating the curlw~t density mid electric field is . I ' = 0.C nrir~we can tlcrivc:

opposcs ihc incidcnt

quency, b) limits

l'hc twin on the right is the ncgarive or the time rate of change ol' the miignctic Clux throngti the surfacc h i n d cd by l h c c o n t o u r dk. This Is i h c mcchnnism by which a changing magnclic flux inipiiiging on a conductor
creatcs cdtly currents. Furtherinurc, the hreritz force lilw states that a inagnctic L'orce results If there is a curmiit flow in n region wiicre there is niagiictic i'liix, by b-'=j'<~13. Using h s c t h i w principles, we can identify the nicchaidsm for creating +z

FfBRUARY/MARCH 2000

41

F/g.3 Electrical model of magnetlc levi!afion system

cvalunting the driving inipetiance at the terminals of tlie coil is shown i n Fig. 3. Rcoil is itie resistance of the coil in free-space due to the finite resistance of the wirc. L is the frequency anrl geometry dependent inductance seen at the coil tcrnmiiinls. R," is the resistance due to eddy-current losses in the conducting platc. Wlicn tlic coil is brought near thc plutc, R, increases and the terminal iiiiluctniice uf thc coil decreases. This is because the field beneath the coil is modified due to induccd cutrcnts. AI high frequencies, the space beneaih the coil IIRS the flux concentratcd bchvccn ihc coil and the plate, atid hence the ineasuretl coil inductance is rcduccd from rhc lrcc-space value. One possiblc Ctmctiond dcpcndcncc lor tlie incasurcd inductancc is:

varies (from the l,(z) term), thcrc is ii resultant force acting on the coil. Using the energy methud, thc foorcc duc to an encrgy ficlrl cui bc casily cnlculntcd. Also, tlie forcc acting on Lllc coil equals t~ic gi+&icnt (or spilinl ciei~ivalivc)of the encrgy ficld. I n lliiR casc, tlic stored inagnctic cncrgy

calculate) 1lic charactcrislic decay length for the terminal inductancc.

E x p r i n i c ? n t a Iresults

Fig.4 Approxlmute

pole locaflonsfor
vertical motion of eddy-currenf

A 1 -cctilimctcr hick aluininunz plate for inductance nieasurernents ntid levitation experirnetits, as a skin dcpth calculation shows that this is the m i i i i n i u i n approximale thickness EO nchievc good rnagnctic sliielding. Pammeters uf tlie copper test coil givcn in Table I .
WRS used

magnetic suspension

Coil inckiclzince measuremen k

changcs as ihc z posilion of the coil varies, anrl ihe magnetic force pulls the coil lowaids a posiilion of higher stored cncrgy. The Iorce acting on the coil in

thc +s direction is:

The term L, is the terrninal inductance of h e coil when it is well away from the plate. The term I.,. nccnunts for the fact that the terininal induclancc decreases when the coil is iicar tlic plate, duc i o the cdrly currents induced iii thc platc. Tlicrclbrc, thc inductance is i function OF Ihc z height of (lie coil l abovc thc conrliicling plate. The decays have the characteristic length scale y. (Note: This is not the only possible mathernalical description, but it yields particularly siiiiple arid surprisingly accurate results.) 111OUI' thought experiment, let's assunle that thr: coil i s drivcn by an AC current source. The magnctic cncrgy stored in tlic inductor is:
I E,,,= - I,(e)r* 2

'1'0 dctcrrnine thc sinhilily ofthc Icvitated coil, we assiinic that the total forcc acting on the coil is ail cquilibi.iuin value plus ii lierturbecl valuc, urh = Fz + .,"2. Furthermore, we assiitne that the position of the coil is given by an q u i libriitni position, plus n sinal1 perlurbatioii, or z = Z(, + 2. Thc cquilibriiirii h c c Fx nutst cqual thc wcight of thc coil, or Mg if thc coil is 10 he stably levitatcd. l3y using our ussumcd coil iiidiictancc given abcive and hy assuiniiig ihat pcrturIiations in z wc sinall (f << y), the Ibrcc wring on thc coil is:

R y the energy method, magnetic forces can be calculated if the change in temiiiid inductance is known. The coil inductance under various test conditions WRS tneasured using B Hewlett-Packnil-rl HP4192A innpedance analyzer. lii it first experiment, the coil inductance in the 30 I-Iz to 5UO I-lz rangc was measured widi the coil L O millimcters above the nlumiiium plate (Fig. 5). The estimate of llic s k i n dcpth shows that there sliauld bc signilicai~l shiclding at 6 0 Hz. IIcncc, thc tcrrninal i n d u c k " should hc grcatly al'l'cctctl. At 60 Hz, the tcrmiiial inductancc is 824 microljenries (down rl.c)111 LIIC ~ C C - S ~ vaiuc or 9x0 ~ W microoHcnrics). 'I'his confirins our suspicioii tlin~1hci.c is significant magnciic shielding at 60 Hz. Tri n second expcritnent, the 60 Hz terininal inductance W R S measured for coil heights of 0,5, 10, 15 and 20 tnillitnctcrs above the plate (Fig.6). The data was ciicvc IIt to ~ i i tcrrnitisl ittducr tancc expression (Eq. 4). The paratneters for the curve fitting are sliowii in Table 2. This data was used to estimite tlic inininiiun currcnt iiccdcd ror coil lift-uff and thc resonant frcqucncy o f
thc suspcnsioii.
Coil "IifLafi" and height Sincc wc now know the terminal inductancc, thc currcnl necded Lo

where we makc usc? of Ikc lac1 lhal c - ~ = I - x for x' << I . 'Ihcrcforc, thcrc is : I restoring force. 1 lor small pcrturliillioils, this suspension a c l s I i kc a inagiiclic spring with spring constan1 k,. Tlic nnlurd I - ~ C ~ L I C I K ~ ofS this sysicin arc 011 the C jw axis (Fig. ns in the cme of ii simple 4) niass and spring. The frequency of oscillation for this system is given by:

achicvc Icvitntion can hc casily estitnntcd. By the energy mcthod, the necessary current to achieve lift-off is calculated to be:

N o k h a 1 cvcn though the current is allcmating, thcrc is average energy storage in the coil (due to the P term). Since this cncrgy vatics as the coil position

This interesting result shows that it i s easy to find the uscill~tioii frcqlrc~icy or the suspension if we cuii incitsurc (or

For this calculalion, ilic icrm dL(z)ldz = L,/y, Thc actual currcnl io achicvc lilt-

42

IEEE POTENTIALS

i'orc, our estimate for I:r eq ii e 11 cy (givcn by Eq. 8) is f,,, = 3.5 IIz. Thc coil was 1cvi lntccl at a height I O
r c s o n ail t

inillimclcrs ahovc 1hc plate and houiicctl, and

21

resuniiiil frequcncy of

approxiinateiy 4 H z was rnensiii-ed.

M~CJtic ne

l3ws?, Magnetic s c a1 i iig . , . , .: .. : .:20'mlIlimet6rs.. : . . . '. .. . . . . . .. . . . .. . . .., . . laws show that large rnngnclic, eletiietits :ire iiiot'e efficient in energy conversion than smaller ones. Convc r sc 1 y , s 111n11- s c n1 e I cvi talion cxpei-imcnts arc Iik c l y to hc v c r y power h u n g r y ((ii' uiiahle to levitate :it all befurc thcy off was approxi oiately 2 1 A (KMS), buro iip). For the test coil, this cflcct with 2hA rcsultirig in 1 millimeter Icv0 can be quantified by coiisidcring tlic itntion height (Fig. 7), aiid 39A resultratio of the lift fotce to the powcr dissiing in 20 ~nilliineter levitation hcight. The I ift-off power neetletl is itppi~xi- pation. For the thin d i s k coil, ~ l i c inductance i.. iH i i ~ ] > Iiiwtcl y pi+c>poi+~x mntely 168 Watts, with 257 Walls tional to ( I , [tic coil radius. needed at h = I O milliinctci*s a n d 577 The lift forcc i s pmporlioiial to Watts needed at h = 20 millinic~ci~s. A d,(z)/dz, which i s a l s o proportional to summary 01' measurcd and calculatetl the coil radius ( f . Thc powcr dissipacurrent needed ;it a givcil Icvilatioii tion is proporiioiial tu thc rcsishricc of height is giver) in 'lablc 3 . the coil. 'lliis, i i i hirn, is proportiotial R e s o r - i a f l t. f ~ ? e q u c x i c y to d b c , the ratio ol:ciirreiit path length of suSpt~r1~;iorl to coil cross-sectional areti. Therefore, the ratio of lift forcc to powcr l o s s i.: The ciirve fit shows that lhc charucpi~oportionalto hc, or tlic cross-sccteristic decay length for intluctancc i s lionnl area of the c o i l . If ill1 coil approximately 20 millimcters. Thcrc. . .. .

5xx*lir-Q

:
'

. . .

'

'

..,A{

lengths are sc;ileil up by the same factor 1, the ratio of lift force to powcr dissipotion iiicreases by the factor Iz, 0 ' the length squiirerl. 1 To achieve lift-off of the test coil in his cxpciiincnt, appi-oxiinately 168 Walls is dissipalcd in Llic coil. Furthcrmarc, as Lhc coil hcals tip, more power rnust be dissipated for. tlic SiiInC levitation height. The reasr)ii is the w i ntl iiig res i s taiice i 1i~:reusesw it11 teinperatu re. (The tetnperatare cue r k cient of the resistmice of coppcr is ;ipproximately = 0.4% per tlcgrcc C.) So, the test coil gets hot, :incl can oiily be run f o r a few seconds at a timc. As shown hy thc iiiagnctic scaling laws a l u g e r coil coiild bc lcvilalcd for lungcr periods oC time.
c<~~lch3sic~rls

We presented R simple demonstralion ol' eddy-current magnetic levitaiioli using a s m a l l c o p p e r coil, ciicrgizcd with 60 11x AC arid levitntetl ovcr an nlaminurn platc. Thc processes ihat gcncixtc inagnctic l'orccs are itlent i l i c d wing M a x w e l l ' s eqiiatiuns. Tlirougli this experinleiit, a inetliod for dctcrrnining lift-off power, levitation height and suspension resonant k c queocy was shown. The principles outlined i n this arLicle have tiunietotis iipplic;lti(lils LO inagiletic levitation, induction hcatiiig and 0th CI' cl cdrod yii tunic processes involving induccd cddy currciils. Scaling laws sliow how to size II siispensioii cniiduc~orand p w c r supply for a givcn lcvitatcd load.

. .

. . . ...

' .

. . . .

..

.. . . . . .

. .. . : . . . . .: .

.. . .

.. .

. .

. .

Fig. 5

Measured termlnal lnductunce, coil ID millimeiers above aluminum plaie

Fig. 6 60 Hz coil inductance, for various coil heighfs above ulumlnum plate. Doffed /in@is curve fit fo datu

FEBRUARV/MARCH 2000

43

Tmnsncrions on Magnelics, vol. 35, no. 3, May 1999. P. J. Gcaty, Magnetic and Electric
Suspensions, British Scientific Instrument Hesearch Association, I964 T. Clark, A. Kondoleon, M. Thompson and R. Thornton, Shaftless, Magnetically Levitatcd Flywheel Energy Storage System, Aermsprrcr Flywheel Workshop, Albuqucrquc, October
Flg. 7 copper coil ievltafed approximately 10 7-8, 1998.

milimefersabove uluminum plate.

Read more about it

S . Earrishaw, Oil thc Nature of the Molecnlar Forces which Rcgnlate the Consti~utioiiof the 1,uminifcrous Etkcr, Twtsactions of rhe Cuinhidge Phihilosr$riod Society, v. 7, 1842, pp. 97- 1 12. 1. R. Powell nnd G. R . Danby, High Speed Trailsport by Magnetically S11spe ndetl Trains , A XME Pir blimtirin No.66 WA/RRS (Deceinbcr 1966). C . A. Guderjahn, S. L. Wipf, H. J. Fink, R. W. Boom, K. E. MacKenzie, D. William, and 1. l~owncy,Magnetic Suspeiision and Guidance for High Spccd Rockets by Superconducling Magiicts, J o i i r i t d of Applied Physics, vol. 40, no. 5, April, 1969, pp. 21 33-2140. H, IT. IColm and I<. D. Thornton, Thc Magneplane: Guided Electromngnctic Fliglil, Proceedings of the 1972 Appiierl S u p i*cmiduc li vii+v Con feren ce, May 1-3, 1172. Marc T. Thompson, High Temperatore Superconducting Magnetic Sirspcnsion For Maglev, Ph.D. Thesis, Dcparlincni of Electrical hgineering aiid Computer Science, Massxhusctls Institute of Technology, May 1997. Marc T. Thompson and Richard D. Tharnton, Modeling UT HTSC Based Iron-Core Flux-C:aiiceling Elcctrodynamic Suspension for Maglev, Procc erliirjis oj~ the 4 th I I Ire m ~7 t io na [ Symp(isiiim o n Magtietic Suspension Technology, Gjfu, Japan, NASA Conference Publication, 1997. Marc T. llioinpson, Richard D. Thorn ion an d A n t h on y Kond oleo n , Scalc Modcl Plux-Canceling EDS Maglev Suspensiori-Part I; Dcsign and Modeling, U X E Trcinsaurioias on Mugrwtics, vol. 35,no. 3, May 1999. Marc T. Thompson aiid Richard D. Thornton, Scale Modcl Flux-Canceling EDS Maglev Suspcnsion-Part 1 : Test 1 Results iu1d Scaling Laws, IEEE

H. U. Wicdcrick, N, Gauthier, 13. C. Campbell and P. Rochon, Maglgnetic Braking: Simple Thcory and Experiment, Amcricari Journal o Physics, f vol. 55, 1gX7, pp. 500-503. 9 M. Adlcr, A Field-Theoretical Approach tn Magnetic Induction Heating of Thin Circular Plates, I E E E Transactions on Magnerics, vol MAG-10, no. 4, Dcccmber 1974, pp, I 1 18-1125. A. E. Fitzgerald, C. Kingsley, Jr., and S. U. Ihnans, Electric Machinery, 5th Edition, McCraw-Hill, 1990. 12. W.Gmvcr, Inclrartmce Ufilculatioris: Working Forinulrxs uiid Tables, Dover Publications, Inc., New York, 1946. Marc T. Thompson, Inductance Calculation Techniqucs-Part I: Classical Methods, Power CoritroE uiid Intelligent Motion, vul. 25, no. 12 December 1999, pp.40-45. Marc T. Thompson, Inductance Cnlculntioti Techniques-Part 11: : . . : . . Appro xi in a t i o n s and Handbook Methods, Power Control m d h l k / k gent Mofion, website http://w w w.
I

sultant and Adjunct Associate Pidcssor of Electrical Engineering a1 Worcester Polytechnic Instiliitc, Worccstcr Massachusetts. At W.F.I., he teaches iniuitivc nicthods for analog circuit, inagnclic, thmmal and power electronics design. His iiiaiii research at M11 concerned the dcsign and test of high- t empcrat uic sup ercond uc t ing suspensions for MAGLEV and the jmplenientation of magnclically-based ridc quality control. Other arms of his res et1rc h and cons U 1t i n g i ti ter e s t include planar magnetics, pnwei electronics, high speed analog design, inductioii heating, 1C pwkuging f r ~ iinprovcd theritial arid electrical pcrhi.niaiicc, 11~1:of scaling laws for electriciil and magiictic dcuigri, atirl high speed laser diodc modulation techniqies. Hc has workcd as a consultniit in nuiialug, electroincchanics, tnechnnical and inugnctics dcsign, and holds 2 patcnts. Currciitly lie works on a variety of consulling projects including high powcr and high speed laser diode modulation, cddy-current brake design for ainusciiicnl appl icati m s , flywheel energy storagc Lo satellites, and magI nctic tracking for inter-body cathetcr positioning and is a constiltatil for Magnernotion, Inc., Uiiitcd States Department of Trmportation, Edward M. Pribonic, 111 IHC.. and Polnroid Corporation.

pcitn.coin,
Akx3LJt

t o h

author

MMCT. Thoinpson (M 92) was born in Vjnalhaven Island, Maine. He received the B.S. E.E. degrcc from the Massachiwits institute of Technology (M.I.T.) i n 1985, .the M.S.H.H. in 1942, tlic Electdcal Engineers degree it1 1994, and the Ph.Il in 1997. Presently he i s an engineering con-

44

lEEf POTENTIALS