PULSED SUPERCONDUCTING MAGNETS

G. Bronca, P. Genevey, F. Kircher,J. Perot, J.P. Pouillange, G. Prost

Departement du Synchrotron Saturne,CEN/SACLAY (France)

Abstract the Rutherford Laboratory and the Depar­tement Saturne (Saclay) in Western Europel

The possibility of using supercon­ducting magnets to increase the energyof present accelerators or to get veryhigh energy in new facilities, has beenstudied in many laboratories allover theworld. We review in the present paper thework which is made concerning the theor­etical aspect of the behaviour of super­conductor in pUlsed field, the presentconductor status, the field tolerancesand correction, the magnet constructionproblems and finally the experimentalresults obtained recently in testingcoils of different sizes and shapes.

I. Introduction

In the general line of the applica­tions of superconductivity to magnets,the field of the pulsed magnets for acce­lerators was quite intensively studiedin the past years. As soon as the theor­etical studies showed the necessity ofusing a few micron twisted filaments toreduce the hysteretic losses and whenthese conductors were obtained from"in­dustries in long enough quantity and withgood quality, many projects were made.Two kinds of facilities were studied:with the first one it will be possible toincrease the energy or the intensity ofpresent or being built machines (Argonne,Batavia, CERN II) by a factor of 2 to 3using the same tunnel. The second possib­ility is to design a new machine with ahigher energy than that existing now, theuse of superconducting magnets giving newperformances and or lower costs (ISABELLE).

In these machines the new aspectwill be the magnetic system and its envi­ronment, namely the multipolar magnets,their cryostats and their electrical andcryogenic supplies. The main characteris­tics of the machines, mainly the magnets,are studied and discussed in the labora­tories working in this field : NationalAccelerator Laboratory (Batavia), ArgonneNational Laboratory, Lawrence BerkeleyLaboratory, Brookhaven National Labora­tory in USA, the Radiotechnical Institute(Moscow), the Electrophysical Insti tute(Leningrad) in USSR , and the Institutefur Experirnentelle Kernphysik (Karlsruhe),

203

The dipoles which will be the lar­gest part of the magnets will be discus­sed in the following chapters, manyresults being also applied to the focu­sing quadrupoles and hi gher order cor­recting lenses.

The maximum field value which hasbeen considered varies from 4 T to 6 T.It should be clearly understood that fora given total radius the energy of themachine is proportional to the field butthat construction problems and cost willalso increase for a higher field. Whenthe final energy has been fixed someeconomical studies show an optimum fieldvalue around 4.5 T to 5 T. However asusual, this optimum is quite flat andthe choice of the field is still open.

The bending magnets must have a ho­mogenous field. Homogeneity is definedthrough the field integral performedalong the magnet for different positionsin the bore. The tolerances which arestated now is a relative field integraldifference of a few 10- 4 in all theregion where particles will circulate.

One should add some space to takelnto account magnet construction tole­rances to this useful region defined bythe beam width, particle oscillation,injection and extraction conditions.Finally the coil aperture we are led to,corresponds to a diameter of 5 to 6 cmfor the Batavia doubler, about 8 crn forthe ISABELLE project whereas the Europeanlaboratories are designing magnets of 8to 11 cm diameter for the CERN II super­conducting alternative.

The last point to be mentioned con­cerns the cycle duration. The basic valuecorresponds to the 8 to 10 s pulse dura­tion of the European 400 GeV conventionalmagnet version. However it is clear thatincreasing this value to 30 to 60 swillhelp in solving many problems such asmagnet technology, low temperature losses,refrigeration, power supply, etc. Thechoice is still being discussed in Europewhile it is clear that the cycle duration

for ISABELLE will be a few minutes.

Approximation of J0 cos eby four blocks

Approximation of Jo cos e byspaced sectors of constant

current density

L...- ..L.lll:.......L...-........"'--...x

y

y

We will review now the work made inthe past years in the study the designand the construction of pulsed supercon­ducting magnets.

II. The production of the field

For all the sources of inhomogeneitytolerances are prescribed so that theireffect may be corrected by systems ofsmall dimensions.

In this kind of magnet, the conduc­tors give the dominant part of the field;so, the location of the currents is mostimportant; consequently, the coil confi­guration must be carefully designed and,during the construction, the location ofthe conductors must be sufficiently accu­rate so as not to be an important sourceof inhomogeneity.

Studies taking into account the ma­gnetic iron shielding are however neces­sary to calculate its contribution to thefield and to minimize the saturation ef­fects.

The c n are called mUltipole coeffi­cients.

y

Concentric loyers of constantcurrent density approximatingeither J0 cos e or intersecting

ellipses

(1 )

(2)z = x + iy

In two dimensions, one can analysethe complex field inside the aperture ofa coil by a Taylor series

i By = 'Z.n=1

B

where

For a perfect dipolar symmetry shape,the dipolar coefficient C1 and its oddharmonics (C3, Cs, ... ) are only present;these coefficients are all imaginary.

Starting from the two configurationsgiving a pure dipolar field (intersec­ting ellipses and cose current densitydistribution on a circular or an ellip­tical boundary), one has to make practi­cally windable approximations. Studiesmade in different laboratories,1 ,2,3,4lead to configurations p~oducing a fieldhomogeneity of a few 10- into 60 to 80per cent of the aperture (fig. 1).

y

Approximation of intersectingelLipses by shaped blocks of

constant current dens~y

Fig. 1. Shapes of coils under constructionin differents laboratories.

204

These results may be obtained eitherby the cancellation of several multipolecoefficients or by balancing their effectin the median plane (fig. 2).

10-3 9,(ll,0)- 9,(0,0)/9,(0,0)

5

-1

-5

One can write :00

K Kx - i Ky =L d n-1 (5)znn=l

that is to define the multipole coeff­icient of a coil as its integrated valuefrom s = - oa to s = 00 . Calculationshave shown that it is possible, to mini­mize these coefficients dn , by shapingthe ends. 6 ,7,8,9,lO However, these endeffects are small for the long dipolestha t one bears in mind now (# 6 m forthe superconducting version of CERN II).

10-3 8,(ll,0)-8,(0,0) 19,(0,0)

ll(em)

-1

-5

Fig. 2. Production of a good field homo­geneity either by cancellation ofseveral multipole coefficients(above) or by balancing theireffect in the median plane (below)

y

l..-._--..L_....1£_~L----'~_L..-.l....-_--.)C

One can also possibly conceivewindow-frame or H-type superconductingmagnets 5 ; the shape of the coils is thenvery simple but one has to shape theshielding to minimize the sextupolarcoefficient at low field and to set add­itional corrections to minimize the sa­turation effects at high field (fig. 3).

Practically, one has to take intoaccount the finite length of the magnetand its ends, wound as to keep free thevacuum chamber. The results given in twodimensions for Bx and By are also validin three dimensions for the field inte­grals K and K along the azimuthal axiss: x y

s= 0'0

K (x,y) ) Bx (x,y,s) ds (3)xs=-00

K (x,y) SJoe> By (x,y,s) ds (4)ys= _00

205

Fig. 3. H-type dipole (Brookhaven Natio­nal Laboratory).

First conceived to canalize the ma­gnetic flux, the iron shielding is alsosuch that it increases the field and itreduces the stored energy; the magnitudeof these two effects depends on the loca­tion of the shielding ; most laboratoriesput the shielding inside the cryostat(cold iron) in order to make the best ofthese advantages. Typically, for a fieldof 5 T in an aperture of 10 cm, the con­tribution of the iron shielding is about1.S T with a c9ysequent reduction of thestored energy.

This laminated iron shielding canbe of circular or rectangular shape. Thepresent computer programs such as GRACY,LINDA, MARE, MAGNET, POISSON, TRIM enableone to locate and to shape the iron shiel­ding in order to minimize saturation

effects; eventually, one can take intoaccount these saturation effects to can­cel one multipolar coefficient; but thisresult is valid for only one field value.

Neglecting the saturation effects,one can prove analytically that a cir­cular and well centered iron shieldingaround a circular symmetry coil improvesthe field homogeneity; for a non circu­lar symmetry coil, the iron shielding cangenerate harmonics which were not presentfor the coil alone.

Other sources of inhomogeneity thanthose coming from the conception of themagnet (reference shape of the coil,finite length, ends, saturation, ... )must be taken into account. They resultfrom the winding of the coils and fromtheir location in the iron shielding. Allthese sources may be listed into threetypes :

- systematic errors for all themagnets.

- random errors from one magnet toanother (they affect the field and itshomogeneity).

- random errors inside one magnet(they affect the field homogeneity).

The sources of inhomogeneity due tothe conception are of the first type andthe sources due to the winding and thelocation are of the three types ; onecan distinguish

- location of the conductors insidethe coil, especially those on the bound­aries.

- location of one half coil comp­ared with the other.

- location of the whole coil intothe iron shielding.

The accepted tolerances in order toget a few lO-~ in the field homogeneityin 70 per cent of an aperture of 10 cmare :

- random location of the conductorswith a root mean square value of about0.3 mm. 12 ,13,14

- location of one half coil comp­ared with the other with a rms of0.05 mm. 13 ,15,16

- location of the whole coil insidethe iron shielding with a rms between 0.1an 0.5 mm. 13 ,16

Copper models have shown that thesetolerances may be kept. 2 ,3,l7

206

For all the sources of field inhomo­geneity, let us say again that it is offirst importance to distinguish the syst­ematic and the random ones. For an acc­elerator with a great number of magnetsthe whole effect on the beam of therandom errors will be less important thanthe effect of the random errors in onemagnet, due to compensation; for the syst­ematic errors, the whole effect will bethe same as the effect for one magnet.

The distinction is however delicatebecause it often depends on the construc­tion process ; on the other hand, formost errors of construction, one part issystematic and the other one erratic.

Whether one attributes the errorsto one magnet or to a great number ofmagnets, the useful aperture, defined byI.6B/B total I ..{.. a few 10- 4, is between 60to 80 per cent of the aperture. 13 ,15,18Other definitions of the useful apertureare possible but lead to the same res­ul ts,15 that is a margin of 10 to 20 mmbetween the end of the useful radius andthe conductors for a total aperture of10 cm.

Due to the nature of the supercond­ucting wires, diamagnetic and circulatingcurrents with very long time-constant aregenerated in the wires after a firstfield-rise. These currents generate aconductor diamagnetic field with a strongsextupolar term and so disturbing duringthe injection.

The theory~9 in reasonable agree­ment with measurements~0,21 shows thatone has to reduce the size of the filam­ents and the critical current density forzero field; incidentally, the higher theinjection energy is, the weaker thisdisturbing effect is.

New theoretical results 22 have shownthat this effect is variable during theaccelerator cycle ; the figure 4 showsthe calculated perturbation due to theremanent field during a triangular cycleup to 5 T for the dipole MOBY with itsiron shielding ; the penetration of thesuperconducting material and the factthat the critical current Jc in the super­conductor is a function of the maximumfield are taken into account.

This variable effect enables us tocorrect this disturbing residual fieldwith dynamic correcting devices.

AS (Gl at 70 per coniof tho aport\n

Fig. 4. Perturbation due to the remanentfield during a triangular cycle(profiles of penetration of thesuperconducting material areshown at different moments of thecycle .)

Several correcting methods are poss­ible to improve the field homogeneity:

- correcting lenses independent fromthe magnets

- multipolar corrections made of Nblocks regularly located within the maincoil aperture ; these corrections are 23supplied independently of the main coil

- corrections made with short-cir­cuited windings; they are located withinthe aperture and the induced currentscorrect instantaneously the unwantedharmonic during the whole cycle. 13Experimental results with a sextupolarsymmetry device (fig. 5) have shown areduction of a factor 6 of the sextupolarterm.

Fig. 5. Self correcting device.(CEN-Saclay)

207

The variable aspect of the conductordiamagnetic field enables one to use thismethod for its correction.

III. Conductor

1. I!!~9!Y

Building pUlsed superconductingmagnets implies using composite conductormade of thin superconducting filamentsembedded for mechanical reasons in a nor­mal material matrix and twisted for elect­rical reasons. The effects of such a con­figuration were studim theoretically fromboth points of view : hysteretic lossesin the conductor with transport currentand stability (flUX jumps, wire movementsetc ... ).

If twisting the conductor may beconsidered as a perfect transpositionwith respect to the uniform externalfield, it is clear that it is not thecase when one considers the field dueto the current in the conductor itself,namely the self field. This imperfecttransposition leads to a change in thecurrent sharing between filaments whichhas been analysed both theoretically andexperimentally.24,2S,26,27

The calculations show that, close tothe maximum transport current for thegiven field, the dissipated power growsrapidly. This effect will have practicalconsequences on the design of the coilcooling.

Contrarily to what was believed someyears ago self field effects do not leadto important extra 10sses 25 ,28,29 andinstability. However, the stability re­mains slightly lower than one thoughtaccording to simple theory.30 However itappears that, for a given matrix to su­perconductor ratio, increasing the totalnumber of filaments means a larger twistpitch due to a larger strand diameter andconsequently more losses ; equally theadiabatic composite stability is lower.It seems that an upper limit will be rea­ched for a conductor of several ten thou­sand of filaments. 26 On the other handa conductor made of filaments locatedonly in the composite outer part shouldbe more stable than a conductor where thefilaments are uniformly located.

The non orthogonality of current andfield due to the twist gives actually alocal increase of the current density j .The measured change in 5c when taken in~oaccount could show in particular casesonly a very slight increase of the losses.

The influence of the temperature onthe losses was studied. 2S Higher tempe­rature means lower current density andlocal change of the electrical field. Theeffect is an increase or a decrease ofthe losses depending on which one of thephenomena is the most important. It hasbeen checked qualitatively.l ,32

The only superconducting materialnow being used for pUlsed magnets is theniobium-titanium alloy. Depending on thealloy characteristics and heat treatment,the critical current densities are quitedifferent, typically fOg a 6 T field theaverage value is 1.2 10 A/m2. Highervalues can be obtained with particularheat treatments and impurity but then themechanical properties lead to fabricationdifficulties (breaks) during wire drawing.

One must remember that for a 6 Tfield no current will flow in Nb-Ti wire;if its temperature is 6.SoK the corres­ponding value is 7.SoK for a 4 T field;stated in other words a 0.2°K temperatureincrease of a conductor in a 6 T fieldmeans a 10 % decrease of its criticalcurrent density. From this point of viewthe materials being studied now : filam­entary Nb3Sn, V3Ga, Nb3Ga are quite prom­ising for the future.

Losses and stability considerationscalled for reducing the filament sizedown to a few microns.

The need of a many thousand ampereintensity in the magnets leads to a totalnumber of S to 10 u filaments of aboutlOS. Today a strand is made of 300 to1000 of these filaments embedded in anormal matrix, the O.S mm diameterstrands being then assembled together(fig. 6). Very new conductors are testedwith a hi 3h number of filaments perstrands. 3

A higher number of filaments allowsthe use of a higher conductor which willfacilitate the fabrication of the finalconductor and give a larger averagecurrent density in the coil. Howeverthere will arise, as already stated,stability and losses problems due to selffield effects. This will limit very like­ly the number of filaments to a few tenthousand. 26

208

Fig. 6. Typical superconducting compositec0nductor (above) and filament(below).

4. ~~!r!~_~~!~r!~1

The choice of the matrix material isdetermined by contradictory requirements.Copper which has good stabilisation pro­perties does not always limit enough theinduced currents. Due to its high resist­ivity, cupronickel exhibits lower extra­losses (smaller induced currents) or fora given percentage of total losses allowslonger twist pitch and eventually blggerstrand diameter. However coils made ofcupronickel matrix conductor were lessstable. 26 A good compromise should be

given by a two component matrix, whereeach filament is directly surrounded bycopper, and then embedded in cupronickel;this conductor is not very stable (it hasnot enough copper) and quite difficult todraw out. Conversely when a very smallamount of cupronickel around each fila­ment and the copper is the matrix theconductor is stable but induced currentsbetween filaments located for one fromthe other, are important and extra lossesmay be obtained for big conductors.

A more sophisticated solution isbeing tested33 (fig. 7). Almost all thelaboratories studying low rise time fieldmagnet using conductors with less than1000 filaments prefer now copper matrix.Many discussions are still important inthis field; the solution will dependalso on the strand assembly difficulties.

Fig. 7. Transverse section of the 13 255filament composite (I.M.I.)

Whatever the choice of the matrixwill be, the matrix to superconductorratio which was reduced down to 1 to 1to allow high average current density isnow usually chosen between 1 to 1 and 2to 1 after experimental results on smallcoils showed some instabilities for toolow ratios.

5. ~!!~~g_~~~~~Q1r

As long as a 10 5 filament conductorhas not been constructed with good coilsperformance, it will be necessary tocontinue strand perfect transposition toget the final high intensity conductor.

209

Braiding or cabling methods can be usedin one or many stages (fig. 8). It alwaysgives low filling factors (50 to 70 %)even with high compaction and obviouslylow average current density. One gets inthe final conductor a current density ofabout 25 000 A/cm2 at 4 T field and16 000 A/cm2 at 6 T field.

Fig. 8. Typical superconducting braid(above) and cable (below).

Whatever the assembling process is,theconductor is more or less subject tostrand movement which leads to degradat­ion and lower effective current in thecoil. Such a trouble can be overcome byimpregnating the conductor itself withan epoxy or a metallic filling material(Pb-Sn, In-Th, Ag-Sn, ... ) and also thewhole coil with an epoxy. Both the strandvarnish insulation, and the epoxy havingbad thermal conductivity properties, agood and efficient cooling must be desi­gned to avoid a temperature increase.When the strand is insulated, the expe­rimental results show no extra losses dueto current between strands, but sometimesdegradation and very often training wasobtained. 1 ,17 Moreover it is hard tocompact efficiently the conductor without

shorts between strands. Very recent testsusin~ ~oly~ondex insulation seem to bepromIsIng.

The presence of a solder shows avery high increase of the measured lossesas obtained with badly insulated strands,but heat treatement bring them back to aacceptable level. So, it could be accep­ted as a solution when the field riserate is low (few kG/s) as the degradationdue to the high losses rate, is not im­portant. It should also be noted that thesolder improves the stability by enthalpyeffect.

Another proposed solution21 consistsin insulating each strand made of a com­plex matrix by an oxide layer. Due to itsresistance, the layer will limit the in­duced currents between strands and thenthe extra losses while allowing an eff­icient high temperature compaction. Itis still necessary to understand whichof the high resistivity material in thematrix or the oxide would be the mostefficient barrier. In the same line, workis done at Brookhaven34 to reduce theinduced currents by having a highly resis­tive bronze layer being formed aroundeach strand by a heat treatment. Itappears to be efficient only if the layeris thin enough. From the frequency depend­ance losses one can conclude that solder­ed conductor will be used only for smallfield rise rate.

Finally impregnation of the wholecoil prevents degradation due to conduct­or or coil movements but sometimes degra­dation was observed and may be due tostresses appearing during the coolingdown.

Withtoday's knowledge, the choiceof the number of filaments, type andmatrix ratio and assembly (insulation,compaction, filling material, etc.) isstill being discussed within the groups.The laboratories are engaged in differentexperimental lines due to different pre­liminary choices concerned with fieldrise rate, size of conductor, coil designetc. It seems very likely that comparingthe work which will be done in the twocoming years will lead to the right con­clusion for each project.

IV. Construction problems

1. Co i 1

Whatever the shane of the approxim­ation (starting from J o cos e distribut­ion or intersecting ellipses), one has to

210

build the coil very accurately and bendup the ends. The winding process must fitthe conductor size (1000 - 5000 A) andits shape. A very flat and wide braidenables us to wind coils with one turnper layer (up to about 4 T ; for higherfield, practical considerations lead touse more than one such coil). A squarecable (or rectangular with an aspectratio close to 1) lends itself to awinding in pancakes or double pancakesdirectly on cylindrical shape former(saddle shape)21 ,35 (fig. 9).

Fig. 9. Trial coil winding of a doublelayer pancake coil of AC4.(Rutherford Laboratory)

Special shapes 4 ,35,36 with a constantlength of turn per layer, give a stablemechanical winding with only one pieceof conductor per pole (no joints), (fig.10 and 11). Generally one bends up theends during the winding process but forsome magnets 3 pancakes are wound flat,each end being bent up afterwards (fig.12) .

Construction tolerances requiresophisticated equipments to be reachedand numerous copper models, up to onemeter long, have been carried out 3 ,35to check the effects of constructionerrors on field map.

Although these models are still farfrom an industrial version, it is gener­ally thought the required tolerances canbe achieved.

•

Fig. 10. Coil winding of MOBY dipole(CEN/SACLAy) .

,d....."'''.'. y-''';'

Fig. 11. Coil winding of dipole number 8(Lawrence Berkeley Laboratory).

211

~r",--...._ ".,"-':- ••'~" I"'~..!.!1..'. •._~ __ III.

~. - ,I'

... __~l ....

l ~ '. - ~

I :~:,,~~~ .. ·.:

Fig. 12. Dl dipole magnet(IEKP Karlsruhe).

With copper to superconductor ratiosclose to 1 : 1 needed to reach high cur­rent densities, the conductor becomesvery sensitive to wire movement and thecoils are often impregnated to avoidthese movements. Impregnation materialsare usually filled or not epoxy resins.The impregnation is made either duringwinding directly on the conductor, orunder vacuum for each pancake, or alsounder vacuum for the completed coil ; theconductors are generally kept in placeduring winding by a little amount ofthermosetting material for the last twocases. Special effort is made to buildvery compact coil in order to get anequivalent Young modulus as high aspossible and consequently to minimizedeformations under stresses. Deformationsof the coil give rise to change of thefield map, extra losses and quenches.Cooling is obtained either by means ofhelium passages between layers 2 ,35 or bymea2s ?f heat drains stuck to the lay­ers,2 to drive the heat from within thecoil to the helium bath.

The direction of the cooling passa­ges varies according to the design, vert­ical passage being much better from heattransfer point of view. Some coils 1 usinga very flat and wide braid are just cool­ed along their edges. Coils with heatdrains have their drains protruding dif­ferent ways from the coil into the heliumbath.

OUTERSUPERCONOUCTINGWINDINGS

For impregnated coils and heat draincoils the thermal contact behaviour bet­ween superconductor and resins is veryimportant. This contact must remain goodfor the lifetime of the magnet and man,studies are going on this subject. 21 ,3

Provided thermal contact remainsgood for the life time of the magnet,heat drains have the advantage of givinga good packing factor and to equalize thetemperature within the coil. In case ofquench they make the propagation of nor­mal state through the whole magnet easier,avoidin~ any hot point which could burnand destroy the coi 1.

2. Choice of the current---------------------Many designers of pUlsed supercon­

ducting magnets choose currents close tothe ones used in conventional magnets ;this would enable us to use a similarvoltage division along the tunnel for asuperconducting synchrotron. It is alsothought that with big conductors a betterfilling factor can be achieved for thecoil winding. However, as it has beensaid previously one needs for the momentseveral stages (2 or 3) for the conductorassembly process and this leads to a poorpacking factor of the conductor. There­fore the question can be asked whether aone stage low current conductor with aslight smaller filling factor duringwinding could result in a higher overallcurrent density. The current lead problemswould be much easier and the voltagedivision along the tunnel could be prob­abl~ changed without difficulty.

For the time being, the tendency isto locate the shielding as close as pos­sible to the coil in order to get themaximum increase of the magnetic fieldand to reduce the stored energy. The onlylimit to this close position is satura­tion problems.

This solution makes the cryostatdesign much easier, its outer wall beingthen no more subject to variable field.In addition no other force than the weighthas to be held through the cryostat.

In some designs, one thinks of usingthe shielding directly to contain thebursting forces of coils (typically200 t/m for 5 T and ¢ = 10 cm); it wouldthen serve for both magnetic and mechani­cal purposes. The shielding is designedfrom the magnetic point of view and istoo large from mechanical point of view,but as silicon laminations have to be

212

used to reduce hysteretic losses and asthese laminations become fragile at lowtemperature, the large mechanical dimen­sion is in fact useful. Another advantageof large mechanical dimensions, is thevery small deformations under stresseskeeping a constant field homogeneity asthe field increases (fig. 13 and 14).

Fig. 13. Iron shielding of MOBY dipole.(CEN/Saclay) .

Fig. 14. Coil structure of a two-layerdipole.(Brookhaven National Laboratory)

The thermal contractions of the coiland of the silicon steel are differentand for that in some magnets one prefersnot to use directly the shielding tocontain the bursting forces. The shiel­ding is then cut in two parts and must besupported by an external clamping struc­ture.(fig. 15). Stainless steel tie rodsare used to join the clamps together.Disc springs under the nuts of the tierods serve to take up differential ther­mal contraction and thus keep the coilalways in a state of compression, butunfortunately in using springs understresses, this structure can move by alittle amount.

This gives typically for a 5 T and 10 cmaperture dipole 90 J/cycle/meter which isroughly the superconductor coil losses if10 ~ filaments are used.

At same field but smaller aperturethe amount of iron losses compared to thecoils losses is less but still importantThe higher the maximum field in a cons­tant aperture is, the higher the amountof iron losses is.

Of course, the comparison is stillworse if 5 u or less filaments are used.These figures are only calculated and onehas to wait realistic measurements tocheck them.

The cryostat design is considerablysimplified if the i~on shielding is athelium temperature 1 ; then only theinner wall is under variable field. Thelosses per meter in such a tube are :

r = radius of the inner wall tube,

e = thickness of the wall,p resistivity,dB-at= speed of field variation.

With g~ = 1 Tis ; r = 5 cm ; e = 1 mm

and p = 5.10- 7 m (stainless steel) oneobtains P = 0.8 W/m which is not negli­gible ; non metallic tube, or corrugatedstainless steel tube should be used ifthe rise rate is kept fast.

Very few cryostats working horizon­tally for pUlsed magne~ are under const­ruction 17 (fig. 16) ; a special effortmust be made to keep the overall dimen­sions small and for that, welded solu­tions have to be used extensively.

(6)ep

3JT r

2p =( g~ )

Fig. 15. AC4 dipole magnet.(Rutherford National Laboratory)

Important mechanical problems remainto realize stacking of laminations seve­ral meter long at low temperature andvery few solutions which satisfy all re­quirements are coming out.

The magnet support system must bedesigned to have low heat leaks and togive a reproducible position of the ma­gnet with regard to the cryostat aftereach cooling down in order to know fromthe outside where the magnetic axes are.

The main drawback of putting theiron shielding at helium temperature, inaddition to the necessity of cooling downa big amount of material, is the hystere­tic losses.

Taking as a basis ma~netic measure­ments made at Brookhaven3 (at 2 T andHe temperature), losses of 5.10- 2 J/cyclelkG are found for silicon steel (3.25% Si).

Whether to have current leads foreach cryostat or for several of them isnot solved, the first solution allowingan easier replacement in the acceleratortunnel. Studies made on the cooling sys­tem applied to a whole accelerator39seem to give a tendency where the magnetswould operate in boiling helium bath, thecryostats being cooled by distributedrefrigerators.

213

Table 1 Estimated losses

Fig. 16. Cryostat of MOBY (CEN/Saclay) :the nitrogen shielding (above)and the vacuum tank (below).(ALSTHOM Company)

5. g2!!~~!~_9f_1922~2

For a 5 T dipole with an aperture of10 cm which could be used on a 1000 GeVrange synchrotron, the losses are esti­mated as a function of the cycle dura­tion, on the following table 1. The fila­ment si ze is assumed to be 10 u, •

This table shows that for a cycleduration of more than 30 s, the lossesdue to the cryostat and the supply arepredominant; so, it is not necessaryto reduce below 10 u the size of thesuperconducting filaments.

Cycle duration (s) 5 10 20 30 60

S c losses (W/m) 18 9 4.5 3 1.5

I ron losses (W/m) 18 9 4.5 3 1.5

Cryostat 10sses(W/m 4 4 4 4 4

Supply losses (W/m) 4 4 4 4 4

Total (W/m) 44 26 17 14 11

6. E~~!~!!9~_~ff~~!2_9~_~~!~I!~12_~!_19~

!~~I?~E~!!:clE~

Three types of problem arise

- damage of materials (change ofproperties, destruction)

- increase of refrigeration power

- local temperature rises sufficientto make the magnet quench.

13 Typically,for a 1000 GeV synchrotron,10 particles per cycle of 10 seconds ,a beam loss of 2% during extraction wouldgive a mean dose of 10 7 - 108 ra~l peryear and a peak dose of up to 10 radsper year. This peak dose would occurdownstream of the extraction region andis not tolerable because it would leadto a lifetime smaller than a year forinsulants and anyway the temperature risewould be 10 to 100 o K.35 Therefore it isclear that one will have to shield thesemore exposed magnets.

Radiation effects on superconductingmaterial result in change of propertiesbut everything is not yet clear in thissubject. Decrease of 5 to 10 % on criti­cal current was observed on Nb-Ti alongwith the increase of the copper matrixresistivity up to a factor 5040 .

V. EXl?erimental work

As long as many thousand ampereintensity will be needed in magnets andhigh filament number conductor (> 5000)is not yet available, braiding or cablinga large number of wires with a perfecttransposition will be necessary. But suchtransposed conductor have a low fillingfactor and a high compaction is needed toget suitable large current density. ThiscompactiOn sometimes breaks or damagesthe wires and gives cross-over points.

214

For example in compacting a 24 insulatedwire braid filling factor as low as 50%,we have had several breaks and a greatnumber of shorts;17 also a cable of oxi­de insulated wires had four broken wires(compaction 60%). These considerationsfavour a conductor with a low number ofwires; nevertheless improvements in com­pacting are expected in the future.

Many solenoids have been built tounderstand the superconductor behaviourand to measure losses.

The characteristics of some recentcoils are listed below as an example.

TABLE 2

Laboratory Brookhaven34 Karlsruhe 41 L. B. L. 42 Saclay17

Coil name 42 24 75 M1 M3

Coil

inner diameter (cm) 2.5 2.5 2.4 6 3.90 3.90

outer diameter (cm) 7.5 7.5 8.4 10 12.3 12.3

length (cm) 4.5 4.5 5.3 6 8.20 8.20

B > at I c (T) 5.3 5.3 5 - 6.9 6.6

potted - yes no - yes yes

Composite

Cu/Sc 1.3 1.3 3 : 1 - 1.0 1 .0

filament diameter ( \.L ) 7 7 13. 11 10 10

filament number 210 210 361 - 1000 1000

composite diameter( \.L ) 200 200 0.5 279 440 440

twist pitch (mm) 2 2 0.4 - 2 2

Conductor

type braid braid - braid braid braid

number of wires 33 33 1 48 24 24

transposition length (mm) 70 - - - 60 60

insulation metal formvar - !copper oxide formvar formvar

I1)C/I C 1 .0 0.7-0.9 1 - 0.95 0.96

I/I c 1 .06 0.6 1 - 0.23 0.94

for rise time (s) - - - - 2 2

Many other solenoi~have been built inother laboratories but their characteris­tics have not been published or have notbeen available in time to be listed here.

Insulated wire assembly :

In almost all the laboratories theorganically insulated wire coils exhibi­ted erratic behaviour. Some coils had ahigh degradation (l/lc = 0.5 to 0.7), afew of them worked correctly. Dependingon the particular case the reasons oftheir behaviour could be different. Forexample with a three wire cable (Cu/Sc1 : 1), the coil was improved (no more

215

degradation) when it was tightly rewound.The wire movement is for a part the rea­son of the observed training. In an otherexperiment,a slightly modified loss measu­rement apparatus gave the overall magne­tization curve of the coil showing up theconductor movement. 21 Besides the wireis more movement sensitive when the cop­per to superconductor ratio is low (closeto 1 : 1). In a more general way a betterbehaviour was noticed when the Cu/Scratio is nigher.

A typical example of impregnated or­ganically insulated wire coil behaviourwas observed on coil M3 using a 24 insu­lated strand braid. 17 In a first test ahigh degradation was obtained (I/I c =0.49)

42A

0-

6 -

I'"1/'"

&~:""'".. ~~o "'~_~.__ I__.I L-.-_L----.l..-_o 10 20 30 40 50 W ro

B, kG

COIL LOSSFOH 8secRISE TIMECOHHECTEO

10·~ rOH VOLUII1(

12

14 .----,~---,------r-___,_--___._---,--_,

With metal impregnants, suitableheat treatment producing intermetalliclayer at the surface and small transpo­sition length, the coupling may be lowe­red so that the extra losses with smallB (few kG/s) are of the same order aswith organic insulation. 34 Figure 17shows the influence of different inter­metallic layers and transpositionlengths . The coil 42 A has no interme­tallic layer and coils 42 A through 42 Dhave progressively thicker layers. Coils42 D and 43 are identical except for theconductor which has different transposi­tion length (5.2 cm for coil 43 insteadof 7 cm).

are only losses in the composite itself(superconductor and copper) and mechani­cal losses but cabling, compacting andwinding may introduce shorts. With cou­pled wires,extra losses are due to cur­rent loops between wires. Theoreticalapproach may be made in ~2e same way asfor multifilament wires. ,45

and by measuring the currents in eachwire (this measurement tending to makethem all equal) higher degradation wasnoticed (III = 0.25). It was found thattwo wires haa lower current and by dis­connecting these wires almost no degra­dation was observed (1/1 c = 0.96) even ata B = 2 Tis. A possible explanation isthat these two wires were broken and thattheir current was transferred through thecross over points to the other wires thenheating the conductor. These troubleswould have been avoided if the wire insu­lation had not been locally destroyed.

Coupled wire assembly :

Coupled wire conductor behaves as awire having a large number of filamentsbecause it allows current transfer bet­ween wires. It should be more stable thaninsulated wire conductors. Moreover as itis fUlly transposed, it is perhaps lesssensitive to self field instability thata more th~n ten thousand filament conduc­tor will be. The metal filled conductorallows this coupling but does not limitthe induced currents and leads to lossincreases. The use of oxide insulation orintermetallic layer lowers this troublekeeping about the same stability. Theexperimental results of AC3 0ttde insula­ted conductor (I~c/lc = 0.95) and metalimpregnated conductor coil (1/1 c = 1)34seem to agree with that explanation. Itis also the case of coil M1 made of thesame braid as M3 but during winding, theinsulation was destroyedby the aluminacharged epoxy allowing a high couplingbetween wires. Small degradation was obs­erved (IDc/lc=0.9S)even with three bro­ken wires but in pUlsed condition theextra losses lead to an important degra­dation (1/1 c = 0.3 to 0.4) due to tempe­rature increase.

In general channel or heat draincooled coils exhibit no degradation inDC condition but depending on the losslevel heat drain coils may chiefly havesome degradation in pUlsed condition dueto temperature increase. 21 ,43

The magnetization losses are nowwell understood and experimental measu­rements agree with theory within a factor1.3 to 1.6. The extra losses due to selffield do not appear experimentally to beas i'Rortant as one thought a few yearsago. However current sharing betweenwires is very much disturbed when theconductor is not fUlly transposed. 44With no coupled (insulated) wires there

Fig. 17. SolenoId coil losses vs fieldfor various heat treatment andmetallized conductor.

(Brookhaven National Laboratory)

Figure 18 gives the measured lossesfor different rise times.

216

16

w

':~....Ju>-u......(f)

W....J::>0J

..4

00 70

1. ~~!'Q:P~

In the project of the 300 GeV ma­chine it was involved the use of super­conducting magnets to increase the ener­gy to about 1300 GeV ; two solutions havebeen proposed:

a) the missing magnet solution :

after the 200 GeV stage is completed,superconducting magnets would be instal­led in the gap of the lattice rather thanconventional magnets. The energy beinguprated to 500-600 GeV and perhaps 700­800 GeV if it is possible to supply con­ventional and superconducting magnetssimultaneously.

By replacing in a second stage con­ventional magnets, the energy will beuprated to 1000-1300 GeV.

Fig. 18. Solenoid coil losses vs fieldfor various rise times.

(Brookhaven National Laboratory)

The oxide insulation seems suffi­cient to obtain the resistance needed tolimit the inducedcurrents. Experimentallyit was shown that such an insulated braidhad, at 1 TIs and 3 T max field, losses1.5 time the losses at B = 0 ; and witha three component wire cable the lossesat 1 TIs and 3 T maximum field are 1.2time the losses at B = O.

All the experimental results are notyet fUlly understood and are difficultto compare, the experimental conditionsbeing not well known; however it seemstoday that a poorly coupled conductor(oxide or any resistive layer) whichallows a good stability and low extralosses is the preferred solution.

VI. Superconducting dipole

Pulsed superconducting magnets havebeen built in different laboratories forover five years.

The development of those pUlsedmagnets have been pushed up by the pro­jects of superconducting accelerators.If some of these projects seem to bewithdrawn the others are quite promising.

217

b) the separate ring solution :

a separate superconducting ring is builteither in the same tunnel above the 200GeV ring, or in an adjacent tunnel, theinjection energy being 200 G~V.

The main parameters are listed intable 3.35

Three European laboratories (IEKPKarlsruhe, Rutherford HEL, CEN Saclay)are engaged in the development of suit­able prototype dipole magnets. The GESSS(Group for European Superconducting Syn­chrotron Studies) collaboration coordina­tes the work of these laboratories.

The main parameters of completedand in project dipoles are listed intable 4.

DT (fig. 19) has a flat race-trackcoil wound on resin bonded fiber glassformer and is not potted in resin. Thehigh level of losses (13 W at 5 s risetime) is due to eddy currents and fila­ment couplings in soldered cable. Thismagnet will be soon tested with the samecoil but potted in epoxy resin.

The coil D2 and D3 will have thesame saddle shape with the ends shapedso as to cancel high mUltipole coeffi­cients in field integral. Both will havethe same field and aperture but D3 willbe twice longer. For D3, the choice ofthe conductor is not frozen yet.

TABLE 3 - Different options for superconducting synchrotrons

Missing magnetSolution 2nd Stage Higher Separate Ring

1st Stage 2nd Stage Energy Variant

Machine Energy 500 1 000 1 330 1 000(GeV)

Injection Energy 10 10 10 200(GeV/c)

Magnetic field in 4.5 4.5 6 4.5dipoles (T)

Diameter of magnet 100 - 110 100 - 110 100 - 110 70 - 80aperture (mm)

Stored Energy in 450 900 1 800 500dipoles (MJ)

Machine cycle time:lC 11 22 44 12(s)

Average flux 9 x 10 11 4 x 1011 2 x 1011 8 x 1011(protons/s)

x Machine cycle time assumed to be 3 times magnet ramp time.

Fig. 19. DT dipole magnet.(IEKP Karlsruhe)

218

~~~~::~~:~.~~~~~:

AC3 : (fig. 20, 21)

Fig. 20. AC3 dipole magnet.(Rutherford National Laboratory)

GESSS laboratory

DesignationYear of completion

Magnet

coil aperture or diameter (mm)

length (m)pulsed field achieved (T)orspecified

at field rise time (s)

cold iron

Conductor

type

No of strands

individual strand diameter(mm)

filament by strand

filament si ze (iJ,)

Nb-Ti : Cu : Cu-Ni

cable filling

operating current (A)

coil cooling

TABLE 4

Karlsruhe Saclay

DT D2 D3 MOBY ALEC1972 1972 1973 1972 1973

80 x 40 80 80 100 1100.4 1.4 2.8 0.5 1.53.2 4.5 4.5 6.0 5.5

5 10 - 20 3 - 10 5 - 15 5 - 15

yes yes yes yes yes

cable cable cable braid cable

10 12 - 24 360.5 0.54 - 0.44 0.5

1045 1000 1000 1000 1000

10 12 - 10 101 :1.5:0.2 1:1.1:0 - 1 : 1 : 0 1:1.6:0solder solder - resin resin

1035 1520 - 1500 2500

channel channel channel -heat heatdrain drain

TABLE 4

GESSS laboratory RUTHERFORD

Designation AC3 AC3 modified AC4 AC5Year of completion 1971 1972 1972 1973

Magnet

coil aperture or diameter (mm) 100 75 90 90length (m) 0.4 0.4 0.8 1.5

pUlsed field achieved (T) or 3.5 4. 1 4.5 4.5specified

at field rise time (s) 1 1 3 3

cold iron no no yes yes

Conductor

type cable cable cable cable or flatened cable.

No of strands 90 90 2S S - 1S

individual strand diameter(mm) 0.4 0.4 0.85 1.1 - 0.7

filament by strand 1045 1045 2035 10.000-20.000filament size ( 1..L ) 8 8 12 5

Nb-Ti : Cu : Cu-Nj 1 :1.3:0.2 1:1.3:0.2 1:1.3:0.2 1:1.3:0.2cable filling resin resin resin resin

operating current (A) 5100(DC 4500(DC) 5400 4000coil cooling heat drain heat drain channel channel

219

2. U.S.A.

MOBY and ALEC will be tested in acryostat with a horizontal bore axis.

The main-parameters are listed below(table 5)46,47

TABLE 5

285

220

937

30216

10 5

1206.1014

0.6

4

0.03

3.280

Average radius of curvedsection (m)Experimental straight sectionlength (m)

Total circumference length (m)

Injection energy (GeV)Each ring proton energy (GeV)

Equivalent proton energy(GeV)Accelerator rise time (s)Each ring stocked protons

Injection dipole field (T)Maximum field (T)

Field variation (Tis)Length of dipole (m)Aperture diameter (mm)

200 GeV intersecting storage accele-...................................rator ISABELLE (Brookhaven National.............. Laboratory).

Those intersecting storage accele­rators will use the AGS as an injector.The ring consists in two interlaced race­tracks each made of two half circles witha regular lattice and two long experimen­tal straight sections. The orbits in theregular lattice are separated by 0.2 m inthe vertical plane.

~~~~~~.~~l?~~::

The so-called constant perimetermethod is used for winding each layer.This technique allows to build an entirepole without internal joint in a verycompact way. MOBy17 (fig.10,13) is thehighest field magnet under construction.Its expected field homogeneity is 10- 3 in50% of the coil aperture. ALEC has a grea­ter length(1.5 m)but a slightly lowerfield. The design field homogeneity is4.10-4 in an aperture of 70 x 40 cm. Theiron yoke is located further to avoidsaturation effects. Considerable indus­trial participation is involved in theconstruction of ALEC.

Fig. 21. AC3 winding.

(Rutherford National Laboratory)

The winding is made of six doublepancakes connected in series. It reached3. 5 T central field (3 ..9 T peak field) ata current of 4580 A. Two additional pan­cakes,reducing the aperture diameter from100 mm to 75 mm,has allowed to reach acentral field of 4. 1 T (4. 7 T peak field)at a current of 4000 A. Particular atten­tion was paid to minimize the losses(44 J in charging the magnet to 3 T and

discharging). Less attention was paid tofield homogeneity and industrial winding.

AC4 44 (fig. 9,15) has two doublelayers of winding per pole. A split ironyoke is used, the two halves being heldby a cylindrical stainless steel clamp.The expected field homogeneity is 4.10- 4over 60 mm diameter.

AC5 44 is based on the AC4 design. Itwould have six double pancakes rather thanfour to allow safety margin on operatingcurrent. The conductor will be either arectangular 5 to 7 strand cable with about20.000 filaments per strand or a flatte­ned cable of 9-15 strands with less fila­ments by strands. A solid laminated ironyoke is studied. The dipole will be testedhorizontally.

220

The first approach for the dipolefield was an arrangement of magnet coilsto allow one ~9il to use the return fluxof the other.

Because of the slow rise time, BNLis developing a metal filled su~ercon­

ducting multifilamentary braid. 4 Thewinding shape results from the cos 8 dis­tribution and usesa single layer for a4 T dipole. 48 The ~haracteristics of di­pole models being studied now are listedin table 6. 49

132 132

13.2x 0.4 13.2 x 0.4

1 20

2 270 2 780

95 100

24 31

3.4 4 TABLE 7

Name

The magnets have in each quadranttwo conductor blocks of the same widthbut with different current densities. Thetechnique used is the so called constantperimeter. The main characteristics ofthe dipole (fig. 11, 22) are listed intable 7. 50 ,51

Fig. 22. Pulsed dipole nO 8.(Berkeley National Laboratory)

2 G

metalfilled

7

0.8 : 1

200

braid

2 P

poly­bondex

10

2 : 1

200

braid

TABLE 6

Name

Insulation

Filament size (l.L)

Cu/Sc

Wire diameter (l.L)

Conductor type

Wire number

Conductor dimensions(mm x mm)

Minimum rise time (s)

Current at maximumfield (A)% of short sample

(1O- 12 0.cm)

Current density

(kAI cm2)

Obtained maximum fieldin aperture (T)

It is planned to build four more modelsto check construction tolerances and testnewer metal filled braid.

Two I S A dipole models are planned,of similar design but 8 em aperture, 90cmlength and six current blocks (3400 A).Each unit will be tested in a verticaldewar then mounted end to end horizontallyto be tested for an extended period. Thefirst one will be ready for testing inSeptember 1972.

~~~::~~:.~::~:~:~.~~~?:~!?:~

Although the main program aimed atthe use of superconducting transport ele­ments for bevatron areas, an extensiveprogram of development work is also di­rected towards the construction of super­conducting magnets for possible protonsynchrotron applications.

coil aperture (mm)

length (m)

operating current (A)DC field (T)

max. field in end (T)

pulsed field achieved(T)

at field rise time (s)

coil coolingcoil iron

ConductortypeNo of strands

strand diameter (mm)cable filling

63.5

0.40

920

3.94.5

3.5

0.5

channelyes (soft iron

wire)

cable133 (19 x 7)

0.2Ag-Sn solder

221

The main characteristics of thedipoles (fig. 23) are listed in table 8.

TABLE 8

~~~::~~~~~~~~~¥.~~::~~~::.:~~¥.~~:

~~:.~::~.¥:~~~:~~.~~~~~:~~:~~(Argonne National Laboratory)

SECTION BBSECTION AA

TABLE 9

\-§tt'li i5JU A '-,

·WoO.UllJ4CICET··~--R"'OI"'TION SHElO

, r--CRTOSTAT-, r-lROH SHlELD- ....'rDlPOLE M4GH[T, QUADRUPOlE MAGNtT­

_VIIlCIJlAll CtlAIIII8ER

Fig. 23. Proposed dipole for the Z G S(Argonne National Laboratory)

TYPICAl 8 FT. SECTtOk

~~~~~~~~.~~~:~::~~~:.~~~~:~~~:~

An energy- doubler concept has beenproposed to uprate the energy to 1000 GeV53,54. This superconducting ring would bemounted 700 mm above the main ring in thesame tunnel. The superconducting magnetswould have very small bore (5 cm) and thefield would be used from 2.2 T to 4.5 t;this simplifies the magnet technology.

A quadrupole 55 has been tested. Itscharacteristics are listed in table 9.

10

1.45

2.9

500

15 to 20

inner diameter (cm)

length (m)

maximum field (T)

operating current (A)

current density(kA/cm2)

A superconducting stretcher ring hasbeen proposed52 to increase the intensityof the Z G S. The beam duty cycle wouldbe about 100% and a factor of five higherrate of acquiring data would be possiblein many experiments. The average intensi­ty of the Z G S would be doubled (theA G S operating without flat top). Al­though the magnets will not be optimizedfor pUlsing they will be pulsable with along rise time (20 to 30 s).

The channels are formed with the Bstage coated epoxy insulation wrappedaround the conductor. In the next testthe soft iron will be replaced by lami­nated iron yoke.

A new magnet with a length of about1 m is planned to be constructed and tes­ted in an horizontal cryostat cooled bya helium refrigerator.

Conductor

type

strand number

filament number

filament diameter ( ~ )

cable

7

700

10

bore (cm)

length (m)

current (A)

gradient (Tim)

% of J c at Bmax

2.5

0.5

800

40

57

Conductor

The dipole are wound of flat pan­cakes with turned up ends. number of strands

Cu/Sc

insula Hon

1 21

3 : 1

formvar

222

Fig. 26. Horizontal cryostat.(Radiotechnical Institute Moscow)

The two first steps use conventionalmagnets but all the plans are made to beable to replace these magnets by super­conducting ones keeping the same ring. Another way is a mUltistage scheme. 57 Thedipole length will be about 6 m. A modeldipole of 6 cm aperture and 70 cm lengthhad been completed. 58 ,59,60 The techniqueis a cosine distribution of 2 mm diameterwire arranged in thin stainless steelgrooves (fig. 25). The magnet is placedhorizontallyin a fUlly demountable cryos­tat (fig. 26). The study of this type ofmagnet is continued by an epoxy impregna­ted version. 59p

At the Radiotechnical institute(Moscow) the study of high energy syn­chrotrons has been carried out for seve­ral years, the aims of the program beinga synchrotron of energy higher than 1000GeV. The general principle given at Ere­van conference is the uprating of a 1000GeV.56 The main characteristics arelisted in table 10.

TABLE 10

1st step 2nd step 3th ste

Proton energy 500 1 000 4 000(GeV) 5 000

Intensity 2.10 12 3.1013 10 13(proton/s)Diameter (m) 5,435 5,435 5,435Max. field (T) 0.8 1.6 6.0-8.0Aperture (mm) 40x66 40x66 o 66Pulse duration 6 3 12

(s)

Injection energ) 6 18 18(GeV)

Fig. 25. Dipole cylinder with grooves forthe superconducting cable.

(Radio technical Institute Moscow)

223

VII. Conclusion

More knowledge is today availablefor the design of new machines. Both therequirements and the methods to fulfilthem did not yet reach a complete agree­ment between all the laboratories. On theother hand so many experiments have beenmade that it is not possible up to nowto compare all their results and to drawa final precise conclusion. However con­siderable progress has been made in thetwo past years in the understanding ofthe behaviour of pulsed coils.

Our feeling is that in the comingmonths a large effort will be continuedspecially in shifting more and more theconstruction from the laboratories toindustry and more will be known aboutoperational behaviour of such magnets.

REFERENCES

1. A. Van Steenbergen - "Superconductingsynchrotron development at B.N.L." ­Proc. 8th Int. Conf. on High EnergyAccelerators, CERN 1971 - p. 196

2. D.B. Thomas - "Superconducting pUlsedmagnets for a 1000 GeV synchrotron."­Proc. 8th Int. Conf. on High EnergyAccelerators. CERN 1971 - p. 190

3. H. Btechna, M.A. Green, W. Heinz ­"Superconducting synchrotron magnetperformance in a closed circuit cryo­genic system." - Proc. 8th Int. Conf.on ~igh Energy Accelerators, CERN 1971p. 183.

4. G. Bronca & al - "Studies and construc­tion of superconducting magnets appliedto synchrotrons of more than 1000 GeV"Proc. 8th Int. Conf. on nigh EnergyAccelerators, CERN 1971 - p. 177

5. G. Parzen & K. Jellett --"A small aper­ture high field bending magnet." ­B.N.L. 16 006, AADD 178 (July 1, 1971)

6. J. Perot - "Coil end effects of elec­tromagnetic lenses without iron." ­SLAC TN-69-20 (November 1969)

7. G. Bronca & al - "Fringing fields andiron saturation in air core magnets."­3rd Int. Conf. on Magnet Technology,Hamburg 1970, p. 64.

8. J.H. Coupland - "Equations and formu­lae for magriets with air cored win­dings of saddle coil type." - RHEL/R203 (August 1970).

9. M.A. Green - "The elimination ofhigher mUltipoles in the two dimen­sional and integrated fields of con­ductor-dominated dipole and quadrupolemagnets with iron shells." - UCID 3493Eng. note M 4373 (January 15, 1971).

10. R.B. Meuser - "End effects in super­conducting beam-transport magnets". ­Proc. 1971 Particle Accelerator Nat.Conf. Chicago (March 1971), p. 677.

11. J. Perot - "Caracteristiques de dipo­les supraconducteurs dans la gamme 4a 6 teslas." - GATS/71-82 Bis (Septem­bre 1971).

12. T.C. Randle - "Estimates for the iielderrors from coil winding irregulari­ties in superconducting magnets." ­RHEL/R 197 (June 1970).

13. F. Kircher - "Contribution a l'etudedes causes d'inhomogeneite de champdans un dipole supraconducteur pulseet proposition d'une methode de cor­rection." - These soutenue a l'Uni­versite de Paris Sud le 7 Juin 1972.

14. G.Ries - "Field errors due to windingtolerances in a superconducting ben­ding magnet." - BSG - Notiz 71/1(February 1971).

15. G. Parzen - "Effects of random errorsin coil locations in a high field su­perconducting accelerator." - B.N.L.16 616, AADD-187 (February 25, 1972).

16. M.A. Green - BSG notiz 71/7 (1971)

17. G. Bronca & al - "The Saclay super­conducting pulsed dipole MOBY." ­These Proceedings.

18. GESSS M.D. Working Group - "Prelimina­ry studies for a 1000 GeV superconduc­ting extension of the European accele­rator". - Proc. 8th Int. Con£. on highEnergy Accelerators,CERN 1971,. p. 171.

19. M.A. Green - "The theory of residualfields in two dimensional dipoles andquadrupoles without iron shields." ­UCID 3508 (January 20, 1971).

224

20. M.A. Green - "Magnetic measurementsof a superconducting dipole magnet."VCID 3494 (December 15, 1970).

21. M.N. Wilson & al - "AC3, a prototypesuperconducting synchrotron magnet."1972 Applied superconductivity Conf.Annapolis., p. 277.

22. J.L. Duchateau - "Etude du champ re­manent perturbateur dans les aimantssupraconducteurs multipolaires." ­SEDAP/72-109 (ler Aout 1972).

23. G. Parzen & K. Jellett - "High fieldbending magnets for Isabelle." ­B.N.L. 16677 - AADD 188 (February 29,1972).

24. J.P. Pouillange & G. Prost - "Lescomposites multifilamentaires." ­GATS/71-24 (Mars 1971).

25. J.P. Pouillange & G. Prost - "Effetdu champ magnetique exterieur et duchamp propre sur le comportement d'uncomposite multifilamentaire alimenteen courant pulse a longue periode."­SEDAP/72-10 (Janvier 1972).

26. M.N. Wilson - "Filamentary compositesuperconductors for pulsed magnets."1972 Applied superconductivity Conf.Annapolis. p. 385.

27. P.F. Smith - "Superconducting synchro­tron magnets" present status, Proc.8th Int. Conf. on High Energy Accele­rat 0 r s, CERN 1971 - p. 35.

28. F. Voelker - Privatecommunication.

29. H. Brechna - "Pulsed superconductingmagnets." - 1972 Applied superconduc­tivity Con£. Annapolis. p. 226.

30. G. Bronca & al - "Problems and recentdevelopments concerning A.C. super­conducting magnets for synchrotronaccelerators." - 3rd Int. Con£. onMagnet technology, Hamburg 1970,p. 701.

31. P.A. Hacq - "Influence de l'angleentre B et J sur le courant critiqued'un fil supraconducteur." - GATS/71­101 (Octobre 1971).

32. G. Prost - "Influence des conditionsde refroidissement. 11 - SEDAP/72-42(Mars 1972).

33. J.R.A. Popley & al. - "A new filamentsuperconducting composite." 4th Int.Cryogenic Ingin. Conf. Eindhoven.(May 1972).

225

34. A.D. McInturff & al - "Pulsed fieldlosses in metal filled superconduc­ting multifilamentary braids." ­B.N.L. 16516, Journal of Applied Physics43, 3546 (1972).

35. GESSS Group - "Towards an Europeansuperconducting synchrotron" - GESSSl(May 1972).

36. W.S. Gilbert - "Test results on pul­sed superconducting synchrotron typedipole magnets." - Proc. 8th Int.Conf. on High Energy Accelerators,CERN 1971 - p. 206.

37. P. Brauns & al - "Measurement ofthermal resistance between materialspotted in epoxy resins." - 4th Int.Cryogenic Engin. Conf. Eindhoven(May 1972).

38. A.D. McInturff & J. Claus - "Lowtemperature measurements." - 3rd Int.Conf. on Magnet Technology,Hamburg1970. p. 45.

39. J.W. Dean - "The cryogenic aspects ofthe proposed European 1000 GeV super­conducting synchrotron." - RHEL/M/ A2 2.

40. H. Brechna & W. Maurer - "Irradiationeffects in superconducting synchro­tron magnet." - KFK 1468.

41. G. Ries & H. Brechna - "A.C. lossesin superconducting pulsed magnets."­KFK 1372 (1972).

42. W. Gilbert & al - "Coupling in super­conducting braids and cables." - 1972Applied superconductivity Conf. Allna=polis, p. 486.

43. H. Artiguelongue & al - "Synthese desresultats d'essais de petites bobinessupraconductrices alimentees en cou­rant pulse." - SEDAP/72-64.

44. P.F. Smith - "Superconducting synchro­tron magnets present status." - Proc.8th Int. Conf. on High Energy Ac~rators, CERN 1971 - p. 35.

45. G.H. Morgan - "Theoretical behaviourof twisted mu1ticore superconductingwires in a time varying uniform magne­tic field." - J. Appl. Phys. -±..Lp. 3673 (1970).

46. J.P. Blewett & al - "Preliminarybasic parameters proton storage ringproject." - CRISP 71-14 (July 1971).

59. V.P. Alexeev & al - "Design of supe~­conducting magnets for accelerators .NTD 711 A.

V.P. Alexeev & al - "Some aspec~s ofthe superconducting magnets deSIgnfor high energy protron synchrotron".Proc. 8th Int. Conf. on High EnergyAccelerators CERN 1971 - p. 228.

60.

A.L. Mints & al - "A 1000 GeV cyber­netic accelerator status report." ­Pro~7th Int.Coni.on High Energy Acce­lerators (1969) Erevan USSR, Vol. 1 ­p. 60.

A.A. Vasiliev - "On the constructionpossibility of the accelerators. andstorage ring using superconductlngmagnets." - Proc. 7th Int. Con£. onHigh Energy Accelerators (1969) ErevanUSSR, Vol. 2 - p. 656.

58. CERN Courrier n012 - Vol. 10, PP 386­387 (December 1970).

57.

56.

51. W.S. Gilbert - Private communication.

52. Superconducting (beam) stretcher ringfor the zero gradient synchrotron.Argonne National Laboratory (December1971).

48. W.B. Sampson & al - "Superconductingsynchrotron magnets." - ParticleAccelerators 1, 173 (1970).

49. W.B. Sampson - Private communication.

53. R.R. Wilson - "1000 GeV with super­conducting magnets at Batavia." ­Physics Today (May 1971) - p. 19.

SO. L.B.L. program, W.S. Gilbert,without reference (April 1972).

47. J.P. Blewett - "200 GeV intersectingstorage accelerators." - Proc. 8thInt. Conf. on High Energy Accelera­tors. CERN 1971 - p. 501.

54. R.A. Carrigan Jr. - "Can an energydoubler be used implement a 350 GeVstorage ring system at NAL ?" - FN233 (N.A.L.) - (July 1971).

55. R. Sheldon & B. Strauss - "0.5 meterprototype energy-doubler quadrupolemagnet." - FN 235 (N.A.L.) - (July1971).

226